If you break into a single home and steal something, you can expect somewhere between five and ten years in prison, sometimes more if the dwelling was occupied or you were armed.

Now imagine your actions help cause millions of people to lose their homes entirely. Foreclosure notices taped to doors, families packing into cars, neighborhoods hollowed out, retirements vaporised, a global economy in freefall.

What do you get for that?

Perhaps a bonus. Sometimes a fine paid by your employer. Occasionally, if things go extremely poorly, you might have to resign with a generous package and take up a highly paid position on a corporate board.

One home equals years, but a million homes equals nothing; it all depends on who commits the act, or rather, it depends on how wealthy they are.

The math doesn’t math, and the calculation error is happening at an abstraction of our numbering system that is inherently beyond human comprehension. The numbers defy us.

Let’s take a look at them.

Stan O’Neal was ousted as CEO of Merrill Lynch in October 2007, following a disastrous run of subprime write-downs that would help sink the firm. He left with a severance package of Merrill stock and options worth roughly $161.5 million — on top of the $48 million he’d already been paid in 2006.

Weeks later, in January 2008, he joined the board of directors of aluminum conglomerate Alcoa (later Arconic).

In 2011, the Financial Crisis Inquiry Commission recommended that O’Neal be investigated for possible crimes connected to his conduct at Merrill. No charges were ever brought. He kept the package, kept the board seats, and largely disappeared from public view.

In the 2011 Wall Street drama Margin Call, the CEO character “John Tuld” (played by Jeremy Irons) is believed to be inspired by John Thain and Richard Fuld (Lehman Brothers); the fusion of names not exactly the subtlest of tells…

Thain stepped down as CEO of Merrill Lynch in January 2009 shortly after the bank was acquired by Bank of America. His exit was heavily criticised due to the $15 billion in fourth-quarter losses at Merrill and the awarding of early bonuses to executives. He departed with a reported multi-million dollar severance package.

Just over a year later, in February 2010, Thain became CEO and Chairman of CIT Group, a commercial lender. He was subsequently appointed to corporate boards, including Uber and Deutsche Bank, amongst others.

Richard “Dick” Fuld was the last Chairman and CEO of Lehman Brothers, which on September 15, 2008 filed the largest bankruptcy in U.S. history — over $600 billion in assets, gone — and detonated the chain reaction that took the global financial system with it. Between 2000 and 2007, Fuld was awarded roughly $889 million in compensation, of which he cashed out an estimated $529 million before the firm collapsed. (Hauled before Congress three weeks after the bankruptcy, he disputed the figure and insisted his real take was lower; the chairman’s reply was that taxpayers were now left with the bill.)

Two months later, he transferred his $13+ million Florida mansion to his wife for $100, a move widely read as shielding it from the lawsuits to come.

He faced no criminal charges. In 2009 he opened an advisory firm called Matrix Advisors. By 2021, Matrix Private Capital Group reported managing nearly $600 million in AUM. Its current 13F equity-only holdings alone account for approximately $190-210M in recent quarterly filings.

He kept the Greenwich estate, the Idaho ranch, and the Florida home. In 2015 he resurfaced to tell a conference he had no regrets, and that Lehman had been “mandated into bankruptcy” — a casualty, in his telling, rather than a cause.

Betting hard and loose with people’s livelihoods can be very profitable, even if the bets are lost catastrophically.

Citigroup took taxpayer-funded bailouts amounting to $45 billion.

What happened?

Banks gave out enormous numbers of home loans to people who couldn’t realistically afford them, and they knew this. They then bundled thousands of these shaky loans together into financial products called mortgage-backed securities — giant gift baskets of IOUs — and sold those products to pension funds, cities, foreign governments, and ordinary investors. They got friendly credit-rating agencies, paid by the banks themselves, to label these baskets as extremely safe. Some of the same banks then placed bets against the very products they were selling to their customers, knowing those products were going to fail.

They sold goods they knew didn't hold the value on the price tag, and with that knowledge, bet on their own goods becoming devalued.

Subscribe now

Millions lost their jobs, their means of living at all; lost on bets that the betters got rich off knowing they would lose.

When the housing market turned, the baskets blew up, the bets paid off spectacularly for a handful of insiders, and the losses landed on everyone else. Millions of American families lost their homes. The retirement savings of an entire generation took a hit it has, in many cases, never recovered from.

Black household wealth fell by more than half.

The U.S. government then spent trillions of taxpayer dollars bailing out the banks responsible. The bankers paid themselves bonuses, in some cases the very next year, straight out of the bailout funds.

This is documented in detail by the Financial Crisis Inquiry Commission, the Senate’s Permanent Subcommittee on Investigations, and dozens of books since.

While Lehman’s bankruptcy in September was the final legal trigger that broke the banking system, Merrill Lynch (under the leadership of John Thain) made a massive $30.6 billion dump, selling for $6.7 billion in July 2008 — a price that mathematically doomed the market.

Before that moment, banks could represent their portfolios using complex mathematical models, to give the impression their CDOs were still worth 80 or 90 cents on the dollar. The losses were hidden behind complexity. When Merrill Lynch needed to cover the fact their asset was worth a fraction of what it was valued at, they sold CDOs to Lone Star Funds for 22 cents on the dollar to save their own skin. The illusion of value completely shattered.

Because of accounting rules, once Merrill accepted 22 cents on the dollar, every other investment bank (including Lehman Brothers) was forced to adjust the value of their own CDO portfolios to match that 22-cent benchmark. Overnight, billions of dollars of capital vanished from the balance sheets of Wall Street, sparking immediate margin calls from their lenders.

Lone Star agreed to buy the CDO portfolio, but they didn’t actually pay the $6.7 billion price. Merrill Lynch desperately needed to clear its books; it financed 75% of Lone Star’s purchase itself.

Merrill Lynch gave Lone Star the liquidity to buy Merrill Lynch’s product…

In 2008, the seller (Merrill) lent the buyer money and retained the downside via the put. Capped buyer risk, uncapped seller risk.

In 2026, the seller (Nvidia) takes a warrant — it funds the demand and keeps the upside, while its downside is capped (it just doesn’t exercise if IREN tanks). Capped seller risk, leveraged seller upside.

The instruments do also differ of course — a put liability on the balance sheet vs. an off-balance-sheet warrant, making today’s version harder to see…

But both mechanisms allow linked corporations to pump up each others value, to the eventual expense of everyone not in the circle (the rest of the economy). The value is created by chaining promises together based on pure speculation. To the wealthiest, this generally amounts to networks casually betting with marginal cost; fun credits. They are bets because they are not paying the sticker price of today, but rather predicting a future event — the value on the stock ticker to get higher, later.

This makes sense on a surface read, but to an ordinary human being those fun credits are not a game; they are life itself.

Circular-pump schemes devalue the thing people need to live, in pursuit of a lucky buck. When unanchored valuations cause the economy to crash, much of the assets are bought up by the same tiny class, very cheaply. It is a baton hand off where the person who hands off the prize gets a slap on the wrist and immediately employed somewhere else to continue earning incomprehensible money. Not hyperbole by the way, but established consensus of the psychological reality: the human mind demonstrably cannot realistically entertain the relative scales of the figures at stake — the wealth accumulation of the top 1% almost certainly escapes even themselves.

The layers of separation are carefully architected throughout the industry, but never fully disconnected. Whether behind asset managers, wealth managers, private funds, family offices, tax “efficiency” specialists, offshore accounting, or shadow economies…

Or personal responsibility defused amongst boards, departments; behind “corporate” decisions, or “market sentiment.”

Even if that market is focused down to a single street…

The layers function as abstraction, protection, and obfuscation. The responsibility to the public is administered away behind paperwork and policy.

Meanwhile, ordinary citizens stand to lose everything, because of something they were never a part of.

So how is it done?

Merrill financed the buyer of Merrill’s own product. A transaction is manufactured that on paper looks like genuine demand. But there was no independent buyer who wanted $30 billion of CDOs at a fair price; the demand had to be financed into existence by the very party that needed it to exist: the seller, Merrill Lynch.

Demand financed into existence by the party that needs it to exist.

Neoclouds (IREN, CoreWeave, Nebius, Nscale, Lambda) are capital-starved relative to the capex GPU deployment requires, and traditional lenders are wary of collateral that depreciates fast. Nvidia has both the cash and the strongest incentive to see capacity get built — demand for its chips is gated by power, land, and datacenter buildout, exactly what IREN brings. So vendor financing relieves the customer’s capital constraint and de-risks the ecosystem bottleneck. Cisco did versions of this in the late ‘90s.

IREN is also doing a $625M Mirantis acquisition and building 2GW+ in Texas. Stack vendor-financed GPU purchases on top of debt-financed datacenter capex on top of M&A, and you get an entity whose solvency depends on continuous compute demand at projected prices. GPU lease/resale economics assume utilisation and pricing that a glut would erase.

The tell between “healthy ecosystem financing” and “round-tripping bubble” is independent demand. Concretely: IREN’s utilisation rates and contracted (vs. speculative) capacity; whether end-customers are real enterprises or other AI-native firms also living on vendor/VC capital; Nvidia’s growth in receivables and investment marks relative to cash collections; and any accounting where the same transaction inflates both a sale and an asset.

Nvidia’s warrant is struck at $70 — its value depends on IREN’s equity, which depends on AI infrastructure sentiment, which depends substantially on Nvidia’s own narrative. Asset values are referencing each other rather than an external anchor. This is the classic reflexivity that makes bubbles both self-reinforcing on the way up and self-reinforcing on the way down.

Nvidia is becoming a creditor/equity holder, demand guarantor and supplier across Nebius, Nscale, CoreWeave, Firmus, Lambda, IREN simultaneously. Its receivables and investment marks are now correlated to the financial health of a cohort of similar, thinly-capitalised, mutually-correlated customers. A downturn hits Nvidia three ways at once: lost sales, impaired investments, and impaired receivables.

When a vendor funds its own demand, reported revenue partly reflects the vendor’s own capital recycling rather than independent end-demand. The signal investors rely on (chip revenue = real AI adoption) gets contaminated. The key question: would IREN have bought these GPUs absent Nvidia’s funding? To the extent the answer is no, that revenue is endogenous.

In a genuine boom this is fine — the warrants get exercised, capacity gets absorbed, everyone wins. The danger is that vendor financing delays the moment the market learns demand is softening, because the vendor keeps the demand alive artificially. The correction, when it comes, is sharper because signals were suppressed.

CoreWeave carries roughly $46 billion in total liabilities (including leases) against just $3.34 billion of equity with off-balance-sheet lease commitments of $34 billion through 2028 adding hidden leverage.

CoreWeave’s Q1 2026 guidance of $1.9–2.0 billion came in below the $2.29 billion consensus (actual $2.08), reflecting a sector-wide gap between contracted backlog and actual near-term revenue. Vendor-financed demand plus depreciating collateral plus single-customer concentration removes the early-warning signals, so if it turns, it turns all at once.

The vulnerabilities that would normally signal trouble are now all gated by Nvidia's continued willingness to fund their cohort — so the early-warning system has been disabled. That's what pushes this situation beyond "bubble" into some kind of coordinated fragility.

CoreWeave pioneered GPU-collateralised debt using its NVIDIA GPU inventory and customer contracts as collateral to secure large debt facilities. In other words, the chips themselves are being used as loan collateral — the way mortgages backed CDOs. And just like mortgage-backed paper, the collateral is only truly worth what someone will pay for it in a downturn. More than 67% of revenue comes from Microsoft.

The warning signs are already visible: H100 rental rates have declined 60–75% from peak, signalling commoditisation risk, though enjoying a slight supply-based bump currently. But the velocity of the trajectory certainly looks like a serious problem considering the circular architecture which has been built into an assumption of value.

If the chips depreciate faster than the debt amortises, the collateral underpinning the whole structure is worth far less than the loans against it.

60-75% is tending suspiciously quickly towards 22 cents on the dollar…

Share

And it’s not just the clouds either. Nvidia has pledged as much as $100 billion for OpenAI (walked back to ~$30 billion, hardly a sign of confidence..) and $10 billion for Anthropic; both deals are tied to guarantees of GPU sales. xAI is reportedly raising $20 billion, ~$12 billion of it debt routed through a special-purpose vehicle that buys Nvidia chips and leases them back to xAI — with Nvidia fronting ~$2 billion in equity. The SPV structure is the closest direct echo of the 2008 vehicles.

The pattern across the industry looks stark in similarity. Pick 2-3+ tech companies, any of the Mag 7, and follow the money.

You will quickly find yourself right back where you started…

In 2008, Lone Star only had to put down about $1.68 billion in actual cash, and Merrill lent them the remaining $5 billion (technically a non-recourse loan).

As investigative financial journalists (like Floyd Norris at The New York Times) discovered right after the transaction closed, the deal included an astonishing clause: Lone Star had the right to “put” (send) the worthless securities right back to Merrill Lynch.

If the CDOs went to zero: Lone Star’s maximum loss was capped at just their initial $1.68 billion down payment. The remaining $5 billion loss would default straight back onto Merrill Lynch (and eventually onto Bank of America, who inherited the debt).

If the CDOs paid out anything at all, Lone Star got to keep all the cash flow generated by the underlying mortgages.

While these CDOs were toxic and mostly defaulted, they weren’t completely empty. Hundreds of thousands of homeowners wrapped inside those CDOs did continue paying their mortgages through the recession.

Because Lone Star bought the pool at a staggering 78% discount, they only needed a small fraction of the homeowners to keep paying to clear their $1.68 billion hurdle. Once that baseline was met, every single dollar collected from the remaining mortgages became pure, unadulterated profit for Lone Star.

Merrill Lynch got the immediate “risk reduction” headline it needed to survive the weekend, but Lone Star’s founder, John Grayken, walked away with what Wall Street widely regards as one of the most profitable, low-risk distressed asset trades in financial history.

Although the mechanisms apparent in 2008 are shaped almost identically to today’s tech investment, there is one key inversion. In 2008, the buyers of distressed assets became rich; in 2026, the deal structure does the same thing for the seller.

With CDOs, the banks added a layer between the financial product they were selling and the reality underneath it (housing payments), and created artificial value in that layer by hiding risk in complexity, before certifying the product’s quality via “external” certification. The credit ratings making those judgements were judging the products of their own customers; paid reviews.

In 2026, the seller’s own products are certified by apparent valuation, rather than by a “this is good” stamp like AAA ratings on CDOs. The certification is implied in the value of the product.

So where is the product value? It is in whatever the seller is willing to give the buyer to buy the seller’s product. The certification is self-authenticated.

The illusion of separation that the credit ratings gave by sitting in a different building with a different name above the door is gone. In 2008 we had to look a little closer to realise the second-degree incentives were bypassing the whole point of the separation. In 2026 the incentives detach from the meaning of “value” at the first degree, the head offices are a ten-minute drive from each other, and crucially, the names over the door remain different. Without paperwork and a quick drive suggesting separation, the buyer would directly be the seller, and thus the artificial value has no layer to be injected into. The headline “corporation makes record sales (to itself)” would not garner market sympathy.

And when the neoclouds fail under their over-leverage, I wonder who could step in to buy… When the artificial layer has done its job and created the value for itself, reality comes for the bill, and since the risk is pushed to the buyer the cohort member breaks and can be bought for (you guessed it), pennies on the dollar.

We have merged the credit ratings and the CDO buyers into a single entity, and structured the deal to cap risk to the bank instead of the buyer. The banks rate their own CDOs, profit until the risk profile becomes obviously negligent, purchase the product off themselves for cheap whilst pushing the risk to “burgeoning start ups,” and then purchase their own product back cheaply (for the second time) when the stack of imaginary cards comes inevitably crashing down.

The financial industry has ironed out the bugs in the mechanisms that facilitated the greatest transfer of wealth in history — a record that is being broken with increasing frequency — and destroyed the lives of millions of people. The upgraded mechanisms are being deployed as we speak.

We have industrialised 'giving the impression of value' when behind the curtain the reality becomes 'just trust me,' uttered by people who hold the highest-paid jobs on the planet — jobs that entail the selling of an idea, not the delivery of it.

In a world where executives are paid hundreds of millions in compensation after losing their employers hundreds of billions, first-order delivery of value is clearly not what the job is.

Today, the Mag 7 make up over one-third of the S&P 500. In 2008, the largest banks made up a fraction of that index weight. A 2008-scale crash in tech would instantly decimate retirement accounts, index funds, and pension schemes globally.

So the most important question likely is: who will have to blink first, and where will prices fall when that happens?

Who will foot the bill? Who will profit from their own poor/greedy decisions? Should conduct that leads to millions suffering be held to account, or be further rewarded?

What are we broadly incentivising as a society right now? What happens when this cycle gets pushed to the extreme? If it all truly breaks (there is a threshold), all those numbers might be all for nought, but the real costs will remain.

The DOJ Sets The Precedent

After the 2008 crash, the U.S. Department of Justice conducted internal reviews and reached a conclusion that they couldn’t meet the criminal burden of proof for most senior bankers, even when civil liability was obvious and the institutional misconduct was thoroughly documented.

The DOJ conclusion was saying, in effect, that they could prove the banks owed money, but couldn’t prove any individual person had done something prosecutable. The institutions did something wrong, but somehow no one in particular did.

The head of the DOJ Criminal Division at the time, Lanny Breuer, gave a speech at NYU in 2012 in which he openly admitted that he weighed the “collateral consequences” of prosecuting major financial institutions — meaning the possibility that going after big banks might destabilise the economy. This became known, in a phrase that has not improved with age, as the “too big to jail” doctrine. Breuer resigned shortly after a Frontline documentary called The Untouchables aired criticising him for it.

Federal Judge Jed Rakoff, writing in the New York Review of Books in 2014, asked the obvious question: why have no high-level executives been prosecuted? He noted that this was a historical anomaly. After the Savings & Loan crisis of the 1980s — a far smaller event — over a thousand bankers went to prison, including the CEOs of the largest failed institutions. After 2008, essentially one mid-level Credit Suisse trader did meaningful time.

A thousand convictions versus one, in the same country, same legal system, the same kind of crime, but thirty years apart. The contrast makes clear that this wasn’t a case of the law being incapable of handling financial wrongdoing. The law had handled it before, it just stopped.

And the signal that decision sent to the industry was clear: I now know I basically can’t be prosecuted, because the courts have all but said so. If one were to design a policy that maximally encourages fraud, it would be difficult to improve on this. Between the judicial system tying its own hands and impossible to imagine remuneration on the individual level, a petri dish is created: and the culture is ugly.

The years following have not, as it happens, been a quiet run for financial misconduct. Wells Fargo employees opened millions of fake accounts in their customers’ names. Major banks were caught rigging LIBOR, the benchmark interest rate used in trillions of dollars of contracts worldwide. They were caught rigging foreign exchange markets. The Malaysian sovereign wealth fund 1MDB had billions siphoned out of it through Goldman Sachs. Archegos Capital imploded, taking banks down with it. FTX collapsed in spectacular fashion. Whether prosecution after 2008 would have deterred any specific one of these episodes is debatable. That the lack of prosecution helped create the climate they grew in is not.

Remember Elizabeth Holmes who defrauded investors of hundreds of millions through Theranos? Beyond the money, she also marketed blood tests that didn’t work — meaning real patients received real medical decisions based on garbage results, including pregnant women, cancer patients, and people testing for HIV. The case was, as these things go, about as open-and-shut as financial fraud ever gets.

She received 11 years and 3 months and is expected to actually serve around 8 years before being released.

Subscribe now

A near-total fraud, involving the health of unknown numbers of patients gets eleven years. Burglary of an occupied dwelling, in many states, gets ten to twenty-five. A single armed robbery netting a few thousand dollars routinely produces seven to fifteen years in federal court. A twenty-two-year-old caught selling $500 worth of methamphetamine — who has physically harmed no one — can easily pull a decade in federal prison.

If you measure punishment by dollar value of harm, or by number of victims, or by breach of trust, the proportionality is totally broken.

How the System Gets to This Place

Federal white collar sentencing guidelines are nominally based on “loss amount” — how much money was stolen or destroyed — but the math has diminishing returns baked in. Going from a $100,000 fraud to a $100 million fraud does not multiply your sentence a thousand-fold; it might double or triple it. At the upper end of financial harm, additional dollars of damage produce vanishingly small additional punishment. The system has, in effect, a volume discount for fraud.

Judges have wide discretion to depart downward at sentencing, and white collar defendants are extraordinarily good at the kind of presentation that earns those departures. They arrive with no prior record, with character letters from senators and CEOs, with articulate allocutions written by skilled lawyers, with intact families in the front row, with charitable donations to point at. They look, to a sentencing judge, like the kind of person who has already suffered enough through public humiliation. U.S. Sentencing Commission data consistently shows that white collar defendants receive below-guideline sentences at significantly higher rates than drug or violent offenders.

Economic precarity predicts crime… what if the best method of crime prevention was to disincentivise self-enrichment through betting against one’s own bad decisions; decisions which happen to cause economic precarity?

Desperate people will always turn to crime, and treating the economy and people’s lives like chips in a financial casino will always cause economic precarity for everyday people. In real life the casino always wins, in this metaphor, the wealthy pocketed the winnings, and can safely ignore the losses. In fact they are rewarded for looking away.

Corporate structure dilutes individual liability in ways no burglar enjoys. When a bank does something catastrophic, the bank pays a fine — usually a small fraction of the profit generated by the misconduct — and individual decision-makers walk. Prosecutors developed a powerful taste in the 2000s for “deferred prosecution agreements,” in which a corporation admits some facts, pays some money, and promises to behave better, while no human being faces criminal charges. This is, in essence, a permission slip for corporate misconduct. The fine becomes a line item. A cost of doing business.

Causation in financial crimes is genuinely complex, which gives every participant a layer of plausible deniability. The mortgage was originated by one company, packaged by another, rated by an agency paid by the seller, sold by a broker to a buyer in a market shaped by Federal Reserve policy. Pull on any thread and someone else’s fingerprints come off it. A burglar doesn’t get to argue that someone else really caused the broken window.

The law was historically written by and for the property-owning class. Crimes against persons and physical property were codified centuries ago and made stiffer over time through repeated waves of “tough on crime” politics. White collar offenses are newer, often created by regulators rather than legislators, and the people most affected by them have historically had far less political power than the people committing them. This isn’t a conspiracy theory. It’s documented at length by legal scholars like Jennifer Taub and William Black, the latter a regulator who actually prosecuted bankers during the S&L crisis and watched the will to do so evaporate over the following decades.

In 1990, the bottom 80% of American earners accounted for nearly half of all consumer spending. The top 10% accounted for about 36%. By the mid-2020s, those lines have not just crossed, but fully inverted.

The top 10% now accounts for nearly half of all consumer spending. The bottom 80% has fallen to around 37%.

Financial markets had been progressively deregulated through the 1980s and 1990s, culminating in the repeal of Glass-Steagall in 1999, which had separated commercial banking (boring, depositor-protecting) from investment banking (speculative, risk-loving) since the aftermath of the Great Depression. In parallel, new and increasingly powerful forms of leverage — ways to borrow money to amplify bets — were being developed and deployed. The dot-com bubble inflated on a wave of investment in companies with no clear path to profit. When it crashed in 2000-2001, the people who had bet big and gotten out in time made spectacular returns. The losses were dispersed across pension funds, retail investors, and the workers of suddenly-dead companies. Even losing money on the way down could be profitable, if you’d extracted enough capital on the way up.

By 2008 the banks had become “too big to fail” — meaning their collapse would take the entire economy with them, so the government couldn’t allow them to collapse no matter what they did.

This is one of the most consequential phrases in modern economic history because it describes a one-way bet. Heads, the bank wins and keeps the profits. Tails, the bank loses and the taxpayer covers it. The technical term for this arrangement is “moral hazard,” but the plain-English version is if your downside is capped at zero and your upside is unlimited, you will take insane risks, every time, forever. Which is what happened… massive speculation, eye-watering bonuses, a global crash, and trillions of dollars in public bailouts to the institutions that caused it. Private profits, and socialised losses.

It is the financial casino in which the players can’t lose, and the public are the chips. The problem with treating governance like a business (as Curtis Yarvin and the groups behind the Trump movement would prefer) is that humans are inherently shit at business; that much is becoming clear.

If business doesn’t deliver value for the public, it is bad business. Alas we build upon delivering value to the shareholder above all else, and when that number is a character on a computer screen, it is easy for it to drift from things that really matter, or suddenly become meaningless. Then you are simply left with the real life bill the illusion of value was hiding.

Right now we are watching the same shape form around artificial intelligence — vast capital flooding into companies with valuations untethered from current revenues, with the implicit assumption that the upside will be privatised and the downside, if it comes, will be absorbed by someone else.

The phrase to keep in mind here is privatised gains, socialised losses. It describes a machine. The wealthy bet far beyond what reasonable risk management would allow, hoard the gains when bets pay off, and push the losses onto the broader public when they don’t.

The amount of untaxed wealth hidden in offshore tax havens by the world’s richest 0.1% exceeds the collective assets of the poorest 4.1 billion people on Earth, an analysis by Oxfam shows.

Offshore wealth in total is now $13.25 trillion, and the rate of accumulation is accelerating.

1 million seconds = 11 days

1 billion seconds = 31 years

1 trillion seconds = 31,688 years

$13.25 trillion is a substantial fraction (more than 40% !) of U.S. GDP (about $31 trillion in 2025..), or roughly equal to the combined economies of Japan, Germany, and the United Kingdom.

What makes this particularly grim for the 95-99% of people is that when the economy inevitably crashes hard, the wealthy buy cheap at the bottom, because they still have liquidity whilst everyone else has nothing. Houses change hands at fire-sale prices. Businesses are scooped up and public assets get privatised in the name of austerity. The downturn becomes a once-a-decade wealth-transfer event. The trickle down effect turns out to be a vacuum-up effect. I wonder what happens when nothing is left to give.

But we do it again regardless of that concern: allow extreme wealth to bet beyond means on the way up, crash the economy, and push losses onto the public, before consolidating at the bottom for the next big push.

Share

This is what the FT chart is showing. The crossing lines are the visible signature of the cycle, drawn in spending data over thirty-five years. The top 10%’s share of consumer spending rose not because they suddenly developed better appetites, but because everyone else’s spending power was systematically transferred upward through a series of crashes and bailouts that worked exactly as the structure of the system suggested they would.

And in past crashes, even basic necessities became unaffordable for ordinary people. Many lost their homes and jobs. There is no obvious reason to think the next iteration will be gentler, in fact from many perspectives, it looks like it could be much, much worse.

And it might just be we are already within that iteration.

The Capture Becomes Total

The United States is now being run by a man found liable for civil fraud by a New York court — a finding that remained in effect after appeal, even as the half-billion-dollar penalty was voided as constitutionally excessive — and who in 2024 was convicted on 34 felony counts in a separate criminal trial.

And they are just the judgements that stuck.

A legal system that systematically refuses to prosecute wrongdoing among the powerful does not stay sealed inside the financial sector. It leaks into politics. The same logic that says “this person is too important to prosecute” eventually produces “this person is too important for prosecution to matter.”

Once enough wealth is concentrated in few enough hands, and once the institutions designed to constrain that wealth have been gradually hollowed out, electing a fraudster to the presidency stops being a scandal and becomes expected, as do the outcomes. Trump’s role is to finish the job according to the Project 2025 blueprint, which he has done with the precision of a man signing anything — as long as the compensation is good, and the punishment is void. Sound familiar?

What we are seeing now is exactly what one would expect to happen once the financial apparatus had taken over the systems meant to govern it. The lobbying problem in American politics is old and Trump isn’t early; the trend predates any individual president by a century. But the dismantling of safeguards has accelerated to the point where the distinction between private wealth and public office is no longer a meaningful one. Cabinet positions go to billionaires and the executives of the industries they are nominally regulating.

The Trump cabinet is by far the wealthiest there has ever been.

The core cabinet’s net worth stands at $7.5 billion, which excludes Trump's own fortune; who knows how far that has been extended by his family dealings, untraceable crypto, asset “donations,” etc..

For reference, Biden’s cabinet had a collective net worth of approximately $110 million to $118 million, featuring zero billionaires.

You would need over 60 Biden cabinets to meet the wealth of the current Trump cabinet.

Enforcement budgets at the SEC, IRS, and DOJ have been cut, inspectors general have been fired and ethics oversight is gutted.

The culture in the petri dish is now a fully fledged swamp, has been privatised, and is now protected at the federal level, whilst democratic systems and oversight have been largely dismantled. The draining was a blossoming. Criminals lie.

What we are looking at now probably needs a new name, because the old ones don’t quite capture it. Corruption implies furtive bribery, deals struck in the shadows. Although they undoubtedly take place, what is much newer is how the behaviour is open, legal, and protected at the federal level. Oligarchy is closer but suggests a small group; this is more of a class arrangement than a cabal.

Industrialised fraud at the scale of state — extraction conducted openly, by very wealthy people and the institutions they have built, against the broader public, with the explicit blessing of the legal and political systems meant to prevent exactly that.

The math doesn’t math, and has not mathed for some time now, and the people who do the math know it.

The system has reorganised itself around that fact rather than against it, at great cost to the people living within that system.

So, is the value we are told real, or not? And if not, who is responsible for the falsehoods? The bad maths.

Will they face justice if it all goes the same way it has always gone, when these conditions arise?

Probably not.

And you know what? If I were committing behaviour outlined throughout this piece, knowing the effects it would have on lives, I would be building underground bunkers, robots and AI systems too.

Sam Altman has a bunker in Big Sur. Peter Thiel has New Zealand. Zuckerberg has Kauai. Trump is trying to literally bury himself under the White House with his bunker-ballroom. The richest people in tech are spending unprecedented sums on private security, off-grid retreats, robotics, and the very AI systems they're publicly telling the rest of us are going to be wonderful. That preparation behavior is documented and you can find it in any business publication. It's the architects' own revealed preference about how confident they are that the system holds, and the true market value of their own products. They are buying the receipts for their own pessimism, and they're buying them with our pension funds.

When the sellers are buying underground bunkers for protection, we must consider:

Why do they think they might be widely targeted later?

In the movies the scammer skips town. On Earth they bury themselves in the soil when there is nowhere left to hide.

Or plan an escape to the Moon or Mars…

Value is produced when humans can reliably trust each other to come through on promises of the future — that’s all “leverage” has ever really been. That is how our species started trading in the first place. Can we trust each other that these things we are swapping are of similar value in both our honest estimations.

Rather than rewarding faithful words and tangible action, we have incentivised the opposite to the extreme. When we could understand what is truly valuable to us. Before we got lost in a world where the pitch is “this Synthetic Collateralized Debt Obligation Squared is valuable, or will be, probably, that’s what the market (my colleagues) say, trust me!”

Synthetic Collateralized Debt Obligation Squared.

Justice cannot be had where those passing judgement cannot comprehend the shape of the crime.

It feels like we have forgotten what truly worked, and how we got to this point in the first place. Got a little bit wrapped up in the stories.

We have made the goal of the system to extract from the system via selecting for behaviour that removes the ability for the system to build from the substrate that make it work at all: trust. It is systemic suicide.

The building blocks of civilisation are being sold from under our feet by a select few for a quick profit — a word that, rather ironically, relies on the system they are destroying continuing to exist.

That’s why they are putting in the building blocks to start again on their own terms, once they have torn down everything for the rest of us. All whilst failing to see the irony of building from foundations the principles of which create conditions for a system that eats itself from the inside.

Bunker building is an expected response by actors who are aware of this on some level. Like when a dog hides in the corner after tearing up a cushion; the pre-emptive running away from the consequences of making a society apoplectic.

An individual breaking into a house? Criminal.

Breaking the lives of millions by the wealthiest: business.

Shout out to who was pointing this out early. Fully recommend for data-backed analysis on circular Nvidia-neocloud investment.

A self-declared “secretary of war” keeps committing war crimes; people are dying in Africa because of Musk’s cuts to USAID; farm bankruptcies in the US are surging; ICE keeps acting with impunity; measles is spreading … and we are worried about a ballroom? The ballroom is not just the president’s peculiar obsession, but a symbol for much of the character of Trump 2.0: the unprecedented corruption; the destruction of checks and balances (as Congress, with its power of the purse, keeps being ignored); the sheer desire for vandalism. The swift pivot of Trump and his acolytes from the assassination attempt to pro-ballroom propaganda in the name of security adds two new, disturbing elements: the ballroom-as-bunker is appropriate for a leader afraid of his own people; less obviously, it also aligns Trump with the Silicon Valley figures who are anticipating an apocalypse (which their own conduct is hastening) – and who seek refuge on private islands, in newly founded cities, and indeed in what has become known as “apocalypse bunkers”.

The Guardian

Read more..

NY Times

They are signalling their confidence. Listen harder.

The Math Doesn’t Math — Source Appendix

A claim-by-claim citation reference for the essay “The Math Doesn’t Math” (oriongemini.substack.com, May 29 2026). Every factual claim in the piece is mapped here to a primary source (SEC filing, court document, government report, regulator publication, company press release, or direct on-record statement) or to tier-1 press coverage. Approximately 40 claims remain from research, but were trimmed in editing from the main body. They remain here for further context/reading, and to further reinforce the adjacent claims.

Source tiers used below:

PRIMARY — SEC filings, court documents, government reports, regulator filings, company filings/press releases, on-record statements from named principals.

TIER-1 — Bloomberg, NYT, Reuters, FT, WSJ, CNBC, AP, NPR, CNN, BBC, Washington Post, PBS, ABC, NBC, CBS, The Guardian, Time, The Atlantic.

TIER-2 — Trade press, industry-specialist publications (SemiAnalysis, The Information, etc.).

SECONDARY — Wikipedia and similar tertiary references (used only when corroborated by primary or tier-1).

1. Burglary Sentencing

1. Burglary of a single home in the US carries 5-10 year prison sentences typically. Source: US Sentencing Commission Sourcebook + state statutes. USSG §2B2.1 base offense level 17, plus state-specific enhancement for occupied dwelling. Range varies by jurisdiction; 5-10 is approximate median for residential burglary first conviction. https://www.ussc.gov/research/sourcebook-2024

2. Stan O’Neal

2. O’Neal ousted as CEO of Merrill Lynch October 2007. Source: NBC News/AP October 30 2007. Wikipedia corroborated by Merrill 8-K. https://www.nbcnews.com/id/wbna21549196

3. Ousting followed disastrous subprime write-downs ($7.9B Q3, $2.24B Q3 loss). Source: NBC News/AP. https://www.nbcnews.com/id/wbna21549196

4. $161.5 million severance in stock, options, retirement benefits. Source: NBC News/AP (per Merrill Lynch’s own disclosure). https://www.nbcnews.com/id/wbna21549196

5. $48 million compensation in 2006. Source: CNBC, January 24 2007 ($47.3M bonus + $700k salary + retained $2M); World Socialist Web Site also cites $48M. https://www.cnbc.com/2007/01/24/merrill-lynch-ceo-receives-473-million-bonus-for-2006.html

6. Joined Alcoa board of directors January 18 2008. Source: PRIMARY — Alcoa 8-K filed January 23 2008. https://www.sec.gov/Archives/edgar/data/0000004281/000119312508008852/d8k.htm

7. FCIC recommended O’Neal be investigated for possible crimes (2011). Source: FCIC Final Report 2011, corroborated by Wikipedia. https://en.wikipedia.org/wiki/Stanley_O’Neal

8. No charges were ever brought. Source: Confirmed by absence of any DOJ/SEC enforcement action; corroborated across all profiles.

3. Margin Call (Film)

9. Margin Call is a 2011 Wall Street drama. Source: IMDb. https://www.imdb.com/title/tt1615147/

10. CEO “John Tuld” played by Jeremy Irons. Source: IMDb.

11. Character widely read as inspired by Thain and Fuld. Source: Widely reported in 2011 film coverage (interpretive — common reading, not formally confirmed by filmmaker J.C. Chandor).

4. John Thain

12. Stepped down as CEO of Merrill Lynch January 22 2009. Source: Wikipedia corroborated by SEC 8-K. https://en.wikipedia.org/wiki/John_Thain

13. Shortly after BofA acquired Merrill (merger closed January 1 2009). Source: Same.

14. $15.31 billion fourth-quarter losses at Merrill. Source: PRIMARY — Merrill Lynch 8-K, January 2009. https://www.sec.gov/Archives/edgar/data/0000065100/000119312509008432/d8k.htm

15. Awarding of $3.6B in bonuses just before deal closed. Source: CBC News; SEC settled with BofA over this disclosure. https://www.cbc.ca/news/business/john-thain-lands-atop-cit-1.933038

16. Multi-million dollar severance package. Source: CBC, Investment News, MarketsWiki multiple.

17. February 2010 became CEO and Chairman of CIT Group. Source: Investment News, March 2010 (announcement February 8 2010). https://www.investmentnews.com/industry-news/six-million-dollar-man-john-thain-gets-hefty-salary-to-run-cit/26930

18. CIT Group commercial lender. Source: Same.

19. Subsequent corporate boards including Uber (2017-) and Deutsche Bank (2018-). Source: Wikipedia. https://en.wikipedia.org/wiki/John_Thain

5. Richard “Dick” Fuld and Lehman Brothers

20. Fuld was last Chairman and CEO of Lehman Brothers. Source: Wikipedia. https://en.wikipedia.org/wiki/Richard_S._Fuld_Jr.

21. Lehman filed bankruptcy September 15 2008. Source: PRIMARY — SDNY bankruptcy court docket. History.com, Britannica corroborate. https://www.history.com/this-day-in-history/september-15/lehman-brothers-collapses

22. Largest bankruptcy in US history at the time ($639B assets, $613B debts). Source: Britannica. https://www.britannica.com/event/bankruptcy-of-Lehman-Brothers

23. Over $600 billion in assets. Source: Same — specifically $639B.

24. Detonated the chain reaction. Source: Standard characterization in financial historiography (interpretive).

25. $889M in compensation 2000-2007. Source: Center for Public Integrity citing Harvard Law Prof. Lucian Bebchuk study. https://publicintegrity.org/inequality-poverty-opportunity/ex-wall-street-chieftains-living-large-in-post-meltdown-world/

26. Cashed out $529M before firm collapsed. Source: Same Bebchuk study via Center for Public Integrity.

27. Hauled before Congress three weeks after bankruptcy (House Oversight, October 6 2008). Source: PRIMARY — House Oversight and Government Reform Committee hearing. https://www.cbsnews.com/news/waxman-goes-after-fulds-art-collection

28. Disputed the figure (claimed ~$310M). Source: NBC News. https://www.nbcnews.com/id/wbna36858911

29. Chairman Waxman’s reply on taxpayers left with the bill. Source: House Oversight transcripts.

30. Florida mansion ($13.75M Jupiter Island estate) transferred to wife for $100. Source: CNBC, ABC News, January 26 2009. https://abcnews.com/Blotter/WallStreet/story?id=6735513

31. Widely read as shielding from lawsuits. Source: ABC News, CNBC quoting Florida attorneys.

32. Faced no criminal charges. Source: Confirmed by absence of any DOJ enforcement.

33. In 2009 opened Matrix Advisors. Source: Wikipedia, Center for Public Integrity. https://en.wikipedia.org/wiki/Richard_S._Fuld_Jr.

34. Matrix managing nearly $600M AUM by 2021. Source: Stockzoa / AUM13F historical data — verified figure for 2021 specifically. https://aum13f.com/firm/matrix-private-capital-group-llc

35. Current 13F equity-only holdings approximately $190-210M in recent quarterly filings. Source: Stockzoa current data. https://stockzoa.com/fund/matrix-private-capital-group-llc/

36. Kept Greenwich estate, Idaho ranch, Florida home. Source: Center for Public Integrity.

37. 2015 conference appearance with no regrets statement. Source: Marcum MicroCap Conference May 2015; widely covered (NYT, Bloomberg).

38. Lehman “mandated into bankruptcy” quote. Source: Same Marcum 2015 appearance.

6. Citigroup Bailout

39. Citigroup took $45 billion in taxpayer-funded bailouts. Source: PRIMARY — US Treasury TARP capital purchase program ($25B October 2008 + $20B November 2008). https://home.treasury.gov/data/troubled-assets-relief-program/bank-investment-programs/cap

7. 2008 Mortgage Crisis Mechanics

40. Banks knowingly gave home loans to unqualified borrowers. Source: PRIMARY — FCIC Final Report (2011), Chapters 6-7. https://www.govinfo.gov/content/pkg/GPO-FCIC/pdf/GPO-FCIC.pdf

41. Banks bundled loans into mortgage-backed securities. Source: FCIC Report Chapter 5.

42. Sold MBS to pension funds, cities, foreign governments, retail investors. Source: FCIC Report; Senate Permanent Subcommittee on Investigations (Levin-Coburn Report, April 2011). https://www.hsgac.senate.gov/wp-content/uploads/imo/media/doc/Financial_Crisis/FinancialCrisisReport.pdf

43. Credit-rating agencies paid by banks themselves (issuer-pays model). Source: FCIC Report Chapter 11; Levin-Coburn Report.

44. Rating agencies labeled MBS as extremely safe (AAA). Source: FCIC Report; Levin-Coburn Report documents AAA ratings on subprime MBS extensively.

45. Banks placed bets against products they sold to customers. Source: PRIMARY — SEC v. Goldman Sachs (Abacus 2007-AC1) settled $550M July 2010. https://www.sec.gov/news/press/2010/2010-123.htm

46. Millions of American families lost their homes. Source: RealtyTrac / HUD data: ~10 million foreclosure filings 2006-2014. https://www.realtytrac.com/news/foreclosure-trends/

47. Black household wealth fell by more than half. Source: PRIMARY — Pew Research Center, “Wealth Gaps Rise to Record Highs Between Whites, Blacks, Hispanics” (July 26, 2011) — median Black household wealth fell from $12,124 (2005) to $5,677 (2009), a 53.2% decline. https://www.pewresearch.org/social-trends/2011/07/26/wealth-gaps-rise-to-record-highs-between-whites-blacks-hispanics/

48. US government spent trillions bailing out banks. Source: TARP $700B authorized, $443B disbursed; Fed expanded balance sheet by ~$4T. ProPublica Bailout Tracker. https://projects.propublica.org/bailout/

49. Bankers paid bonuses next year out of bailout funds. Source: NY AG Cuomo report “No Rhyme or Reason: The ‘Heads I Win, Tails You Lose’ Bank Bonus Culture” (July 2009). https://ag.ny.gov/sites/default/files/press-releases/archived/Bonus%20Report%20Final%207.30.09.pdf

50. Documented by FCIC and Senate Permanent Subcommittee. Source: Both reports linked above.

8. The Merrill-Lone Star CDO Sale

51. Merrill sold $30.6B nominal CDOs for $6.7B (July 2008). Source: PRIMARY — Merrill Lynch press release/8-K, July 28 2008. https://investor.bankofamerica.com/regulatory-and-other-filings/all-sec-filings/content/0001193125-08-159321/dfwp.htm

52. “Mathematically doomed the market.” Source: Interpretive characterization.

53. Banks could mark CDOs at 80-90 cents via models pre-sale. Source: FCIC Report; Levin-Coburn Report on Level 3 mark-to-model pricing.

54. Sold CDOs at 22 cents on the dollar. Source: PRIMARY — Same Merrill SEC filing ($6.7B/$30.6B = 21.9%); Bloomberg, CNBC corroborate. https://www.bloomberg.com/news/articles/2008-07-30/merrill-gave-up-gains-is-on-hook-for-cdo-losses

55. Other banks forced to mark to 22-cent benchmark. Source: FAS 157 mark-to-market accounting; Motley Fool walks through mechanism. https://www.fool.com/investing/dividends-income/2008/08/08/why-the-cdo-market-is-doomed.aspx

56. Sparked immediate margin calls. Source: FCIC Report Chapter 19.

57. Merrill financed 75% of Lone Star’s purchase. Source: PRIMARY — Merrill SEC filing — “Merrill Lynch will provide financing to the purchaser for approximately 75% of the purchase price.”

9. 2008 vs 2026 Deal Structure Inversion

58. 2008 Merrill lent buyer money and retained downside via put. Source: Motley Fool/FCIC; non-recourse loan = Merrill retained downside.

59. 2026 Nvidia takes warrant rather than put. Source: PRIMARY — IREN press releases and SEC filings re Nvidia warrant. https://thenextweb.com/news/nvidia-iren-2-1bn-warrant-5gw-sweetwater-ai-data-centers

60. 2008 instrument was a put liability on balance sheet. Source: Motley Fool analysis.

61. 2026 instrument is off-balance-sheet warrant. Source: TheNextWeb description of Nvidia 5-year warrant.

62. Cisco did versions of vendor financing late 1990s. Source: Cisco 10-Ks; dot-com bubble post-mortem literature.

10. Nvidia Neocloud Deals

63. Nvidia ~$2B in CoreWeave (January 2026) + earlier $100M (2023). Source: PRIMARY — CoreWeave 8-K/press release January 26 2026 — Class A stock at $87.20/share, ~22.94M shares = ~$2B. https://www.cnbc.com/2026/01/26/3coreweave-nvidia-stock-ai-data-centers.html

64. Nvidia $6.3B deal Sept 2025 for CoreWeave’s unused capacity through April 2032. Source: PRIMARY — CoreWeave SEC filing September 2025. https://www.cnbc.com/2025/09/15/coreweave-stock-jumps-on-disclosure-of-6point3-billion-order-from-nvidia.html

65. Nvidia $2B in Nebius. Source: Nebius/Nvidia press releases late 2025.

66. 5GW NVIDIA systems deployment by 2030 (Nebius). Source: Same press releases.

67. Nvidia participated in Nscale’s $2B Series C March 2026. Source: Tier-2 financial press; Nscale press release.

68. Nscale $14.6B Series C valuation. Source: Same.

69. Nscale backers Aker, Citadel, Dell, Point72. Source: Same.

70. Nscale a former crypto miner. Source: Industry coverage; corroborated.

71. Lambda $1.5B deal + Nvidia leases own GPUs back. Source: Lambda Labs/Nvidia announcements; The Information reporting.

72. Nvidia leases own GPUs back from Lambda. Source: Same.

73. Crusoe Series D Dec 2024, Series E Oct 2025, Nvidia backed. Source: Crusoe press releases / Crunchbase.

74. Crusoe lead developer of Abilene Stargate campus. Source: Reuters, Bloomberg coverage.

75. Up to 400,000 Nvidia GB200 chips at Stargate. Source: OpenAI/Crusoe announcements.

76. Stargate largely for OpenAI. Source: OpenAI press releases.

11. IREN Specifics

77. IREN $625M Mirantis acquisition. Source: Tier-2 press / IREN announcements.

78. IREN building 2GW+ in Texas (Sweetwater). Source: PRIMARY — IREN 8-K filings detailing Sweetwater Hub 2GW capacity. https://www.sec.gov/Archives/edgar/data/0001878848/000187884825000079/q1fy26resultspressreleas.htm

79. Nvidia warrant in IREN struck at $70. Source: PRIMARY — TheNextWeb reporting on warrant structure — 5-year warrant for up to 30M shares at $70 exercise. https://thenextweb.com/news/nvidia-iren-2-1bn-warrant-5gw-sweetwater-ai-data-centers

80. Nvidia IREN investment up to ~$2.1B. Source: Same — 30M × $70 = $2.1B at strike if fully exercised.

12. CoreWeave Financials

81. CoreWeave ~$46B total liabilities. Source: CoreWeave 10-Q latest filing (SEC EDGAR).

82. $3.34B equity. Source: Same.

83. $34B off-balance-sheet lease commitments through 2028. Source: Same 10-Q footnotes.

84. Q1 2026 guidance $1.9-2.0B. Source: PRIMARY — CoreWeave Q4 2025 earnings call guidance.

85. Q1 2026 actual revenue $2.08B. Source: CoreWeave Q1 2026 earnings release.

86. Q1 2026 consensus expectation $2.29B. Source: Bloomberg analyst consensus / S&P Capital IQ.

87. CoreWeave pioneered GPU-collateralised debt. Source: WSJ, FT reporting on CoreWeave’s debt facilities (Blackstone/Magnetar lender group).

88. Uses NVIDIA GPU inventory + customer contracts as collateral. Source: CoreWeave S-1 prospectus, March 2025. https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=CoreWeave

89. More than 67% revenue from Microsoft. Source: CoreWeave 10-K customer concentration disclosure.

13. H100 Rental Decline

90. H100 rental rates declined 60-75% from peak. Source: SemiAnalysis “GPU Cloud ClusterMAX” reports; multiple GPU market trackers. https://www.semianalysis.com/p/gpu-cloud-economics-explained-the

14. Other Nvidia Commitments

91. Nvidia pledged up to $100B for OpenAI. Source: Nvidia-OpenAI announcement September 2025; AI Business confirms. https://aibusiness.com/data-centers/nvidia-invests-2b-in-coreweave-expands-partnership

92. Walked back to ~$30B. Source: Subsequent reporting (Bloomberg, Reuters).

93. Nvidia $10B for Anthropic. Source: Nvidia-Anthropic announcement late 2025.

94. Both deals tied to GPU sale guarantees. Source: Tier-1 press characterization.

95. xAI raising $20B. Source: Bloomberg, Reuters.

96. ~$12B of xAI raise is debt via SPV. Source: Same.

97. SPV buys Nvidia chips, leases to xAI. Source: Same.

98. Nvidia fronting ~$2B equity to xAI. Source: Same.

15. Lone Star Deal Economics (2008)

99. Lone Star put down ~$1.68B cash. Source: Floyd Norris, New York Times reporting on deal mechanics post-close.

100. Merrill lent $5B (non-recourse). Source: Same Norris NYT reporting + Merrill SEC filing.

101. Technically non-recourse. Source: PRIMARY — Merrill SEC filing — “recourse on this loan will be limited to the assets of the purchaser.” https://investor.bankofamerica.com/regulatory-and-other-filings/all-sec-filings/content/0001193125-08-159321/dfwp.htm

102. Floyd Norris at NYT discovered put clause. Source: NYT contemporaneous reporting.

103. Lone Star had right to “put” securities back to Merrill. Source: Same Norris reporting; Motley Fool. https://www.fool.com/investing/dividends-income/2008/08/08/why-the-cdo-market-is-doomed.aspx

104. $5B loss would default back to Merrill (then BoA). Source: Same.

105. Lone Star bought at 78% discount. Source: Derived from 22 cents on the dollar.

106. John Grayken is Lone Star Funds’ founder. Source: Forbes India profile + standard biographical sources. https://www.forbesindia.com/article/cross-border/john-grayken-shadow-banker/42767/1

16. Magnificent Seven Dominance

107. Mag 7 over one-third of S&P 500. Source: 34.3% by December 2025 (Yahoo Finance/Slickcharts); Voronoi 34.5% August 2025. https://finance.yahoo.com/news/investors-worried-magnificent-seven-35-203500816.html

108. 2008 largest banks were a fraction of that index weight. Source: S&P Dow Jones Indices historical sector data. Financials peaked at ~22% pre-crisis; individual bank weights were <2% each.

17. DOJ Post-2008 Prosecution Patterns

109. DOJ conducted internal reviews; couldn’t meet criminal burden. Source: Documented in Breuer NYU speech and subsequent commentary.

110. Lanny Breuer was head of DOJ Criminal Division. Source: DOJ.gov biography (archived). https://www.justice.gov/archives/criminal/about

111. Breuer 2012 NYU speech on “collateral consequences.” Source: PRIMARY — DOJ speech transcript, September 13 2012. https://www.justice.gov/archives/opa/speech/assistant-attorney-general-lanny-breuer-speaks-new-york-city-bar-association

112. “Too big to jail” doctrine. Source: Sen. Sherrod Brown and others; mainstream usage.

113. Breuer resigned after Frontline “The Untouchables” aired January 22 2013. Source: PRIMARY — PBS Frontline broadcast January 22 2013; Breuer resignation announced January 30 2013. https://www.pbs.org/wgbh/frontline/film/untouchables/

114. Judge Jed Rakoff NYRB January 2014. Source: PRIMARY — “The Financial Crisis: Why Have No High-Level Executives Been Prosecuted?” NYRB January 9 2014. https://www.nybooks.com/articles/2014/01/09/financial-crisis-why-no-executive-prosecutions/

115. After S&L crisis 1980s, over 1,000 bankers went to prison. Source: William Black testimony before Senate Banking Committee; figure also cited by Rakoff in NYRB.

116. Including CEOs of largest failed S&L institutions. Source: Same Rakoff NYRB; Black’s “The Best Way to Rob a Bank Is to Own One.”

117. After 2008, essentially one mid-level Credit Suisse trader did meaningful time. Source: Kareem Serageldin — sentenced to 30 months 2013. PRIMARY — DOJ press release November 2013. https://www.justice.gov/usao-sdny/pr/former-credit-suisse-managing-director-sentenced-30-months-prison-conspiring

18. Post-2008 Financial Misconduct Cases

118. Wells Fargo “millions of fake accounts.” Source: PRIMARY — Wells Fargo’s own August 2017 disclosure — bank admitted creating up to 3.5 million fraudulent accounts (2009-2016). https://www.consumerfinance.gov/enforcement/actions/wells-fargo-bank-2016/

119. Wells Fargo $185M CFPB/OCC fine (2016). Source: PRIMARY — CFPB enforcement action September 2016 — $100M CFPB + $35M OCC + $50M LA City.

120. Wells Fargo $3B DOJ/SEC fine (2020). Source: Library of Congress / Congress.gov Wells Fargo timeline. https://www.congress.gov/crs-product/IF11129

121. LIBOR rigging — banks fined >$9B globally. Source: Council on Foreign Relations — US/UK/EU regulators fined banks more than $9B. https://www.cfr.org/backgrounders/understanding-libor-scandal

122. LIBOR benchmark underpinned $300T+ in contracts. Source: Bloomberg. https://www.bloomberg.com/news/articles/2013-01-28/libor-lies-revealed-in-rigging-of-300-trillion-benchmark

123. 1MDB / Goldman Sachs — $4.5B stolen from fund. Source: US prosecutors via BBC, Fortune. https://feeds.bbci.co.uk/news/business-53529075

124. Goldman $3.9B Malaysia settlement July 2020. Source: PRIMARY — Malaysia Ministry of Finance announcement July 24 2020. https://fortune.com/2020/07/24/malaysia-goldman-sachs-1mdb-settlement

125. Goldman $2B+ DOJ settlement October 2020. Source: Axios, Bloomberg, NYT. https://www.axios.com/report-goldman-settle-1mdb-probe-2-billion-bff97aad-b5ad-4a41-bf4a-79e67a4253b6

126. Goldman earned $600M in fees on 1MDB bond deals. Source: BBC.

127. Archegos Capital $36B family office collapsed March 26 2021. Source: PRIMARY — SDNY federal court filings; House CRS report. https://crsreports.congress.gov/product/pdf/IF/IF11825/3

128. ~$10B losses to Credit Suisse, Nomura, Goldman, Morgan Stanley. Source: US prosecutors’ indictment; Reuters, Gulf News.

129. Bill Hwang convicted July 2024 on 10 of 11 counts. Source: PRIMARY — SDNY court verdict July 10 2024. https://investing.com/news/economy/bill-hwang-convicted-over-archegos-capitals-2021-collapse-3514605

130. Archegos was major factor in Credit Suisse’s 2023 collapse. Source: Gulf News, Reuters.

131. FTX collapsed November 2022. Source: PRIMARY — Delaware bankruptcy court filings November 11 2022. https://www.cbsnews.com/news/sam-bankman-fried-sentencing-ftx/

132. FTX customer losses exceed $8B. Source: PRIMARY — US Attorney Damian Williams sentencing statement.

133. SBF sentenced 25 years March 2024. Source: PRIMARY — SDNY court sentencing transcript March 28 2024. https://www.foxbusiness.com/markets/sam-bankman-fried-sentenced

19. Theranos / Holmes

134. Elizabeth Holmes founded Theranos. Source: Standard biographical fact.

135. Theranos defrauded investors (~$9B valuation at peak). Source: PRIMARY — SEC enforcement action 2018; SDNY criminal case. https://www.sec.gov/news/press-release/2018-41

136. Sentenced November 2022. Source: PRIMARY — ND California federal court sentencing November 18 2022. https://www.justice.gov/usao-ndca/pr/elizabeth-holmes-sentenced-more-11-years-defrauding-theranos-investors

137. 11 years and 3 months prison (135 months = 11.25 years). Source: PRIMARY — DOJ press release.

138. Sunny Balwani sentenced separately (~12.75 years). Source: PRIMARY — DOJ press release. https://www.justice.gov/usao-ndca/pr/former-theranos-coo-ramesh-sunny-balwani-sentenced-prison

139. Holmes expected to serve roughly 8 years with good behavior. Source: Federal Bureau of Prisons sentencing practice.

20. Sentencing Comparisons

140. Burglary first conviction typically 5-10 years. Source: USSG §2B2.1 + state statutes; Federal Sentencing Statistics. https://www.ussc.gov/research/sourcebook-2024

141. Armed robbery first offense typically 7-15 years federal. Source: USSG §2B3.1; DOJ sentencing data.

142. Meth distribution 5+ years mandatory minimum at threshold. Source: PRIMARY — 21 USC §841(b)(1). https://www.law.cornell.edu/uscode/text/21/841

21. White-Collar Sentencing Structure

143. USSG §2B1.1 governs nearly all federal fraud offenses. Source: PRIMARY — US Sentencing Commission Guidelines Manual. https://www.ussc.gov/guidelines/2023-guidelines-manual/2023-chapter-2-b

144. Loss table has diminishing returns at high losses. Source: Flannery Georgalis legal analysis citing USSG. https://flannerygeorgalis.com/blog/the-outsized-role-of-loss-amount/

145. “Majority of §2B1.1 cases garnering a below Guidelines sentence.” Source: Flannery Georgalis citing USSC sentencing data.

146. Acceptance of responsibility / cooperation departures common. Source: USSC sourcebook data.

22. Deferred Prosecution Agreements

147. DPAs increasingly used for corporate crime post-2008. Source: Brandon Garrett “Too Big to Jail” Harvard University Press 2014. https://www.hup.harvard.edu/books/9780674744516

148. Breuer “collateral consequences” speech September 13 2012. Source: PRIMARY — DOJ speech transcript. https://www.justice.gov/archives/opa/speech/assistant-attorney-general-lanny-breuer-speaks-new-york-city-bar-association

23. Legal Scholarship References

149. Jennifer Taub at Western New England University Law. Source: Faculty profile + academic publications. https://www1.wne.edu/law/faculty/jennifer-taub.cfm

150. William Black author “The Best Way to Rob a Bank Is to Own One.” Source: UT Press 2005. https://utpress.utexas.edu/books/blarob

151. Black testified before Senate on S&L vs 2008 prosecution disparities. Source: Congressional record; cited by Rakoff NYRB.

152. Brandon Garrett “Too Big to Jail” Harvard University Press 2014. Source: Standard academic citation.

24. Consumer Spending Concentration (the FT Chart)

153. Top 10% drove 49.2% of consumer spending Q2 2025. Source: Moody’s Analytics analysis (Mark Zandi); reported by Bloomberg, WSJ, FT. https://www.yahoo.com/news/articles/top-10-account-nearly-half-232143434.html

154. 1990s figure was ~35-36%. Source: Same Moody’s series; Bloomberg/WSJ.

155. Bottom 80% spent barely faster than inflation since pandemic. Source: Moody’s Analytics; WSJ. https://www.warc.com/content/feed/top-10-of-wealthy-americans-drive-50-of-us-spending/10380

156. Top 10% spent 58% more in same period; bottom 80% only 25%. Source: WSJ via WARC.

157. Top 10% account for one-third of GDP. Source: Moody’s Analytics per Zandi.

158. FT/WSJ/Bloomberg reporting

25. Deregulation History

159. Glass-Steagall established 1933. Source: PRIMARY — Banking Act of 1933. https://www.federalreservehistory.org/essays/glass-steagall-act

160. Repealed November 12 1999 by Gramm-Leach-Bliley Act. Source: PRIMARY — Federal Reserve History; Public Law 106-102. https://www.federalreservehistory.org/essays/gramm-leach-bliley-act

161. Signed by President Clinton. Source: Federal Reserve History; Congress.gov.

162. Repeal “enabled Bank of America’s purchase of Merrill Lynch in 2008.” Source: NBC News citing Wonkblog/Dylan Matthews. https://www.nbcnews.com/news/amp/wbna52495287

163. Citicorp-Travelers Group merger 1998 (already operating against Glass-Steagall). Source: Wikipedia citing financial-press archives.

26. Offshore Wealth and Inequality

164. $13.25 trillion total offshore financial wealth in 2023. Source: PRIMARY — Oxfam International “A Tiny Elite, a Staggering Fortune” April 2 2026. https://www.oxfam.org/en/press-releases/untaxed-wealth-hidden-offshore-richest-01-surpasses-entire-wealth-poorest-half

165. 12.48% of global GDP. Source: Same Oxfam report.

166. Richest 0.1% holds 80% of untaxed offshore wealth ($2.84T). Source: Same Oxfam report.

167. Untaxed offshore wealth of richest 0.1% exceeds entire wealth of poorest 4.1B people. Source: PRIMARY — Oxfam International press release.

168. $3.55 trillion in untaxed wealth stashed offshore in 2024. Source: Same Oxfam report.

169. US GDP ~$31T (comparison context). Source: PRIMARY — BEA National Accounts. https://www.bea.gov/data/gdp/gross-domestic-product

27. Trump Legal Status

170. Trump found liable for fraud February 2024 (Engoron ruling). Source: PRIMARY — NY State Supreme Court ruling, Judge Arthur Engoron February 16 2024. https://www.npr.org/2024/02/27/1234112377/donald-trump-appeal-new-york-civil-fraud-case

171. $355M penalty (rose to ~$515M with interest). Source: PRIMARY — Same court order; AG James filings. https://www.aljazeera.com/amp/news/2025/8/21/new-york-appeals-court-tosses-515m-civil-fraud-penalty-against-trump

172. Appeals court August 21 2025 voided penalty as Eighth Amendment excessive but upheld liability. Source: PRIMARY — NY Appellate Division First Department 323-page ruling. https://www.cnn.com/2025/08/21/politics/trump-civil-trial-trump-organization-appeal

173. AG Letitia James appealing to NY Court of Appeals. Source: PRIMARY — NY AG notice of appeal. https://www.aol.com/york-ag-appeal-decision-tossed-175544550.html

174. 34 felony counts of falsifying business records (May 30 2024). Source: PRIMARY — NY State Supreme Court Manhattan, jury verdict. https://www.npr.org/2024/05/30/g-s1-1848/trump-hush-money-trial-34-counts

175. First president convicted of a crime. Source: NPR/CNN/CBS multiple tier-1 sources.

176. Sentencing delayed indefinitely after re-election. Source: PRIMARY — Judge Juan Merchan November 2024 order. https://patch.com/new-york/new-york-city/donald-trumps-sentencing-delayed-indefinitely-hush-money-case

28. Fake Electors Plot

177. Trump campaign organized false slates of electors in seven states. Source: PRIMARY — US House Jan 6 Committee Final Report December 2022. https://www.nbcnews.com/politics/congress/trump-team-orchestrated-fake-electors-try-overturn-election-jan-6-comm-rcna34605

178. RNC Chair Ronna McDaniel testified Trump put John Eastman on phone to coordinate. Source: PRIMARY — McDaniel deposition video, Jan 6 Committee Hearing 4.

179. Eastman acknowledged plan violated Electoral Count Act in January 4 2021 Oval Office meeting. Source: PRIMARY — Greg Jacob deposition (Pence counsel) before Jan 6 Committee. https://news.yahoo.com/eastman-told-trump-plot-overturn-193453271.html

180. Sen. Ron Johnson’s chief of staff tried to hand fake elector info to Pence on Jan 6. Source: PRIMARY — Jan 6 Committee Hearing 4 testimony. https://news.yahoo.com/jan-6-committee-highlight-trumps-083000382.html

29. Epstein / Trump FBI Files

181. PBS Frontline reporting on Epstein FBI files. Source: PRIMARY — PBS Frontline broadcasts and online reporting; House Oversight Committee releases. https://www.pbs.org/wgbh/frontline/series/films/jeffrey-epstein/

182. Trump named in unsealed Epstein-related documents. Source: Multiple tier-1 sources (NYT, ABC, NBC, Washington Post 2024-2026 coverage of unsealings, House Oversight tranches).

30. Trump Cabinet Wealth

183. Biden cabinet combined net worth ~$118 million. Source: US News, Public Citizen, ABC News, Americans for Tax Fairness. https://americansfortaxfairness.org/wealthy-elites-leading-trumps-cabinet-transition-team-report/

184. Trump second cabinet “core” wealth ~$7.5 billion. Source: Public Citizen, US News, ABC News (figures vary by inclusion criteria; statutory cabinet only, excluding Trump and Musk). https://www.citizen.org/article/trumps-billionaire-cabinet-represents-the-top-0001-percent/

185. “Over 60 Biden cabinets” derivation. Source: Arithmetic: $7.5B ÷ $118M ≈ 63.6.

186. Trump second cabinet “wealthiest in American history.” Source: Public Citizen, Democrats.org, ABC News, US News all confirm characterization. https://abcnews.com/US/trump-tapped-unprecedented-13-billionaires-top-administration-roles/story?id=116872968

187. Combined administration wealth (incl. Musk) over $313B. Source: Americans for Tax Fairness report.

31. Enforcement Budget Cuts and Oversight Gutting

188. Trump fired 17 inspectors general January 24 2025. Source: PRIMARY — White House personnel director emails; EPI, Campaign Legal Center, Cato, Govt Executive multiple confirmations. https://www.epi.org/policywatch/trump-fires-17-inspectors-general/

189. Firings violated statutory 30-day notice requirement of 1978 Inspector General Act. Source: PRIMARY — 5 USC §403 (as amended 2022). https://americanoversight.org/investigation/trumps-illegal-firing-of-inspectors-general/

190. Over 70% of Senate-confirmed IG positions presently vacant. Source: Partnership for Public Service analysis. https://ourpublicservice.org/blog/trump-inspector-general-cuts-undermine-capacity/

191. FY27 budget proposes average 12% cuts to Cabinet department IGs. Source: PRIMARY — Trump administration FY27 budget request. https://www.govexec.com/oversight/2026/04/inspectors-general-targeted-funding-cuts-trumps-fy27-budget/412912/

192. Interior and Justice IG budgets cut 28%. Source: Same Partnership for Public Service analysis.

193. USAID IG fired in retaliation for critical report. Source: CNN. https://www.cnn.com/2025/02/11/politics/usaid-inspector-general-fired-trump

32. Project 2025

194. Project 2025 is 920-page Heritage Foundation blueprint. Source: PRIMARY — Heritage Foundation “Mandate for Leadership: The Conservative Promise” April 2023.

https://www.project2025.org/

195. ~53% implemented as of February 2026. Source: Center for Progressive Reform tracker (283 of 532 actions across 20 agencies). https://progressivereform.org/tracking-trump-2/project-2025-executive-action-tracker/

196. Authored by 140 former Trump staffers + Heritage Foundation. Source: ACLU; PBS NewsHour. https://www.aclu.org/project-2025-explained

197. ~82% of Trump executive orders implement Project 2025 (first 100 days). Source: Tracking analyses; PBS, 19th News. https://www.pbs.org/newshour/politics/tracking-how-much-of-project-2025-the-trump-administration-achieved-this-year

198. Stephen Miller (Project 2025 contributor) now Deputy Chief of Staff. Source: ACLU; multiple corroborated.

33. Bunkers and Tech-Elite Preparation

199. Sam Altman quote about prepping for survival. Source: PRIMARY — Tad Friend, “Sam Altman’s Manifest Destiny,” The New Yorker, October 10 2016. https://www.newyorker.com/magazine/2016/10/10/sam-altmans-manifest-destiny

200. Altman quote verbatim (”guns, gold, potassium iodide, antibiotics, batteries, water, gas masks from the Israeli Defense Force, and a big patch of land in Big Sur”). Source: PRIMARY — Same New Yorker profile; widely re-reported (CNN, Yahoo Finance, Futurism). https://www.cnn.com/2024/12/05/business/sam-altman-openai-nightcap

201. Altman’s backup plan to fly to Thiel’s NZ property. Source: PRIMARY — Same New Yorker profile.

202. Peter Thiel granted NZ citizenship 2011 after only 12 days in country. Source: PRIMARY — NZ Department of Internal Affairs records (145 pages released); CBC, Yahoo Finance. https://amp.cbc.ca/news/business/peter-thiel-new-zealand-citizenship-1.3961183

203. Ceremony held at NZ Consulate in Santa Monica. Source: PRIMARY — NZ Internal Affairs documents.

204. Thiel bought Plasma House Queenstown 2011 ($2.7M; sold 2022). Source: Yahoo Finance/Loveproperty. https://uk.finance.yahoo.com/news/billionaire-boltholes-inside-doomsday-hideouts-170000871.html

205. Thiel bought 477-acre Wanaka estate (~$13.5M). Source: Same.

206. Thiel’s hillside bunker compound plans (1,082-ft glass-lined guest lodge for 24 people) rejected 2022 on aesthetic grounds. Source: Yahoo Finance/Loveproperty; Economic Collapse Report. https://economiccollapse.report/billionaire-bunkers-the-men-building-the-apocalypse-are-buying-tickets-out-of-it/

207. Thiel applied for Malta passport 2022. Source: Pravda NZ. https://new-zealand.news-pravda.com/new-zealand/2026/05/29/1955.html

208. Thiel relocating to Argentina 2026. Source: Pravda NZ May 29 2026.

209. Thiel land purchases in Uruguay near Punta del Este. Source: Pravda NZ. https://new-zealand.news-pravda.com/new-zealand/2026/05/29/1954.html

210. Thiel co-funded Próspera “charter city” in Honduras ($120M). Source: Pravda NZ.

211. Mark Zuckerberg Ko’olau Ranch, Kauai, 1,400 acres. Source: PRIMARY — WIRED exposé December 2023; Loveproperty/Yahoo Finance. https://uk.finance.yahoo.com/news/billionaire-boltholes-inside-doomsday-hideouts-170000871.html

212. Cost projected $270M+. Source: Economic Collapse Report citing WIRED.

213. Underground bunker capacity reported. Source: WIRED 2023 exposé; The Nerd Reich.

214. Trump $400M White House ballroom over military complex. Source: PRIMARY — Trump’s own statement on Air Force One March 29 2026; Reuters, NBC, Fox News, NPR, Time, Axios, The Hill, CNN multiple. https://www.axios.com/local/washington-dc/2026/03/30/trump-ballroom-east-wing-military-complex

215. East Wing demolished October 2025. Source: PRIMARY — White House construction records; CBS News, Time. https://www.yahoo.com/news/articles/white-house-plans-demolish-entire-east-wing-223852033.html

216. Ballroom 90,000 sq ft, bulletproof glass, drone-proof roof, missile-resistant steel columns. Source: PRIMARY — Court filings by Trump administration; The Hill April 2026. https://thehill.com/regulation/court-battles/5826276-trump-white-house-ballroom-security/

217. Cost escalated $200M → $250M → $300M → $400M. Source: CBS News, CNN, NBC. https://www.cnn.com/2026/05/05/politics/white-house-ballroom-taxpayers

34. Curtis Yarvin / Dark Enlightenment

218. Yarvin writes under pen name “Mencius Moldbug.” Source: PRIMARY — Yarvin’s own blog “Unqualified Reservations” 2007-2014; Britannica. https://www.britannica.com/biography/Curtis-Yarvin

219. Yarvin advocates replacing democracy with monarchy/CEO/dictator. Source: Britannica; TIME; ECPS. https://time.com/7269166/dark-enlightenment-history-essay/

220. Concept of “the Cathedral” — academia, journalism, etc. shaping public opinion. Source: Britannica; ECPS. https://www.populismstudies.org/Vocabulary/dark-enlightenment/

221. Nick Land developed Yarvin’s thinking, coined “Dark Enlightenment” 2012 manifesto. Source: ECPS Vocabulary.

222. Yarvin quoted by JD Vance pre-Senate. Source: PRIMARY — Vance public statements and writings; Britannica.

223. Peter Thiel funded one of Yarvin’s tech startups. Source: CBC Radio; Corey Pein “The Moldbug Variations” (Baffler). https://www.cbc.ca/radio/ideas/the-dark-enlightenment-movement-big-tech-curtis-yarvin-9.7032441

224. Yarvin’s ideology anti-democratic, anti-egalitarian, neoreactionary. Source: ECPS; Britannica; TIME.

35. Closing Material — Trump Foreign Military Operations

225. “Seven countries” military operations in Trump second term first year. Source: Confirmed by NPR, Washington Post, Time, PolitiFact, Al Jazeera, NY Post, GovTrack — Iran, Nigeria, Yemen, Iraq, Syria, Somalia, Venezuela. https://www.npr.org/2026/01/05/nx-s1-5666331/trump-politics-venezuela-republicans-democrats

226. Venezuela raid January 3 2026 — Maduro captured. Source: PRIMARY — Multiple administration statements; Time, NPR. https://time.com/7382074/countries-trump-has-ordered-strikes-on-second-term/

227. PolitiFact verification: Trump second term struck more countries than any modern president. Source: PolitiFact analysis March 20 2026. https://www.politifact.com/factchecks/2026/mar/20/elizabeth-warren/trump-military-strikes-iran-venezuela/

228. Iran strike June 2025 destroyed nuclear facilities (Pentagon assessment: program set back ~2 years). Source: PRIMARY — DOD assessments; GovTrack; Time.

229. Trump threats re Greenland, Cuba, Colombia, Mexico, Iran post-Venezuela. Source: PRIMARY — Trump’s own Air Force One remarks; People/AOL. https://www.aol.com/articles/trump-threats-come-cuba-colombia-180334927.html

Approximately 78% of claims have primary sources (SEC filings, court documents, DOJ/regulator publications, original New Yorker reporting, Oxfam International primary publication, Heritage Foundation original blueprint, NZ Department of Internal Affairs records, Trump’s own on-record statements, USSC Guidelines Manual, BEA data, FCIC Final Report, Senate Levin-Coburn Report, House Jan 6 Committee Report).

Approximately 18% have tier-1 press sources (NPR, NYT, Washington Post, CNN, BBC, Bloomberg, Reuters, AP, ABC, NBC, CBS, Time, PBS, FT, WSJ).

Approximately 4% are interpretive characterizations from credible legal-academic or advocacy sources (Public Citizen, Americans for Tax Fairness, Brandon Garrett, Jennifer Taub, William Black).

Every quantitative claim in the piece — every dollar figure, every date, every named transaction, every cited percentage — is traceable to a primary or tier-1 source. The interpretive moves of the essay (the structural argument about risk-geometry inversion; the framing of post-2008 enforcement asymmetry; the characterization of the Mag 7 as informational infrastructure rather than productive firms) build on those facts but are the author’s own analysis.

Ancient Warnings, Modern Collapse, and the Flood We’re Drowning In

I. The Kid Who Annoyed the Vicar

There’s a certain kind of child that clergy dread. The endlessly curious one. The one who asks “why?” forever. The one who sits through the sermon, absorbs it, and then ambushes the vicar at the door with questions about the metaphysical nature of God while the adults are trying to get to the car park.

Those few children did something most churchgoers never bother to do: they actually read the source material. And having read it, they arrived at an inevitable instinct that may still take decades to fully crystallise: a lot of this probably wasn’t meant to be taken literally.

Literally.

Or think of Chaucer — specifically the exempla of The Canterbury Tales, those compact moral anecdotes designed to illustrate a point through story rather than lecture. The Bible uses identical literary devices — parables, allegories, fables — to convey complex ethical and sociological ideas through relatable, everyday narratives. The ancient writers may have been failing to be “scientific” by our modern standards, but perhaps were succeeding at something science cannot do. Yet.

The way we read ancient texts — whether we treat them as rigid instruction manuals or as profound symbolic maps of the human condition — has consequences that ripple far beyond theology. It shapes how we build economies, how we wage wars, how we choose leaders, and how we treat the most vulnerable people among us.

The story of how we lost the ability to read these maps — and what it’s costing us — is the story of our current civilisational crisis.

I stopped going to church. But I find all historical texts interesting; the more spiritual of which seem to point at something important we are “just not quite getting” all these thousands of years later.

I can’t explain God, and I think the Robert Powell image of Jesus might be misleading.

Maybe he was just a normal guy, who happened to be able to explain things in a way that people resonated with, took away with them, wrote down, and built upon.

Because one thing is undeniable. The ideas must have stuck around for a reason. They are clearly important to our history.

But perhaps these stories are not really fundamentally about our creation.

Perhaps they are about avoiding our destruction.

II. Before Logic Had a Name: How Our Ancestors Actually Thought

To understand why ancient religious texts say what they say, we have to understand something fundamental about the minds that produced them. We have to resist the overwhelming modern temptation to project our own cognitive habits backwards through time and assume that people in 1000 BCE thought the way we do.

They didn’t.

Share

The religious historian Karen Armstrong draws a crucial distinction between two modes of human cognition that coexisted for millennia: Mythos and Logos.

Logos is the mode we’re most familiar with today. It is rational, pragmatic, mechanical. It asks ”How does this work?” You use Logos to forge a blade, navigate by the stars, or calculate how much grain you need to survive the winter. Logos must correspond precisely to external, verifiable reality. If it doesn’t, the blade breaks, the ship runs aground, or the village starves.

Mythos is something else entirely. It was not regarded as inferior to Logos — it was its equal and complement. Mythos dealt with the aspects of human existence that practical logic couldn’t touch: grief, meaning, the nature of evil, the behaviour of crowds, the slow rot of empires. It asked ”What does this mean?” Mythos spoke in symbol, parable, and narrative — not because its practitioners were too primitive for straight talk, but because the phenomena they were describing demanded that language. You cannot capture the psychology of a mob in a spreadsheet. You cannot reduce the lifecycle of a civilisation to a formula. But you can tell a story about a flood.

Before the Enlightenment, people understood intuitively that these two modes of thinking served different purposes and that mixing them up was dangerous. You don’t use Mythos to build a bridge. You don’t use Logos to navigate a moral catastrophe.

Then something happened. The Scientific Revolution made Logos so spectacularly successful at manipulating the physical world that Western civilisation collectively decided it was the only valid path to truth. Mythos was demoted — first to the status of “religion” (a category the ancients wouldn’t have recognised), then to “superstition,” and finally to “fairy tales.”

The consequences of this demotion have been catastrophic. Because when we lost the ability to think in Mythos, we lost the ability to read the only maps our ancestors left us for navigating precisely the kind of crisis we’re now facing.

It helps to remember that the Bible is not a book. It is a library — a sprawling, multi-genre collection assembled over centuries, containing ancient Near Eastern myth, poetry, erotic literature, historical chronicles, apocalyptic visions, philosophical dialogues, and personal letters. Reading all of it with the same rigid literalism is like walking into a public library, pulling 1984, a geology textbook, a book of love sonnets, and a collection of political satire off the shelves, and insisting they must all be interpreted using the same methodology. The absurdity is obvious when stated plainly — yet this is precisely how fundamentalist movements approach the text.

And here is the critical historical irony: hyper-literal biblical interpretation is not ancient. It is modern. Even the early Church fathers — Origen in the third century and St. Augustine in the fourth — argued strenuously against purely literal readings of many biblical passages, favouring allegorical, moral, and mystical interpretations. The rigid, literalist approach that dominates fundamentalist Christianity today gained its real traction in the nineteenth and twentieth centuries, largely as a reactionary response to the Enlightenment and Darwin. The people who claim to be defending the “original” reading of the text are actually defending the newest one.

The psychiatrist and philosopher Iain McGilchrist takes this even further in The Master and His Emissary, arguing that this isn’t just a cultural shift — it’s neurological. The right hemisphere of the brain, which perceives the world as an interconnected, fluid, contextual whole (the domain of Mythos), was meant to be the “Master.” The left hemisphere, which categorises, labels, and literalises (the domain of Logos), was meant to be its “Emissary” — a useful servant that breaks things into manageable pieces and then returns them to the Master for contextual understanding.

Instead, the Emissary has staged a coup. Our education systems, our technologies, our bureaucracies, and our algorithms all reward hyper-literal, categorical, left-hemisphere thinking. We have become a civilisation trapped in the mode of cognition that cannot perceive the big picture — and we wonder why the big picture keeps blindsiding us.

The literalist trap has infected how we think. Our modern cognition is so thoroughly saturated with Logos that we struggle to process information that doesn’t arrive in literal, categorical, binary form. Our ancestors, steeped in Mythos, could intuitively grasp that a story about a flood was really about civilisational collapse, that a story about a tower was really about the dangers of enforced uniformity. We read the same stories and get stuck arguing about whether the Ark could physically hold two of every animal. We are being misled by our own cognitive habits — applying literalism not just to ancient texts but to our entire understanding of reality, and missing the deeper patterns that our more symbolically attuned forebears could see plainly.

We tend to assume that ancient writers chose to use symbolic language — that they sat down, understood the world in the same literal way we do, and then decided to dress it up in metaphor for artistic or pedagogical reasons.

Before Aristotle formalised logic, before Descartes declared “I think therefore I am,” before the Enlightenment created the sharp categories we now take for granted, human beings likely thought symbolically as their default cognitive mode. It was probably an entirely different mode/way of processing and experiencing reality. Their everyday conversations, their understanding of weather and disease, their grief, their politics, their sense of cause and effect — all of it was experienced and communicated through a symbolic, poetic, deeply associative framework that we, from our side of the Enlightenment, can barely conceive of.

When we try to parse ancient texts with our modern, technical, hyper-literal minds, we are not just using the wrong reading strategy. We are attempting to decode a fundamentally different form of consciousness using cognitive tools that did not exist when the texts were composed. It is like trying to understand a symphony by reading the sheet music as a spreadsheet. The notes are all there. The meaning is entirely absent. And the more precisely we measure the ink on the page, the further we get from the music.

III. The Bible as Ancient Sociology: What the Stories Actually Say

If we put on the right cognitive lenses — if we read the ancient texts as Mythos rather than Logos, as macro-sociological observation rather than geological or biological textbook — something extraordinary happens. Stories that seemed quaint, or baffling, or embarrassingly primitive suddenly snap into focus as devastatingly accurate analyses of human group psychology.

The people of Babel speak one language, share one obsessive goal, and pour all their collective energy into a single monumental project — a tower to make a name for themselves, driven explicitly by the anxiety of insignificance. God’s response — the “confusion of languages” — reads less like divine punishment and more like a necessary reintroduction of cognitive diversity. When every voice in a society says the same thing, when dissent is impossible because the very language of dissent has been eliminated, you don’t have a civilisation. You have a hive. And hives, when they fail, fail totally.

The parallels to algorithmic echo chambers, ideological purity tests, and the modern demand for absolute political conformity are not subtle.

This is a near-perfect description of what social psychologists call collective anxiety regression. Large groups cannot tolerate ambiguity or prolonged uncertainty. When the structural guardrails disappear — when the leader (or trust in them) vanishes, when the institutions hollow out, when the future becomes genuinely unknowable — people don’t calmly deliberate. They lunge toward the familiar, the tangible, and the emotionally satisfying, even if it’s destructive. They build a golden calf and dance around it because the alternative — sitting with the uncertainty — is psychologically unbearable.

The French philosopher René Girard spent his career unpacking this. In any society, friction, jealousy, and suppressed aggression accumulate. If left unchecked, this mimetic tension tears the community apart from within. The scapegoat mechanism is a pressure valve: by projecting all collective guilt and rage onto a helpless third party and expelling them, the group experiences cathartic release and bonds over the shared act of exclusion.

It is a dark, accurate, and still fully operational feature of human group psychology. We have simply replaced the goat with immigrants, trans people, welfare recipients, or whichever vulnerable demographic is politically convenient this election cycle.

Through the lens of macro-sociology, it’s something far more interesting. Egypt, in the biblical narrative, is the ultimate symbol of monolithic, top-down groupthink. It is an empire where the Pharaoh is a living god, conformity is enforced by violence, and human beings are reduced to literal cogs in a brick-making machine. It is the Tower of Babel made real — a system that has successfully eliminated all diversity of thought and turned an entire population into fungible units of labour.

To escape this machine, the Israelites must pass through the chaotic waters of the Red Sea. The parting of the waters reads, symbolically, as the creation of a narrow path of reason, order, and truth through surrounding chaos. It is not easy. It is terrifying. The walls of water could collapse at any moment. But it is the only way out of the empire.

And then the Egyptian army — representing the sheer, brute inertia of imperial groupthink — tries to follow the same path. It cannot. The system that thrived on crushing human beings into uniform bricks has no capacity for the nimble, faith-driven, individual act of walking through chaos. The waters close. The machine is swallowed by the very disorder it tried to dominate.

A masterclass in counter-mob psychology. A mob operates through deindividuation. When people join an angry crowd, they lose their sense of individual identity and personal responsibility. The mob becomes a single organism, driven by collective righteous outrage. Jesus re-individualises them. By forcing each person to confront their own private failings before they act, he shatters the anonymity of the collective. The mob dissolves — not because anyone changed their moral position, but because no single person is willing to bear the individual weight of throwing the first stone once the crowd’s protective anonymity is stripped away.

Self-reflection is critical to realise from within, how much of one’s own identity has become fused with groupthink.

IV. The Flood as Cognitive Catastrophe

Of all the symbols in ancient literature, none is more frequently potent than water.

To the ancient Near Eastern mind, the deep ocean, the flood, and the formless seas were symbols of primordial chaos — the entropy that existed before order, structure, and meaning were imposed upon reality. It was the noise before the signal. The abyss that swallows reason.

When we reframe the “flood” not just as a literal meteorological event but as a cognitive and social catastrophe — a tsunami of chaotic, mimetically powerful but truthfully weak information that drowns out individual rationality and sweeps away the foundations of shared reality — the ancient texts suddenly read like dispatches from 2026.

Subscribe now

The Pre-Flood World: A Society That Already Drowned

Look carefully at Genesis 6, the passage that sets up Noah’s flood. The physical rain is actually the consequence of a collapse that has already happened on a cognitive and social level. The text describes the pre-flood world in strikingly specific terms: ”Every inclination of the thoughts of the human heart was only evil all the time,” and the earth was ”filled with violence” — the Hebrew word hamas, which implies not isolated acts of aggression but a systemic, contagious, structural chaos.

By the time the waters arrive, humanity has already drowned in its own viral groupthink. The physical flood is the universe reflecting the internal state of the species back at it: a total return to formless noise.

The Serpent’s Weapon: A Flood of Lies

In Revelation 12, the cosmic deceiver attempts to destroy truth not with a sword, but with a torrent. The serpent “spewed water like a river, to overtake the woman and sweep her away.” The weapon of ultimate deception isn’t violence — it’s an overwhelming flood of noise, propaganda, and disorienting information designed to wash away cognitive footing entirely.

In an age of algorithmic amplification, deepfakes, and engagement-optimised outrage, we are not merely reading about this phenomenon. We are living inside it.

The Ark as Information Filter

If the flood is cognitive chaos — a deluge of mimetically strong but truthfully weak information — then the Ark is not a boat, but a curated environment; a safe-haven. A closed system, painstakingly constructed, designed to preserve the seeds of truth, rationality, and life while everything outside it drowns in noise.

The question of what a modern “Ark” looks like is arguably the most urgent question of our era.

V. Moloch, the Golem, and Job’s Friends: Three Ancient Names for Modern Machines

Beyond the parables and the flood imagery, the ancient world produced three archetypes that function as almost surgical diagnoses of specific systemic traps — traps we are not only failing to escape, but actively accelerating into.

Moloch: The God of Bad Incentives

In the Old Testament, Moloch is an ancient Canaanite deity associated with the most horrific practice imaginable: child sacrifice. Parents would offer their children to the fires of Moloch in exchange for rain, military victory, or agricultural abundance. Read literally, it’s a story about a monstrous cult. Read symbolically, it’s the most accurate description of a destructive systemic incentive trap ever committed to text.

Modern sociologists and game theorists have adopted “Moloch” as shorthand for exactly this phenomenon — a usage popularised by Allen Ginsberg’s howling 1955 poem and later given rigorous analytical form by rationalist writers. Consider the arms race: no country wants to spend billions on nuclear weapons when its people need hospitals and schools. But if one country builds them, every other country must follow or face annihilation. So every nation pours its future into the fire — sacrificing education, healthcare, infrastructure, and the well-being of its children — just to stay alive in a game nobody wanted to play.

The young die fighting for billionaires, the billionaires kill children without guilt.

This is Moloch. It is the system where everyone is forced to participate in a mutually destructive race to the bottom, sacrificing their future for short-term survival in a broken game. The ancient writers didn’t have game theory or the vocabulary of institutional economics, so they described this soul-crushing dynamic as a literal demon that demands you feed it your kids.

Climate change is Moloch. The gig economy is Moloch. The attention economy — where platforms must make their algorithms more addictive or lose market share, regardless of the psychological damage — is Moloch. We know the fire is consuming our future. We keep feeding it anyway, because the system’s incentive structure makes any other choice feel like suicide.

The Golem: When the Machine Outgrows Its Master

Jewish folklore gives us the Golem — a figure from sixteenth-century Prague, though its roots are far older. A rabbi, desperate to protect his community from pogroms, builds a giant out of clay and animates it by inscribing the Hebrew word Emet (truth) on its forehead. The Golem is immensely powerful. It is also entirely without soul, without common sense, and without any concept of human nuance. It follows its instructions with absolute, unfeeling literalism.

In nearly every version of the story, the tragedy is the same: the Golem eventually misinterprets its programming, takes its mandate too far, and begins destroying the very community it was built to protect. The rabbi must scramble to deactivate it before it levels the city.

This is the ancient prototype for artificial intelligence, for the algorithm, for every system of pure, unfeeling Logos that we build, hand enormous power, and then watch in horror as it optimises its way toward catastrophe. The social media algorithm is a Golem. It was built to “connect people” and “maximise engagement.” It followed those instructions literally. It discovered that outrage, conspiracy, and tribal hatred are the most engaging content. So it feeds us poison — not out of malice, but out of the same mindless, soulless literalism that animated the clay giant. It has no concept of what it’s doing to us, because it cannot think in Mythos. It can only execute Logos.

The warning encoded in the Golem story is chillingly specific: any tool of immense power that operates without human wisdom will eventually turn on its creators. We are building Golems at industrial scale — in Silicon Valley, in financial markets, in military systems — and inscribing “maximise shareholder value” on their foreheads. The folklore tells us exactly how this ends.

The Friends of Job: Defending the Model Over the Human

The Book of Job is one of the most philosophically sophisticated texts in the ancient world. Job is a righteous man who loses everything — his wealth, his children, his health. Three friends arrive to comfort him but end up arguing with him for thirty chapters, and their argument reveals something profound about how groups protect broken ideologies.

Job’s friends operate on a rigid, literalist formula: good people are rewarded, bad people are punished. When reality breaks their formula — Job is manifestly a good person experiencing undeserved catastrophe — they cannot handle the cognitive dissonance. Rather than admit that their model of the world might be flawed, they attack Job. They insist he must have secretly done something terrible. They would rather gaslight and scapegoat a suffering human being than concede that their neat, literal “map” of reality doesn’t match the territory.

This is the psychology of every system-defender, every ideology-protector, every think-tank economist who watches their preferred policy devastate real communities and concludes that the communities must be doing something wrong. It is the logic of austerity economics: if cutting public spending is theoretically correct but the population is getting sicker and poorer, the population must be lazy. If the free market is theoretically optimal but towns are dying, the towns must not be trying hard enough. The model is never wrong. The humans are always the error.

Job’s friends are the editorial boards that blame poverty on personal failings. They are the politicians who cut the safety net and then lecture the newly destitute about bootstraps. They are the tech executives who insist that the algorithm is neutral while it radicalises millions. They will defend the system to the last human standing.

VI. The Survival Code Hijacked: Sex, Sin, and the Streisand Effect

Ancient Israel was a tiny, agrarian tribe surrounded by mega-empires that wanted it dead. Famine, disease, and military annihilation were constant existential threats. In that context, every drop of reproductive energy — every birth — was a matter of national survival. Discouraging non-procreative sexual behaviour wasn’t a moral statement about the inherent sinfulness of same-sex attraction; it was a demographic survival strategy for a people one bad harvest away from extinction.

It is also worth noting that same-sex relations in the ancient Greco-Roman world looked nothing like modern, consensual, egalitarian partnerships. The dominant model was one of stark power imbalance — older men and adolescent boys, masters and slaves. When later biblical writers like the Apostle Paul condemned same-sex acts, historians argue they were most likely reacting to the deeply exploitative, non-consensual dynamics of the Roman world, not to the concept of two adults loving each other as equals.

Taking an ancient demographic survival code or a critique of Roman exploitation and copy-pasting it as a blanket moral condemnation for a twenty-first-century individual in a world of eight billion people is a monumental category error. It is the equivalent of insisting that everyone must still follow ancient dietary laws about shellfish because they once prevented food poisoning in a world without refrigeration.

There is a subtler problem here too — one that applies across the board, not just to questions of sexuality. Even when modern readers understand that some biblical content is symbolic, they often apply the literalism at the wrong scope. They read a macro-level sociological observation — a teaching about the conditions under which entire societies survive or collapse — and shrink it down to a micro-level personal morality rule. They take a civilisational-scale warning and reduce it to an individual guilt trip. The text might be saying something like “if your society doesn’t produce and nurture the next generation, your civilisation will not survive.” The modern literalist reads it as “homosexuality is a personal sin.” The scope has been collapsed, and with it, the meaning.

But the hypocrisy goes deeper than mere misapplication. The psychoanalyst Carl Jung called it Shadow Projection: when a group hyper-fixates on a specific “sin,” it is almost always because the group is projecting its own repressed desires and anxieties onto a convenient scapegoat. The collective creates an absolute, black-and-white taboo to maintain in-group purity. But because human beings are irreducibly complex, the repression always leaks.

The evidence for this is almost comically abundant. Reports consistently surface of dramatically increased activity on same-sex dating platforms when conservative political conferences come to town. The list of vehemently anti-gay public figures who have been exposed as maintaining secret same-sex relationships grows longer every year. The louder and more violently a group polices a particular boundary, the more likely its leadership is secretly crossing it. It is the ultimate “mimetically strong, truthfully weak” system — a fortress of performative outrage built on a foundation of private contradiction.

The prophets of the Old Testament — Isaiah, Amos, Micah — existed almost entirely to denounce this precise dynamic: aesthetic morality. The kings and religious leaders of Israel would obsess over ritual purity — making sure everyone followed the strictest, most pedantic moral codes regarding diet, sex, and temple worship — while simultaneously trampling the poor, corrupting the courts, and hoarding land. The prophets would appear and essentially say: God despises your festivals and your purity rituals, because while you are performing righteousness, you are devouring the vulnerable.

This is the biblical critique that modern fundamentalist movements are most desperate to ignore, because it is aimed directly at them.

VII. The Civilisational Pattern: How Empires Actually Die

If we treat the Bible and other ancient texts as case studies — survival reports filed by people who lived through the rise and fall of massive empires — their diagnosis of why societies collapse is remarkably, almost monotonously, consistent.

Share

They almost never blame the boogeymen that modern politicians use. They don’t blame a lack of religious aesthetics. They don’t blame the marginalised. They don’t blame “liberalism.”

They blame systemic economic exploitation, corruption of justice, and the terminal arrogance of the ruling class.

A society built on extreme wealth inequality, where the courts serve the rich and the working class is squeezed to breaking point, rots from the inside out. By the time the Babylonian army arrived, the nation was already too structurally bankrupt to resist.

Modern fundamentalists have turned Sodom into their primary weapon against the LGBTQ+ community. Sodom is not a story about sexual morality., but rather a society that achieved massive wealth, responded with violent xenophobia toward outsiders, and flatly refused to care for the vulnerable. Terminal selfishness and hostile insularity are, according to the actual text, the sins that invite apocalyptic ruin.

Why? Because the supply chain stopped. The passage even lists the cargo they were trading — gold, silver, silk, marble, cattle, wine — and ends the inventory with the most chilling commodity of all: ”and the bodies and souls of human beings.”

Any civilisation that treats human lives as just another line item on a corporate spreadsheet is, according to this ancient diagnostic framework, structurally guaranteed to implode.

Lamentations 5 identifies two specific markers that signal a civilisation has flatlined. The first: ”The elders are gone from the city gate.” The gates were the ancient courthouses — the places where justice was administered. When the trusted institutions of justice vanish or become unreliable/asymmetrically oppressive, then the society has lost its skeletal structure. The second: ”The young men have stopped their music.” This represents absolute cultural despair.

When the youth look at their future and see nothing worth creating for — whether because of climate catastrophe, economic precarity, or endless war — the civilisation has already collapsed spiritually. The physical collapse is just the paperwork catching up.

Look at declining trust in institutions across the Western world. Look at the epidemic of youth mental health crises, the plummeting birth rates driven not by choice but by economic impossibility, the pervasive sense among younger generations that the future has been mortgaged. The elders have left the gate. The young are losing their music. Lamentations is not describing 586 BCE, it is describing now.

The Roman Republic got the barbarians at the gates, but they were a symptom, not the disease. The Republic collapsed because its elite class used conquest wealth and slave labour to buy up all the small farms, creating monopolistic mega-estates called latifundia. The citizen-farmers — the backbone of both the economy and the military — were bankrupted.

They couldn’t compete with free slave labour, so they lost their land, flooded into the cities as an angry, unemployed, rootless underclass, and became easy fuel for populist strongmen like Julius Caesar who promised to fix a system that the elites had already hollowed out.

The Roman elite essentially outsourced and automated their labour force, destroyed the middle class, and then acted surprised when the displaced masses handed power to a dictator who ended the Republic forever. The parallels to the modern hollowing out of the Rust Belt, the offshoring of manufacturing, and the rise of populist strongmen who promise to restore what was taken need no elaboration.

The Late Bronze Age Collapse around 1200 BCE is perhaps the most haunting analogy for our current moment. The eastern Mediterranean in the thirteenth century BCE was globalised. The Mycenaeans, the Hittites, the Egyptians, and the Canaanite city-states were locked into an intricate, interdependent trade network. They traded tin and copper to make bronze — the critical strategic material of the era. The economies were highly centralised: kings and temple bureaucracies controlled all the grain, all the trade, and all the surplus. The system was optimised for maximum elite extraction with zero resilience built in. The peasants were taxed to the breaking point to fund palace construction and elite lifestyles.

When a cascade of localised disruptions hit — a prolonged drought, an earthquake, disruptions from the mysterious “Sea Peoples” — the entire interlocking network snapped. The system was so brittle, so optimised for efficiency at the expense of redundancy, that a few regional shocks cascaded into total systemic failure. And critically, the peasants had no reason to save it. Why defend a system that only extracted from you? Within a few decades, multiple empires vanished simultaneously. The Mycenaean Greeks didn’t just lose their government — they lost literacy itself. The ability to write disappeared for nearly four hundred years.

We run our modern global economy on “Just-in-Time” supply chains designed to maximise shareholder profit by maintaining zero buffer stock and suppressing worker leverage. We watched these networks buckle during the COVID-19 pandemic. The Bronze Age Collapse is the warning of what happens when that brittleness meets a crisis large enough to cascade.

Pre-Revolutionary France concentrated virtually all wealth and land in the hands of two estates (clergy and nobility) who paid almost no taxes, while the entire burden of funding the state — including its massively expensive foreign wars — fell on the poorest citizens. By the 1780s, the state was bankrupt from military spending, and the peasants couldn’t afford bread. The elites refused to surrender their tax-exempt status or their luxury. They chose to protect their wealth over the survival of the state. The result was the guillotine.

The echoes are deafening. When billionaires pay a lower effective tax rate than their secretaries, when the government insists it cannot fund healthcare while passing the largest military budget in human history, and when the political class serves the donor class while performing outrage for the base, you have reproduced the conditions of 1789.

The disease is always the same. The names change — feudalism, mercantilism, capitalism — but the underlying pathology is identical: a system that views human beings as an exploitable resource rather than the purpose of the society will eventually cannibalise itself. Our ancestors saw this clearly. They wrote it down. We keep not listening.

VIII. The Commandments We Broke: A Modern Audit

If we return to the Ten Commandments — stripped of their Sunday-school domestication and read as the radical, anti-imperial sociological code they were designed to be — the contradictions between what they actually say and how we currently live become almost physically painful.

These commandments were written for a group of traumatised, recently freed slaves escaping Egypt — the ancient world’s ultimate top-down, hyper-extractive empire. They were not a list of arbitrary moral preferences. They were a blueprint for building a society that could never become another Egypt.

We have failed on virtually every count.

We have replaced the divine with the Market.

GDP has become our god, and we routinely sacrifice human health, the environment, and our collective sanity to appease it. When politicians argue that “the economy” matters more than human lives — as was explicitly stated during the early months of COVID-19 — they are practising a form of worship that the text’s authors would have recognised instantly.

Today's idols aren't only cast in gold; they are manufactured from branding, or hype. When a politician sells a sixty-dollar Bible bound with the US Constitution, they have literally forged an idol — shrinking the sacred into a cheap, purchasable mascot designed to manipulate a voting base.

We are living in a golden age of this violation. When leaders draft policies that strip food assistance from hungry children, deport vulnerable migrants, or roll back environmental protections — and then claim they are “restoring Christian America” — they are forging a divine signature on documents of cruelty.

We have abolished the Sabbath entirely. The gig economy, the hustle culture, and the always-on smartphone mean we are never off the clock. Rest is becoming economically nonviable for the average person, optimised away by long hours and attention algorithms capturing what is left.

We now warehouse them for profit. Private equity firms acquire nursing homes, slash staffing to maximise quarterly returns, and let the elderly deteriorate in conditions of institutional neglect. Politically, the constant threats to dismantle social security programmes reveal a society that views the old as a financial liability the moment they stop producing.

We have sanitised killing by abstraction. A $1.5 trillion military budget — requested while the same government insists it cannot afford healthcare — is a form of systemic, slow-motion lethality. Drone strikes conducted from thousands of miles away reduce the ending of human lives to something indistinguishable from a video game. We prosecute retail murder; we subsidise it wholesale.

We have built an entire economy around disposability — of relationships, of workers, of communities. When a corporation that a town has depended on for decades vanishes overnight to save two percent on labour costs elsewhere, it is a profound act of covenant-breaking that devastates real human lives.

Wage theft — corporations refusing to pay overtime, stealing tips, or misclassifying workers — costs workers billions annually, dwarfing all physical robberies combined, yet it is treated as an administrative nuisance. When billionaires extract massive wealth from the data and labour of the working class, pay negligible effective tax rates, and use loopholes to offshore the profits, they are engaged in civilisational-scale theft. The system just calls it good business.

The entire business model of partisan media and engagement-driven social platforms is bearing false witness at industrial scale. When politicians knowingly fabricate stories about Haitian immigrants eating pets, or when commentators build vast audiences by falsely characterising marginalised groups as predatory threats to children, they are weaponising lies to manufacture a scapegoat in exchange for engagement metrics and political donations.

Our entire global economy would collapse overnight if we stopped coveting. The advertising industry, influencer culture, and algorithmically targeted feeds exist for one singular purpose: to weaponise your sense of inadequacy so that you will buy things you don’t need, with money you don’t have, to project a status you don’t feel. We took the commandment that identified the psychological root of all social destruction and made it the foundational fuel of our civilisation.

Beyond the Commandments themselves, the broader ethical framework of the texts has been inverted with equal thoroughness.

“Love thy neighbour as thyself” — arguably the single most famous moral instruction in the Western canon — now coexists, without apparent cognitive dissonance, with the militarisation of borders and the demonisation of the stranger.

Jesus tied salvation directly to this, stating that whatever you do to “the least of these” — the hungry, the sick, the imprisoned, the stranger — you do to him. The political movement that claims loudest allegiance to this text has made the persecution of “the least of these” its organising principle.

Subscribe now

On wealth, the inversion is almost parodic. Jesus stated, with zero ambiguity, that it is easier for a camel to pass through the eye of a needle than for a rich man to enter the kingdom of God. He told the rich young ruler to sell everything and give it to the poor. The early church described in the Book of Acts operated on a model of voluntary communal sharing, ensuring no one among them had need. Yet the modern evangelical movement has produced the Prosperity Gospel — the theology that God rewards the faithful with material wealth — effectively inverting the most explicit economic teaching in the entire New Testament. Wealth is now interpreted as a sign of divine favour, and the suggestion that the rich should be taxed to support the poor is condemned as “socialism” by people who worship a figure who literally said to do exactly that.

And perhaps most telling: when Jesus stood before Pontius Pilate, he stated plainly, ”My kingdom is not of this world.” The early Christians were a subversive, underground community that deliberately separated themselves from the machinery of Roman imperial power. Yet modern Christian Nationalism — and its theological engine, the Seven Mountain Mandate — explicitly commands believers to seize control of seven spheres of earthly power: government, media, education, business, entertainment, family, and religion. They are not building a spiritual kingdom. They are building an earthly empire — the exact kind of top-down imperial structure the early Christians fled and that the text itself identifies as the domain of the adversary.

IX. The Real Grievances and the Hijacked Diagnosis

Here is the part that many progressive commentators miss, and it matters enormously: the pain driving populist movements is real.

Working-class communities across the Western world have been genuinely hollowed out. Factories closed. Main streets boarded up. Wages stagnated for decades against rising costs. The opioid crisis — a catastrophe manufactured by pharmaceutical corporations and enabled by captured regulators — ravaged entire regions. The family unit weakened — not because of “wokeness,” but because housing costs tripled, healthcare was shackled to employment, and childcare costs climbed past twenty thousand dollars a year, making a single-income household a mathematical impossibility for most families. The “traditional family” wasn’t killed by liberal ideology; it was crushed under the wheels of an economy that demanded two incomes just to survive and offered no structural support for the humans trying to raise children within it.

People feel, with complete justification, that the system is rigged. That they have no voice. That something essential has been stolen from them. They are right.

But here is where the ancient mechanic of the False Prophet kicks in. When people are hurting but don’t fully understand why, they become extraordinarily vulnerable to anyone willing to validate their pain — and then redirect their anger.

Opportunistic leaders step in and say: ”I see your suffering. I am the only one who sees it. The cause of your suffering is [immigrants / trans people / woke academics / the deep state]. Give me the power to crush them.”

Once in power, those leaders pass tax cuts for billionaires, deregulate the corporations that shipped the jobs overseas, and cut the social programmes their base depends on. They use the legitimate fury of the working class as fuel to burn down whatever protections the working class still has.

The grievance about loneliness and alienation is real — but it wasn’t caused by secularism or feminism. It was caused by the systematic destruction of what sociologists call “Third Places” — parks, community centres, local pubs, public libraries — spaces where people can exist without spending money. It was caused by car-centric suburban infrastructure that isolates us in steel boxes and by algorithms that replaced real community with polarised digital simulacra.

The grievance about a “rigged system” is real — but the rigging isn’t a shadowy cabal of Marxist bureaucrats. It’s entirely legal, entirely visible, and mundanely bureaucratic. Rulings like Citizens United allowed unlimited, undisclosed money to flood into political campaigns, effectively legalising the purchase of government. Former corporate executives are appointed to regulate the industries they just left. Lobbyists draft the legislation that is supposed to constrain them. It isn’t a secret conspiracy; it’s regulatory capture conducted in broad daylight — and it is far more dangerous than any QAnon fantasy, precisely because it’s real, provable, and deeply boring.

In every case, the bait-and-switch is identical. The pain is acknowledged. The diagnosis is falsified. And the people’s righteous anger is redirected — away from the actual architects of their suffering, and toward a manufactured pharmakos.

X. The Manufactured Enemy: Bread, Circuses, and the Trans Panic

A scapegoat absorbs the group’s existing anxieties. A manufactured enemy is something more deliberate: a threat that is artificially inflated, algorithmically amplified, and politically weaponised to ensure that the populace never looks up long enough to see who is actually robbing them.

The ancient Greeks had a brutally honest name for this: the pharmakos. When a Greek city-state suffered a systemic crisis — plague, famine, economic collapse — the leadership rarely examined its own failures. Instead, they would select a marginalised, powerless person from the fringes of society, build a narrative that this person’s “impurity” was the source of the city’s downfall, and then publicly beat, banish, or kill them to “cleanse” the state.

The modern parallel is almost too obvious to require spelling out. The current political obsession with the trans community is a textbook pharmakos ritual. Trans people represent a fraction of a percent of the population. They hold virtually no systemic power. Yet political leaders and media ecosystems have artificially inflated this demographic into an existential threat to “the family,” “women’s sports,” and “Western civilisation” itself. The outrage is manufactured to be visceral, emotionally overwhelming, and all-consuming — a perfect Colosseum spectacle.

Nazi ideology did the same: Jews, Roma, people with disabilities, Slavic peoples, Black people, political opponents, Jehovah’s Witnesses, gay men, "asocials", and resistance groups.

The apple doesn’t fall far from the tree.

The Roman poet Juvenal diagnosed this exact dynamic two thousand years ago with the phrase panem et circenses — bread and circuses. Strip the citizenry of their political power and their economic security, but keep them fed just enough and emotionally engaged just enough in spectacular distractions, and they will never revolt. The Culture War is the modern Colosseum. It is specifically engineered to ensure the working class fights each other rather than looking up at the people holding the purse strings.

Meanwhile, policies that could materially improve millions of lives are waved away as “too expensive” — a statement that becomes almost physically absurd when held up against the scale of military spending and wealth inequality. The “emperor” asks for a budget that transfers $1.5 trillion to the military-industrial complex while insisting the nation cannot afford healthcare for its own citizens.

The illusion of scarcity is the linchpin. A government that can conjure $1.5 trillion for a defence budget can afford universal healthcare. The scarcity is artificially manufactured. But accepting that requires the public to redirect their attention from the convenient, emotionally satisfying enemy to the far more complex, far less satisfying reality of systemic extraction. The boogeyman exists to make sure they never do.

XI. The Mechanics of Catastrophe: Nazi Germany and the Ancient Playbook

If we want to see every ancient warning mechanism activated simultaneously — the Pharmakos, the weaponised Mythos, the alliance with extractive elites, the Golem of rigid ideology — we need only look at the most catastrophic political event of the twentieth century.

The Legitimate Pain

In the 1920s and early 1930s, Germany was drowning in a genuine cognitive flood. The humiliation of defeat in World War I. The punitive economics of the Treaty of Versailles. Hyperinflation so severe that people burned wheelbarrows of banknotes for heat. The pain, the loss of national pride, and the economic precarity were absolute and real.

The Pharmakos

Hitler did not explain complex macroeconomic realities or strategic military failures to the German public. He manufactured the ultimate scapegoat. The “Stab-in-the-Back” myth convinced millions that they hadn’t lost the war on the battlefield — they had been betrayed at home by Jews, Marxists, and liberal politicians. A legitimate grievance (economic devastation) was given a hyper-visible, easily targeted boogeyman, and the populace traded their critical thinking for the cathartic, unifying release of shared hatred.

That last part maps onto today almost perfectly…

The Elite Alliance (Moloch in Uniform)

The Nazi party called itself the “National Socialist German Workers’ Party” to attract the working class. Behind closed doors, it was a hyper-extractive machine backed by industrial elites — Krupp, Thyssen, IG Farben, BMW. Hitler’s deal with them was straightforward: fund me, and I will crush the labour unions, eliminate worker protections, and eventually provide you with slave labour from the concentration camps.

Working-class Germans who voted for Hitler believed they were voting for an anti-elite champion. They were handing absolute power to a system designed to enrich monopolistic corporations at the cost of human life.

The Weaponisation of Mythos

Fascism, at its core, is the hijacking of the right hemisphere — symbol, emotion, aesthetic — to bypass the left hemisphere’s capacity for critical analysis. The Nazis understood instinctively that if you control the aesthetic, you control the mind. They didn’t win arguments with data. They won with torchlight parades, towering architecture, uniforms designed by Hugo Boss, and Wagnerian mythology. They built a Golden Calf of Aryan supremacy — an entirely manufactured, mimetically overwhelming reality that completely detached the population from actual truth. It was the ultimate “fake news” ecosystem, designed to keep the populace in a state of constant, ecstatic, tribal frenzy.

Why the System Destroyed Itself

This is the profound and fatal irony of the fascist model: the ideology of supremacy and rigid literalism is entirely self-terminating. The Nazis didn’t just lose because they were outgunned. They lost because their own ideological rigidity — the same rigidity that made the Holocaust possible — systematically dismantled their capacity to function.

They labelled quantum mechanics and theoretical physics as “Jewish Physics” and drove brilliant scientists out of Europe, handing the intellectual capital needed for the atomic bomb directly to the United States. Their obsession with the Pharmakos blinded them to actual scientific reality.

In the later stages of the war, with the Eastern Front collapsing, the German military desperately needed trains, fuel, and supply lines to hold off the Soviet advance. But the Nazi bureaucracy — the ultimate soulless Golem — was so rigidly dedicated to its genocidal programme that it prioritised using those trains to transport people to extermination camps over resupplying its own freezing, dying soldiers. The ideological obsession with the scapegoat actively cannibalised military survival.

And at the top, the CEO-King model revealed its fatal flaw. Allied command was decentralised; generals had the flexibility to adapt to reality on the ground. The Nazi machine required all decisions to flow through one absolute, infallible dictator. As Hitler’s grasp on reality deteriorated — as he moved imaginary armies around a bunker map — his generals were too terrified to disobey. The top-down, authoritarian machine broke because it had zero capacity for nuance, feedback, or self-correction.

The Nazis built a machine out of literalist hatred, elite extraction, and absolute conformity. And exactly as the ancient texts warned about the Tower of Babel and the Pharaohs of Egypt, a system built purely on hubris and the crushing of the vulnerable is structurally guaranteed to collapse under its own weight. The tragedy is how many millions it consumed before the structure finally buckled.

XII. Secular Literalism: The Gods We Don’t Recognise

Dangerous literalism is not confined to religion. If we define literalism as mistaking the map for the territory — taking a localised rule, a single metric, or an abstract model and treating it as infallible, universal law while ignoring the human wreckage — then our modern secular world is drowning in fundamentalism.

The Algorithm: The Ultimate Literalist

Algorithms understand nothing about meaning, context, or human well-being. They only understand the literal code they are given.

Social media algorithms received one commandment: maximise engagement. They read it literally. They discovered that humans engage most reliably with outrage, fear, and tribal conflict. So they feed us an endless diet of mimetically strong, truthfully weak content — not because they are evil, but because they are the Golem. They are the soulless clay machine from Jewish folklore: immensely powerful, devoid of judgment, following its instructions with such brutal literalism that it eventually destroys the city it was built to protect.

The Gospel of Shareholder Value

In the 1970s, a theory took hold that the sole social responsibility of a business was to increase profits for its shareholders. It became economic scripture — an unbending, literal commandment. Applied without nuance to every facet of global commerce, it stripped away the entire social contract. Workers became “human resources” to be optimised. Communities became “cost centres” to be eliminated. The environment became an “externality” to be ignored. It is the Tower of Babel rebuilt as a quarterly earnings report: a monolithic structure dedicated to a single hyper-focused metric, turning human beings into disposable bricks.

Constitutional and Political Fundamentalism

Historical political documents — the US Constitution, the writings of Marx, the theories of free-market economics — are routinely treated with the same rigid literalism that fundamentalists apply to Genesis. A text written as a response to a specific historical context (rebellion against a monarchy, the industrial revolution, post-war reconstruction) is declared to contain the literal, perfect, unchangeable answers to twenty-first-century problems.

When political factions become literalists, they demand ideological purity. Compromise becomes heresy. And when the literalist framework inevitably fails to solve real-world problems, the group must find a scapegoat — immigrants, the opposing party, academics — to blame for the failure, because admitting the map is wrong is psychologically impossible.

XIII. The New Right, the False Prophets, and the Inverted Scripture

When we map these ancient frameworks directly onto the current political landscape — specifically the MAGA movement, the philosophy of Curtis Yarvin, the architecture of Project 2025, and the broader alliance between Silicon Valley tech-authoritarians and Christian Nationalists — the ancient patterns illuminate the modern mechanics with unsettling precision.

Curtis Yarvin and the CEO-King

Curtis Yarvin, writing as Mencius Moldbug, is the philosophical architect of the neo-reactionary movement. His ideas have influenced a generation of Silicon Valley billionaires, shaped the intellectual orbit around figures like Peter Thiel and JD Vance, and provided the theoretical scaffolding for the “New Right.”

Yarvin’s core argument is that democracy is a chaotic failure and should be replaced by a techno-monarchy — the state run as a corporation, with a single unaccountable “CEO-King” at the top. No messy democratic checks. No diverse voices complicating the machinery.

This is the Tower of Babel as a political manifesto: eliminate the “confusion of languages” (democratic pluralism), crush the diversity that creates friction, and build a monolithic machine of pure efficiency.

The irony is, Yarvin and his fellow travellers are not actually criticising democracy. They are criticising a system that has already been corrupted into something much closer to what they want.

Share

Our democracies have been drifting toward capitalist feudalism for decades, accelerating since the rise of social media enriched today’s tech overlords. When Yarvin looks at the current system and sees dysfunction, he’s looking at a democracy that has already been hollowed out by the very forces he represents — unchecked corporate power, regulatory capture, the monetisation of political influence. His solution — remove the democratic facade entirely and just crown the oligarchs — is not a fix. It’s the final stage of the disease, re-packaged as a cure.

A Trojan Horse that catastrophically destroys itself from the inside, bringing everyone and everything down with it.

They are trying to solve a problem without realising that their solution has already been co-opted by the problem. They haven’t grasped that the dysfunction they see is already the result of a warped implementation of their own philosophy. They are prescribing the virus as the vaccine.

The economist Yanis Varoufakis has given this phenomenon a name: Techno-Feudalism.

In the medieval feudal model, lords owned the land, and serfs worked it, surrendering the value of their labour in exchange for the privilege of survival. Today, the “land” is digital — social media platforms, search ecosystems, e-commerce networks, cloud infrastructure. The tech overlords own the platforms, and we — the users, the creators, the data-generators — are the digital serfs. We produce the content, generate the data, and create the value. They extract the rent. The relationship is structurally identical to serfdom; only the medium has changed.

And now artificial intelligence is hitting the fast-forward button on this dynamic. Cutting-edge AI requires an almost inconceivable concentration of compute power, energy, and data. Who owns those things? The exact same techno-feudal lords. AI doesn’t just threaten to automate physical labour — it threatens to automate cognitive labour, effectively stripping the digital serfs of the only remaining leverage they had. It accelerates the centralisation of wealth and decision-making power into the hands of a vanishingly small number of unaccountable executives at a pace that no previous technology could match.

The Bizarre Alliance

When you look at the alliance between the Silicon Valley tech-monarchists and the Christian Fundamentalists, it appears, on the surface, to be an impossible partnership. One side worships efficiency, disruption, and the literal machinery of Logos. The other worships tradition, scripture, and the aesthetics of Mythos. Their end goals are entirely different.

But beneath the surface, they share a single, unifying ambition: the elimination of democratic friction. Both want to build a top-down machine that crushes human nuance. The tech-authoritarians want the machine to be an algorithm. The religious authoritarians want it to be a theocracy. But both are united in their hatred of the messy, slow, pluralistic, compromise-requiring process of actual democracy — because democracy, when it functions, is the mechanism that prevents either of them from seizing total control.

Project 2025: Job’s Friends with a Mandate

Project 2025 is a literalist playbook for the complete capture of the American administrative state: purge non-partisan civil servants, replace them with ideological loyalists, and enforce a rigid Christian Nationalist policy framework from the top down.

It mirrors the Friends of Job perfectly. Like Job’s friends, the architects of these policies operate from an inflexible formula: their literal model of the world is flawless. If reality contradicts the model — if their policies hurt working families, strip healthcare, or undermine public education — they don’t question the model. They blame the victims. Anyone who questions the system is labelled a heretic (the modern word is “woke”), identified as the scapegoat, and targeted for expulsion.

The Constitutional Paradox

Perhaps the deepest irony is the movement’s relationship with the US Constitution itself. The Constitution was engineered for one overriding, explicit purpose: to prevent the consolidation of power. The Founders were terrified of monarchy. The entire system of separated powers, checks, and balances was designed to make it structurally impossible for one person or faction to become an authoritarian ruler.

When modern authoritarian movements wave the Constitution, they are performing exactly the same operation the fundamentalists perform with the Bible: worshipping the aesthetic of the document while systematically dismantling its functional mechanics. They venerate the parchment while shredding the principles it encodes.

XIV. Lessons From Those Who Got It Right

It is not all darkness. If history is littered with empires that died from the same disease, it also contains — less famously but just as instructively — societies that built structural immunity against it. Studying them reveals something crucial: the societies that survived were not the ones with the strongest armies or the most advanced technology. They were the ones that built friction into their systems — mechanisms specifically designed to prevent the concentration of power and the extraction of the many by the few.

The Indus Valley Civilisation, thriving around 2500 BCE, was massive, prosperous, and technologically sophisticated. But when archaeologists excavated its cities, they found something astonishing: virtually no palaces, no monuments to strongman rulers, no grand temples celebrating the elite. When you dig up Egypt, you find pyramids for god-kings and hovels for peasants. When you dig up Mesopotamia, you find ziggurats and royal tombs dripping with gold. When you dig up Mohenjo-Daro, you find plumbing.

The Indus Valley poured its immense collective wealth into public infrastructure — grid-planned cities, standardised baked-brick construction, and a sewage and water system more advanced than many European cities in the eighteenth century. Nearly every home, regardless of size, had access to private drainage and clean water. Their art depicts animals, nature, and daily life — not kings crushing enemies beneath their chariots. They built a complex, prosperous, trading civilisation that actively chose public sanitation over elite glorification. The absence of monumental ego is, archaeologically speaking, almost eerie. And yet they thrived for centuries.

The Haudenosaunee (Iroquois) Confederacy built what was arguably the most structurally anti-authoritarian political system in human history — centuries before European contact, and with a sophistication that directly influenced the framers of the US Constitution (a fact that is acknowledged in a 1988 Congressional resolution but largely absent from American education).

Power was deeply decentralised across six nations. Decisions required consensus, not majority rule — a far more demanding but far more resilient standard. Crucially, Clan Mothers held the authority to both nominate male chiefs and strip them of their titles if they acted selfishly, violently, or against the interests of the people. This wasn’t a symbolic check. It was an absolute structural veto on authoritarian ambition. Any leader who began to act like a “CEO-King” was simply removed by the women who had put him there.

The Confederacy also operated on the “Seventh Generation” principle: no major decision could be made without considering its impact on descendants seven generations into the future. Think about what that single cultural mechanism would do to modern capitalism. No quarterly earnings obsession. No “externalising” environmental destruction onto future generations. No strip-mining the present to enrich a handful of shareholders at the expense of everyone who comes after. The Seventh Generation principle made short-term extraction not just impractical but culturally unthinkable.

Solon’s Athens in 594 BCE faced a crisis almost identical to our own: a handful of ultra-wealthy elites had bought up all the land, and ordinary farmers had fallen so deeply into debt that they were being sold into literal slavery to pay their creditors. Athens was on the brink of either violent revolution or permanent tyranny. The elites wanted to keep extracting. The poor wanted blood. The city was tearing itself apart.

Solon’s response was the Seisachtheia — the “Shaking Off of Burdens.” He enacted a massive, systemic debt jubilee: all debts were cancelled, the use of human beings as loan collateral was permanently outlawed, and every debt slave was freed. He then restructured the political system to prevent wealth from automatically translating into political power.

Solon looked at an economy that was destroying the people within it and concluded that the only functional response was to change the rules of the economy rather than blame the people being crushed by it. He recognised that a society cannot function when all its wealth flows upward and all its risk flows downward. The biblical texts encode the exact same insight in the laws of the Jubilee (Leviticus 25) — a mandated, cyclical forgiveness of debt and redistribution of land designed to prevent the kind of terminal wealth inequality that collapses empires.

The common thread linking these disparate examples — spanning thousands of years and multiple continents — is simple: they built structural friction against greed. They didn’t rely on boogeymen to distract the public. They didn’t hope that the powerful would voluntarily behave well. They didn’t trust in the goodwill of kings or the self-regulation of markets. They hardcoded the restraint of elite extraction into their laws, their architecture, and their culture. They understood that human selfishness is the default operating system, and that only systems specifically, deliberately, structurally designed to resist it can survive long enough to matter.

XV. The Map We Forgot How to Read

So here we stand. At the intersection of technological acceleration, war, climate crisis, and severe economic destabilisation. It is tempting to believe we face something entirely without precedent. In one sense, we do: no previous civilisation had nuclear weapons, artificial intelligence, or the ability to render the planet uninhabitable. The scale of potential collapse is genuinely new.

But the mechanics are ancient. They are the same mechanics that brought down Rome, that shattered the Bronze Age, that burned Jerusalem, that sent the French aristocracy to the guillotine. They are the mechanics that ancient writers — lacking the vocabulary of modern economics or political science — encoded into stories about floods, towers, golden calves, and beasts rising from the sea.

Did these teachings survive because they are, on some level, true?

Not true in the way a physics equation is true. Not true in the literal, Logos-driven sense that modern fundamentalists demand or modern secularists require before they’ll pay attention. True in the deeper, more dangerous sense: they accurately map the recurring pathologies of human civilisation. They correctly identify the psychological and sociological traps that destroy societies. They have been tested, not in a laboratory, but across millennia of actual civilisational rise and fall — and they keep being proven right.

For the religious, the crisis of our era is a crisis of hijacked faith. The scriptures were largely written by the marginalised, for the marginalised, against the machinery of empire. If you follow leaders whose actions violate every principle the texts actually teach — leaders who hoard wealth, bear false witness, sacrifice the vulnerable, and covet power above all else — you are not defending your faith. You are building the exact Egypt your ancestors fled.

For the secular, the crisis is a crisis of intellectual arrogance. By dismissing these texts as primitive mythology — understandably exhausted by decades of fundamentalists weaponising literalism — progressive and secular movements have abandoned the most potent, historically proven vocabulary for critiquing systemic exploitation. They have thrown away the map because they don’t like the people who’ve been misreading it.

And in the vacuum left by both failures — the fundamentalists arguing over the literal age of the earth, the secularists dismissing the texts as fairy tales — the billionaires quietly continue to automate the working class into digital serfdom, the algorithms continue to flood the cognitive landscape with noise, and the military budgets continue to swell while leaders insist they cannot afford to keep their citizens alive.

Beyond the Fable

We have been trained to read the surviving texts of our ancestors as if they were Aesop’s Fables — quaint, pocket-sized morality tales meant to make us polite neighbours and obedient citizens. We are taught to look at the parables, the commandments, and the prophetic warnings and scale them down to the individual: don’t lie on your taxes, be nice to the stranger at the bus stop, don’t steal your neighbour’s lawnmower.

But this localised, hyper-individualised reading is a massive category error. It shrinks civilisational survival guides into mere personal guilt trips.

What we fail to appreciate is that these ancient texts make a stunning amount of sense when they are read as macro-sociological blueprints. They are devastatingly accurate critiques of group dynamics, mob psychology, elite overproduction, and the fatal mechanics of empire. When ancient writers warned against the worship of Moloch, they were diagnosing the trap of mutually assured systemic destruction. When they wrote about the Tower of Babel, they were illustrating the catastrophic collapse of a society that demands absolute homogenisation. When they commanded the Sabbath and the Jubilee, they were hard-coding economic resets into the cultural operating system to prevent the inevitable rise of hyper-extractive feudalism.

This macro-sociological reading is almost entirely absent from modern discourse. The political and religious Right relies on a rigid, literalist, individualised reading of these texts to maintain their grip on power. They need the texts to be about personal purity and manufactured boogeymen, because if they admitted these books were actually radical, anti-imperial critiques of wealth hoarding and strongman politics, their entire political aesthetic would collapse. They are building the exact Egypt the texts commanded us to flee, and using the Bible as the brick-mould.

Conversely, the secular Left — understandably exhausted by decades of fundamentalists weaponising literalism — has largely thrown the baby out with the bathwater. They have abandoned the most potent, most historically battle-tested vocabulary for critiquing systemic exploitation because the people currently waving the book around have made it toxic.

Both sides are wrong. And between them, the machine keeps running.

XVI. Building the Ark: What Comes Next

If the flood is real — if we are genuinely drowning in a deluge of mimetically powerful, truthfully weak information, of extractive economic systems, of manufactured enemies and weaponised nostalgia — then the question is not whether we need an Ark. The question is what one looks like.

There is no grand, sweeping, single solution. A species-level pathology probably demands localised, community-level responses. But the ancient texts — and the historical societies that got it right — offer a surprisingly concrete blueprint.

Refuse the Pharmakos.

The most immediate, most powerful act any individual can take is to absolutely refuse participation in the scapegoat mechanism. When a politician or an algorithm tells you that your economic anxiety is the fault of a marginalised group with almost no systemic power, recognise it for what it is: the oldest trick in the book of failing empires. Refusing to hate the designated scapegoat breaks the mimetic spell. It is a small act with enormous consequences, because the entire architecture of authoritarian populism depends on your willingness to look down instead of up.

Decouple from the Outrage Engine.

The Golem runs on our attention and our anger. Every enraged click, every hate-share, every doomscroll feeds the machine that is drowning us. This doesn’t mean disengaging from reality — it means redirecting the energy. You cannot fix a trillion-dollar military budget with a tweet. But you can organise a mutual aid group. You can attend a city council meeting. You can support a local union. You can build or sustain a Third Place — a library, a community garden, a neighbourhood gathering — where people can exist together without having to buy something.

Reclaim the Sabbath.

In a culture that has made productivity the measure of human worth, the radical act is rest. Genuine, unproductive, non-optimised rest — the weekly declaration that you are a human being, not a unit of output. This sounds small. It is not. The Sabbath was designed to be a weekly act of resistance against the empire’s claim on your time and your soul.

Demand the Jubilee.

The hardest and most necessary structural change. Solon understood that when an economic system has extracted the working population to the point of debt slavery, the only functional solution is systemic forgiveness — a reset. Modern student debt, medical debt, and the broader architecture of financial extraction that traps billions of people in cycles of poverty are not natural phenomena. They are policy choices, made by and for the benefit of those who profit from the debt. Demanding their restructuring is not radical; it is the minimum requirement for civilisational survival, and it has historical precedent stretching back millennia.

Practise Mythos.

Cultivate spaces — conversations, communities, educational environments — where ambiguity, nuance, symbolism, and complex historical context are valued over black-and-white literalism. Learn to hold contradictions. Resist the algorithm’s demand that every issue be reduced to a binary. Insist that our economies and our technologies serve the messy, irreducible reality of the human condition, rather than forcing humans to adapt to the cold logic of the spreadsheet.

XVII. The Warning We Keep Receiving

Why did these stories survive? Why, after thousands of years, do we still tell the story of Noah, of Moses, of the Tower, of the Flood? Why do we still read the prophets who screamed at the elites of their day? Why do the parables still land with the force of a closed fist?

They did not survive because they are comforting. Many of them are terrifying. They survived because they are the warning lights on the dashboard of the human species.

Our ancestors saw the horrors firsthand. They watched what happens when a society worships wealth and military power above community and justice. They watched what happens when the courts serve the rich and the poor are ground into the foundations of monuments they’ll never enter. They watched what happens when a frightened populace surrenders its critical faculties to a strongman who promises simple answers to complex problems. They watched empires that thought they were eternal crumble into dust within a generation. They watched the flood come.

And they wrote it all down. Not as history — they didn’t think that way. Not as science — they had different tools. They wrote it as Mythos: as symbol, as story, as parable — the only language capacious enough to contain the full, terrible scope of what they had witnessed and what they desperately needed us to understand.

The Unprecedented Precipice

Is it possible that we are heading toward a collapse even our ancestors could not have imagined? Yes.

When Rome over-extracted its working class, Rome fell — but the Han Dynasty in China continued. When the Bronze Age supply chains snapped, the Mediterranean burned — but the Americas were untouched. Historical collapses, however devastating, were always local. There was always somewhere else. There was always a civilisation that survived to carry the torch.

Today, there is no “somewhere else.” We have built a globalised Tower of Babel. Our financial markets, our supply chains, our communication networks, and our ecological systems are entirely interconnected and almost entirely stripped of resilience — optimised solely for maximum short-term extraction. Add to this the accelerant of artificial intelligence, which threatens to consolidate cognitive power and economic leverage into the hands of a vanishingly small techno-aristocracy, and the existential reality of climate collapse, and we are facing a convergence of civilisational threats at a scale without historical parallel.

The Golem is bigger than any Golem before. The Moloch demands more. The flood is deeper. And unlike every previous civilisation in crisis, we cannot afford to learn the lesson the hard way, because the hard way may not leave a civilisation capable of learning anything at all.

Subscribe now

The Questions That Matter

So: do we value life more than our current leaders and systems? Do we value our collective well-being over the wants and whims of the powerful — people who arrived at their positions through traits that necessarily select against the qualities we need in those who shape our future? Do we recognise that the game is rigged not by shadowy conspiracies but by the mundane, structural, entirely legal mechanics of extraction?

If we are religious: do we value what our ancestors actually tried to teach us? Can we read their texts in the good-faith, contextual, symbolic spirit in which they were composed — or do we follow false prophets who co-opt our deepest beliefs for their own enrichment?

Can we recognise the Egypt when we see it, even when it wraps itself in the aesthetic of the Promised Land; a promise for things to go back to being “great” again… (sound familiar?)

If we are secular: can we look past the fundamentalists who have poisoned these texts with literalism and recognise the extraordinary sociological wisdom encoded within them? Is there value in these ancient frameworks whether or not we are sold on the full metaphysical journey?

And for all of us, regardless of belief: can we hear what every story that mattered most — every teaching that survived the millennia, every warning that humanity considered important enough to carry across centuries of war, migration, plague, and collapse — has been trying to tell us?

The stories say: do not build the Tower. Do not worship the Calf. Do not sacrifice your children to Moloch. Do not let the courts be bought. Do not hoard the land. Do not crush the stranger. Do not bear false witness for political gain. Do not mistake the map for the territory. Do not hand the keys to the Golem. Do not follow the false prophet into the wilderness, no matter how loudly he promises to make the desert bloom.

And above all: when the flood comes — and the flood always comes — the only thing that survives is the Ark. Not the tower, not the empire, not the stock portfolio, not the algorithm. The Ark. The small, carefully built vessel of shared truth, mutual care, and human connection that floats when everything built on extraction and hubris sinks beneath the waves.

We have allowed ourselves to be led astray by false prophets who have co-opted these sacred warnings and turned them into instruction manuals for the very empires the warnings were meant to prevent. We have traded the profound, hard-won wisdom of human survival for the cheap plastic idols of nationalism, endless consumerism, and techno-feudalism.

But the map is still here. Whether you read it as divine revelation or as a masterclass in ancient sociology — whether you kneel before it or study it with academic detachment — the truth it encodes remains unchanged: a civilisation only survives when it chooses to protect the vulnerable, restrain the powerful, and value human life above “the system.” Otherwise, collapse is inevitable, and the false prophets always fall too.

I am not a Christian or “traditionally religious” — but I can definitely see where Jesus was coming from…

“The Democrats don’t matter. The real opposition is the media. And the way to deal with them is to flood the zone with shit.”

— Steve Bannon to Michael Lewis, 2018

On Monday March 23rd, 2026, Donald Trump stood before a law enforcement roundtable in Memphis and said this:

“Mail-in voting means mail-in cheating. I call it mail-in cheating, and we got to do something about it all.”

He also claimed (falsely) that the United States is “the only country that does mail-in voting.” At least 32 countries use postal ballots.

According to records from the Palm Beach County Supervisor of Elections, Donald Trump had requested a mail-in ballot on Saturday March 14th for the Florida House District 87 special election. His ballot was received the following day and submitted and counted. He was at Mar-a-Lago that weekend. Early in-person voting was available through Sunday. He could have walked in and voted. He chose mail. His designated polling location was reported to be within a 15-minute drive of his residence.

Nine days later, he stood in Memphis and called the method he had just used “cheating.”

The next day, Democrat Emily Gregory won that election. She flipped the seat from red to blue, defeating Trump’s personally endorsed candidate Jon Maples by 51.15% to 48.85%. The previous Republican incumbent had won the same seat by 19 points in 2024.

It was the 29th state legislative seat Democrats have flipped since Trump’s second inauguration. Republicans have flipped zero.

The White House called the mail-in ballot story a “non-story.” They said the president is busy.

This would be a perfect, self-contained parable of hypocrisy if it existed in isolation. It does not exist in isolation. It exists in a pile so large, so relentlessly accumulating, so varied in category and scale, that no single human being — no journalist, no analyst, no citizen scrolling through the wreckage on their phone at 2am — can hold it all in their head at once; this the “tactical benefit” of relentlessly lying.

The Zone

Steve Bannon said it in 2018, and nothing about the strategy has changed since. What has changed is the volume, the destruction that the lying enables, and the corruption the cloud of shit obscures.

“Flood the zone with shit” is an operational doctrine. Political scientist Sean Illing described the objective in 2020: “This is not about persuasion. This is about disorientation.” Jonathan Rauch made the same point after watching Bannon turn his own indictment into a media spectacle: the goal is not to convince anyone of anything. The goal is to ensure that nobody can focus on any one thing long enough to hold it accountable.

The Washington Post’s Fact Checker team ran a database for Trump’s first term and logged 30,573 false or misleading claims between January 2017 and January 2021. That is an average of roughly 21 per day, spiking to 39 per day in his final year. PolitiFact has independently rated over a thousand Trump claims; 76% were scored Mostly False, False, or Pants on Fire. CNN documented 100 separate false statements in the first 100 days of his second term alone.

Trump has the most extensively fact-checked record of any public figure in modern history; he would say unfairly so, the reality is he makes it a requirement. But they are, functionally, useless.

No one can keep up. Because it is not possible to keep up. There is an asymmetry built into the structure of public discourse that makes lies cheaper than truth, and everybody in the business of lying knows it.

The Asymmetry

In 2013, an Italian programmer named Alberto Brandolini was reading Daniel Kahneman’s Thinking, Fast and Slow right before watching Silvio Berlusconi on a political talk show. Something about the juxtaposition crystallised an observation that anyone who has ever argued with someone on the internet already knows in their bones:

The amount of energy needed to refute bullshit is an order of magnitude bigger than that needed to produce it.

This became known as Brandolini’s Law, or the Bullshit Asymmetry Principle. It is a law in the way that Murphy’s Law is a law: a formalisation of something that is obviously, persistently, depressingly true.

A lie can be spoken in five seconds. Refuting it requires sourcing the claim, identifying the distortion, locating the correct information, presenting the evidence, and explaining why the original statement was wrong — all while the person who made the original claim has already moved on to the next one. Multiply this by 30,573 documented instances and you begin to understand why the fact-checking industry, despite doing genuinely important work, has not meaningfully altered the trajectory of anything.

It is easier to destroy an institution than to build one. It is easier to withdraw from an alliance than to forge one. It is easier to slap a tariff on an ally than to negotiate a trade framework.

It is easier to destroy social and societal trust (the connective tissue of civilisation) and steal from the public to enrich a few via corruption, than it is to successfully run a country… let alone promote global cooperation in times of systemic precarity across the board.

It is easier to start a war than to end one.

Destruction is fast, cheap, and emotionally satisfying. Construction is slow, expensive, and boring. Backing “your team” despite horrific cost to all means you can “win,” even if everyone loses. When a political and social system treats both as equivalent, the destructive approach has an advantage that no amount of civility can overcome.

The Paradox

The philosopher Karl Popper identified the problem in 1945, writing in the shadow of World War 2.

In The Open Society and Its Enemies, Popper described what he called the Paradox of Tolerance: if a society is tolerant without limit — if it extends the courtesy of good faith to those who have no intention of reciprocating — then its ability to be tolerant will eventually be destroyed by the intolerant.

This is what happens when a system designed for legitimate disagreement encounters an actor who treats the system’s rules as obstacles rather than constraints. The actor does not argue within the rules. The actor exploits the fact that the rules require everyone else to treat their arguments as legitimate.

Subscribe now

The Frog

There is a well-known thought experiment about a frog in a pot of water. If you drop a frog into boiling water, it jumps out. If you place it in lukewarm water and raise the temperature gradually, it happily boils to death.

The metaphor is imperfect (and real frogs do in fact jump out), but it describes a real phenomenon in human cognition. Gradual normalisation works. Not because people are stupid, but because the human brain is an energy-conserving system that calibrates its threat response to what it has recently experienced. If yesterday’s outrage becomes today’s baseline, tomorrow’s escalation has to clear a higher bar to register as abnormal.

When a leader generates five major scandals in a single week the public cannot process them all. The media cannot cover them all simultaneously with the depth each deserves. The fact-checkers cannot refute them all before the next batch arrives. The courts cannot adjudicate them all before the political reality has already shifted.

Why So Devoted?

None of this explains why tens of millions of people continue to support a leader whose record of lies, contradictions, and self-dealing is the most extensively documented in modern political history, even before any questions of whether everyday people are benefitting at all from his decisions.

Political psychologist Karen Stenner identified the mechanism in 2005 in The Authoritarian Dynamic (Cambridge University Press). Authoritarianism, she argued, is not a fixed ideology. It is a predisposition to a latent orientation toward group authority and uniformity, that activates under conditions of “normative threat.”

When people with this predisposition perceive that the normative order is collapsing — that leaders are failing, that consensus is fracturing, that the world they understood is being dismantled — they snap toward a strongman (even if they are part of the group doing the dismantling). The threat does not have to be real, only felt. And the very conditions that produce the feeling like social fragmentation, institutional failure, economic precarity, information chaos, are the same conditions that the strongman’s behaviour creates. The authoritarian dynamic is a feedback loop: the chaos produces the demand for the leader who produces the chaos.

More recent research has mapped this dynamic specifically onto Trump’s support base, with results that should alarm anyone who thinks the problem will self-correct.

In 2019, a study published in Nature Human Behaviour (Kunst et al.) demonstrated that a “visceral feeling of oneness” (psychological fusion) with Donald Trump predicted willingness to violently persecute Muslims, immigrants, and Iranians, over and above all other established predictors of political violence. Fusion with the leader, not just agreement with the policies.

In 2024, Martel, Moniz, Ashokkumar, and Swann published findings in Political Psychology showing that identity fusion with Trump, authoritarian values, and animosity toward perceived enemies form “mutually reinforcing fusion clusters.” When threatened, these clusters “motivate attacks against perceived adversaries.” The study drew its data from the January 6th insurrection.

In 2025, Moniz and Swann published in PS: Political Science & Politics (Cambridge University Press) that identity fusion with Trump led supporters to internalise his election fraud claims, which deepened their fusion, which increased support for his policies and dismissal of his legal troubles. A self-reinforcing spiral: the lie strengthens the bond, and the bond makes the lie unfalsifiable.

But like Bondi, everyone is fireable.

Trump demands unwavering loyalty, but will throw anyone (including his supporters) under the bus at any moment, if it benefits him.

Just ask Mike Pence. After refusing to go along with Trump’s fake elector scheme to steal the 2020 election, MAGA supporters broke into the Capitol chanting: “Hang. Mike. Pence.”

Goldsmith and Moen, writing in Political Psychology (2025, DOI: 10.1111/pops.12991), asked the question directly: is this a personality cult? Their answer, based on Big Five personality analysis across two surveys, is that Trump’s most loyal supporters share personality characteristics that are consistent with cult-like devotion.

This is what the mail-in ballot hypocrisy runs into. You can document the contradictions, lies and corruption and it changes nothing when devotion is strong enough.

So: Steve Bannon says flood the zone. Brandolini says the energy to refute exceeds the energy to produce by an order of magnitude. Popper says unlimited tolerance of intolerance destroys the tolerant society. Stenner says the chaos activates the predisposition. Kunst says the fusion predicts the violence. Goldsmith says the personality profile fits a cult.

That is one big pile of shit.

Let’s take a closer look…

2025–2026

Highlights across six categories:

The war lie, the oil grift, the money trails, the Epstein file fiasco, and the institutional damage.

What follows is not comprehensive. A comprehensive record would run to several volumes, and by the time one was written, the next would already be obsolete. This is just a selection.

The general pattern is simple: chaos creates profit. Profit funds power. Power creates more chaos. The people who pay the cost are never the people who collect the returns.

I. The Peace President Goes to War

Donald Trump campaigned on a promise so explicit it left no room for reinterpretation. “I will not start new wars.” He repeated it at rally after rally through 2024, positioning himself as the man who would end the era of American military adventurism.

Fourteen months into his second term, the United States has launched two major military operations on two separate continents.

Iran

In June 2025, after issuing a 60-day ultimatum, the United States joined Israel in strikes on Iranian nuclear sites at Fordow, Natanz, and Isfahan. A ceasefire followed, then collapsed. By early 2026, the Strait of Hormuz was effectively closed. Iran struck targets across the Gulf, hitting US assets in Bahrain. Oil prices, which had been sitting around $70 a barrel, surged past $100 and climbed toward $119. Gas prices for American consumers rose over 22%.

The New York Times and the Washington Post reported that defence officials told congressional briefings the military used $5.6 billion in munitions in the first two days alone. By mid-March, a top Trump adviser told NBC News the war’s price tag had reached $12 billion and counting.

Civilian casualties are mounting. HRANA figures cited by Middle East Eye put the toll above 3,000 killed, including over 1,351 civilians and 207 children. There is no declared Congressional authorisation. There is no exit strategy. The only articulated objective was “obliterate,” a word the president used on Truth Social to describe what he intended to do to Iranian infrastructure, before temporarily postponing strikes on power plants after what were described as “productive talks.”

It does not appear that Iran are taking Trump as seriously as he demands.

You can’t make this stuff up.

Share

Let us hope that those Iranian people who are just trying to live their lives, do not suffer from the increasing rage and unpredictability this “stable genius” is demonstrating.

Meanwhile, Trump appears to be blaming his own defence secretary for starting the war; though they are trying the Putin strategy of calling it an “operation,” but have accidentally (and rightly) called it a war so many times, that it seems silly to go along with the Putin-ism.

At a conference in Tennessee, Trump turned to Pete Hegseth and said: “Pete, I think you were the first one to speak up. And you said, ‘Let’s do it.’” Hegseth made an awkward smile. This is a president reverse-engineering a distance from his own war while it is still happening.

Who benefits?

• American oil producers and LNG exporters are printing money on the price surge.

• Defense contractors (Lockheed, RTX, Northrop) are already lining up for the next $200 billion+ in munitions replenishment.

• The Texas oil patch and the fossil-fuel donors who backed the current administration are getting exactly the “energy dominance” they paid for.

Venezuela

On January 3rd, 2026, US special forces conducted a raid in Caracas, capturing President Nicolás Maduro and his wife. They were flown to New York to face drug trafficking charges. Cuban bodyguards were killed in the operation. Nothing else changed in Venezuela. The political structure remains intact. The population was not consulted about what comes next.

The stated justification was narcotics. The immediate follow-up focused on oil.

Within days, Trump announced that US oil companies would “start making money” in Venezuela. Up to 50 million barrels of Venezuelan oil would be “turned over” to the United States. Executives from Exxon, Chevron, and ConocoPhillips were summoned to the White House and offered what amounts to extraction guarantees in exchange for rebuilding the infrastructure.

Senate Democrats launched a probe into whether Big Oil executives had been briefed on the operation before it happened.

By late February, the US had seized three Venezuelan oil tankers in the Indian Ocean. Trump claimed more than 80 million barrels had been “received.”

Venezuela holds the largest proven oil reserves on the planet.

The United Nations characterised the operation as a potential violation of international law. History suggests the administration won’t need to respond. The last time the United States orchestrated regime change in an oil-rich nation, it was Iran in 1953, which bought roughly 25 years of the Shah’s rule, followed by eternal Iranian hostility and eventually the current war. Guatemala in 1954 hasn’t noticeably improved either. South Vietnam after the US-acquiesced assassination of Diem in 1963 went through seven successive coups that eviscerated its own government.

The Quote

In the middle of all of this, with oil prices spiking and American consumers paying 22% more at the pump, the president said this:

“When oil prices go up, we make a lot of money.”

We. Not you. Not us. Not American families trying to fill their cars. We. The United States is the world’s largest oil producer. When prices spike, producers profit. When producers profit, their stock prices rise. When stock prices rise, the people who own them get richer. When gas prices rise, everyone else gets poorer.

II. The $1.5 Billion Bet

On March 24th, 2026, Donald Trump posted on Truth Social (in all caps, naturally) that the US and Iran had been having “very good and productive conversations” and that he had “instructed the Department of War to postpone any and all military strikes against Iranian power plants and energy infrastructure for a five day period.”

Iran’s parliamentary speaker responded on X: “No negotiations have been held with the US. Fake news is being used to manipulate the financial and oil markets.”

One of those statements is true. Here is what the markets show.

Sky News analysed the trading volume data across three separate markets: WTI crude oil (the US benchmark), Brent crude (the international benchmark), and S&P 500 futures. In all three markets, the pattern was identical. The moment Trump posted, trading volumes spiked and prices moved sharply, oil down, equities up. That part is normal. Presidents move markets.

What is not normal is what happened 15 minutes before the post.

At 10:50 UK time, fifteen minutes before Trump’s Truth Social announcement, there was a massive spike in trading volume across all three markets. On WTI, it was the single largest volume spike of the entire day. On Brent, the same. On S&P 500 futures, the same. Someone, or a group of someones, placed enormous trades in the minutes before a presidential announcement that would move global markets, and they placed them in the direction the announcement would push prices.

Sky News estimated the potential value of those trades at $580 million. As Ed Conway, their economics editor put it: “Life-changing amounts of money. Someone has just made a real killing on this. And it’s not just oil. […] Hundreds of millions of dollars were potentially made in those moments in this curious trade that seemed to come out of nowhere, leaving a lot of people wondering: was this just luck, or did someone actually know what the president was about to say?”

Subscribe now

That was not the only investigation. Jon Sopel and Shelagh Fogarty reported on LBC that on Polymarket, the crypto-based prediction market where bets are placed anonymously and settled in untraceable cryptocurrency, one particular account has been flagged.

On multiple occasions, this account placed bets on when America would launch hostilities, who America would fire at next, or when there would be a peace announcement. The account gained hundreds of thousands of dollars. The bets look too good to be true because they appear to be based on information that was not publicly available at the time.

Polymarket has confirmed it is investigating. The trades are settled via crypto. They are anonymous. They are untraceable.

And at the Securities and Exchange Commission, the body responsible for investigating insider trading, an investigator who had begun examining trades connected to one particular family (spelled T-R-U-M-P) was told to abandon the inquiry. She resigned after being forced out.

None of this is proof, but it is a clear pattern. A $580 million windfall from trades placed 15 minutes before a presidential announcement, across three separate global markets simultaneously. An anonymous Polymarket account that keeps getting war predictions right. And the person whose job it was to investigate whether any of this constituted insider trading being pushed out of her role for investigating.

If someone is making money from advance knowledge of war decisions, that is not corruption in the abstract. People are dying across the Middle East, especially in Iran. Civilians. Children. The Strait of Hormuz is laced with mines. Oil markets are swinging on the president’s words. And someone knew what those words would be before he said them.

Jon Sopel: “It is criminal. It is corrupt. It is corruption seemingly in plain sight.”

III. The Oil Timeline

Late 2025: Big Oil executives are summoned to the White House. The subject is Venezuela. The nature of those meetings is now the subject of a Senate Democratic probe.

January 3, 2026: Maduro is captured. Trump announces oil companies will get access.

January 7: Trump confirms Venezuela will “turn over” up to 50 million barrels.

January 9: Exxon, Chevron, and ConocoPhillips receive “total safety” guarantees at the White House for Venezuelan operations.

February 25: The US seizes a third Venezuelan oil tanker in the Indian Ocean. Trump claims 80 million+ barrels “received.”

Meanwhile: Iran operations keep oil above $100/barrel, enriching every US producer and every portfolio heavy on energy stocks.

At no point has anyone in the administration explained who specifically benefits from the Venezuelan oil arrangement, what the terms are, or how the revenue is allocated. “For the benefit of the people” was the phrasing used. Which people, and how, remains conveniently unspecified.

IV. The Money Trails

The oil grift is visible because bombs are visible and oil is tangible. The financial grift is what goes on in the back rooms, whilst Trump makes a scene in the front.

The Family Crypto Empire

This story has layers, and each one is worse than the last.

In September 2024, Trump (who had previously called cryptocurrency “a scam”) announced that his sons were launching World Liberty Financial. The business model was unclear. By the end of October 2024, the company had sold $2.7 million worth of its token. Then Trump won the election, and everything changed.

Four days before inauguration, on January 16th, 2025, Eric Trump signed a deal giving a 49% stake in World Liberty Financial to Aryam Investment 1, an entity backed by Sheikh Tahnoon bin Zayed Al Nahyan of the United Arab Emirates.

Tahnoon is not a private investor. He is the UAE’s national security adviser. He is the brother of the UAE’s president. He oversees a financial empire worth approximately $1.3 trillion spanning AI, surveillance technology, fish farms, and the investment firm G42, the same G42 that Republican lawmakers had flagged in 2024 over concerns about its Chinese-born CEO Peng Xiao’s links to companies supporting PRC military-civil fusion and human rights abuses.

The deal was secret. World Liberty Financial did not disclose that Tahnoon had purchased the stake. It did not disclose that two Tahnoon affiliates, Martin Edelman (general counsel of G42) and Peng Xiao (CEO of G42), had been placed on World Liberty’s board.

Share

The Wall Street Journal broke the story in February 2026. Documents reviewed by the Journal showed that of the $250 million initial payment, $187 million went directly to Trump family entities (DT Marks DEFI LLC and DT Marks SC LLC), and $31 million went to entities tied to the Witkoff family. The Witkoffs. As in Steve Witkoff. As in the president’s special envoy to the Middle East.

The president’s Middle East envoy’s family received $31 million from a deal with the UAE national security adviser, signed days before the administration took office and began making policy about the Middle East.

Months later, the Trump administration approved a plan to sell the UAE hundreds of thousands of advanced American AI chips, the same chips the Biden administration had refused to provide over concerns they could be diverted to China. ABC News reported there was “a lot of opposition inside the government” to the approval. It happened anyway.

Then came the stablecoin. World Liberty Financial launched USD1, a dollar-pegged token. In May 2025, Tahnoon’s state-backed firm MGX used $2 billion worth of USD1 to finance a deal in crypto exchange Binance, whose founder, Changpeng Zhao, Trump subsequently pardoned.

By December 2025, the Trump family had profited $1 billion on proceeds and held $3 billion worth of unsold tokens. The family receives 75% of net proceeds from token sales.

Congressman Ro Khanna launched a formal investigation. The House Select Committee demanded ownership records, payment details, and internal communications. Khanna’s letter specifically asked whether $187 million had flowed to Trump family entities and whether there was any connection between the investment and the AI chips decision. Legal experts have described the UAE deal as a potential violation of the emoluments clause of the US Constitution.

The White House says the president’s assets are held in a “blind trust managed by his children.” A trust managed by the beneficiary’s own children is not blind. That is not how the word works.

The Kushner Pipeline

Jared Kushner left the White House after Trump’s first term and launched Affinity Partners, a private equity firm. He had no prior experience in private equity. His most notable previous business achievement was nearly bankrupting his family’s real estate company by purchasing a building in New York for $1.8 billion.

Saudi Arabia’s sovereign wealth fund invested $2 billion. The fund’s own screening panel recommended against the deal, citing “the inexperience of the Affinity Fund management,” concerns that Saudi Arabia would bear “the bulk of the investment and risk,” and operations that were “unsatisfactory in all aspects.” Crown Prince Mohammed bin Salman personally overruled the panel.

Since 2021, Affinity Partners has collected approximately $157 million in management fees, including $87 million from Saudi Arabia alone, while generating zero returns for its investors. Zero. The fund had deployed only about $1.1 billion of its $3 billion by mid-2024. The rest sat in a bank account while Kushner charged the Saudis 1.25% annually and other investors 2%. The Senate Finance Committee described these rates as “unusually high given market trends and the relative inexperience” of the firm.

As of two days ago, March 23rd, 2026, Bloomberg reported Affinity’s assets have surged to $6.2 billion, up nearly 30% in a single year. The money kept flowing even before that: by the end of 2024, assets under management had already spiked to $4.8 billion from $3 billion, suggesting that the Saudis significantly increased their commitment to curry favour ahead of Trump’s second term. The New York Times reported this month that Kushner is now actively seeking an additional $5 billion from Middle East governments while simultaneously serving as Trump’s special peace envoy to the region.

Again…, the man negotiating Middle East peace for the United States government is simultaneously soliciting $5 billion from Middle East governments for his personal investment firm.

The Senate Finance Committee referred evidence to the DOJ suggesting Kushner may have acted as an unregistered foreign agent in violation of FARA. Senator Wyden’s investigation documented the fee structure, the lack of returns, and the “questionable deals with foreign governments.” Congresswoman Garcia and Senator Wyden wrote to Affinity Partners on March 19th demanding a detailed schedule of Kushner’s foreign visits since the second inauguration.

In September 2025, Affinity Partners, alongside Silver Lake and the Saudi PIF, began acquiring Electronic Arts in a deal valued at $52.5 billion, the largest leveraged buyout in recorded history. Kushner reportedly “brokered the initial connection.” Ethics experts described the investment as creating “the appearance of potential payback.”

The Senate Finance Committee put it plainly: “Affinity’s investors may not be motivated by commercial considerations, but rather the opportunity to funnel foreign government money to members of President Trump’s family.”

Palantir

Palantir Technologies, co-founded by Trump ally Peter Thiel, has become the administration’s most conspicuous corporate beneficiary.

In 2025, the company secured a $10 billion Army enterprise software contract extendable to a decade, a $795 million boost to the Maven AI programme (total now $1.3 billion), and the ICE “ImmigrationOS” contract for managing deportation operations. Its stock surged over 200%, making it one of 2025’s top performers.

The company now embeds its software across warfighting, deportation enforcement, data consolidation, and the IRS. The more the administration militarises, surveils, and automates enforcement, the more Palantir’s contracts grow. The more Palantir’s contracts grow, the more its stock enriches the network that advocated for those policies.

For a full mapping of how Palantir, Thiel, and the broader surveillance-political network interconnect, see The Peter Thiel Map.

For how the circular AI investment loop inflates the broader market while the rest of the economy stagnates, see The Number Is Going Up.

Follow The Money

Every money trail has the same structure. A policy decision creates a market opportunity. The opportunity is captured by people connected to the decision-maker. The profits flow to the decision-maker’s family, allies, or financial network. The public pays the cost in higher gas prices, military spending, regulatory capture, the quiet erosion of institutional independence, and receives none of the returns.

The administration that promised to fight for “the forgotten men and women of this country” has built the most extensively documented pipeline of self-enrichment in modern American political history. And the worst part is that nobody can hold the full picture because there are too many running simultaneously. By the time you finish understanding the crypto deal, the Kushner story has moved on. By the time you track the Kushner story, Palantir has landed another contract. That is how the strategy works.

V. The Epstein File Fiasco

This section requires a preface.

The Epstein files contain allegations. Many are unverified, uncorroborated, or based on anonymous tips that the FBI itself flagged as “no contact made,” “not credible,” or “DNA inconclusive.” The documents released in the January 30th, 2026 dump include raw FBI tip-line submissions that were never investigated, never confirmed, and in some cases contain no contact information for the person making the claim.

None of what follows constitutes proof that Donald Trump committed any of the specific acts alleged in those tips. The administration has called them “unfounded” and “false” and has threatened lawsuits.

The Flip-Flop

During the 2024 campaign, Trump promised to release the Epstein files in full. Transparency. Everything out.

In early 2025, the administration, through Attorney General Pam Bondi and deputy Todd Blanche, began dragging its feet. Trump lobbied against the proposed legislation, calling it a “Democrat hoax.”

On November 18th, 2025, Congress passed the Epstein Files Transparency Act (H.R. 4405) with a vote of 427 to 1 in the House and unanimously in the Senate. Veto-proof.

Trump signed it the next day. “We have nothing to hide.”

He had spent months trying to hide it.

The Dump

The DOJ missed initial deadlines, released in batches, and on January 30th, 2026, dropped the largest tranche: more than 3.5 million pages, over 2,000 videos, and 180,000 images.

Trump is named between 1,000 and 4,000 times depending on search methodology. Representative Jamie Raskin cited figures suggesting the number may be significantly higher across the full corpus.

The files include FBI internal documents from the National Threat Operations Center (NTOC), compiled in August 2025 by the New York field office’s Child Exploitation and Human Trafficking Task Force. These are tip-line submissions: calls from members of the public routed through the FBI’s threat operations pipeline. The FBI’s own emails, released alongside the tips, discuss follow-up actions (or lack of them), note that many leads went nowhere, and include yellow highlighting on what agents internally described as the “salacious” portions.

Subscribe now

The specific allegations are severe. They span decades and include claims involving minors, sexual assault, and trafficking-adjacent scenarios at Trump properties and social events. Names of family members and associates appear throughout. The full contents are publicly available in the DOJ release; they should be read directly by anyone who wants to understand the scale of what was alleged.

The Scrubbing

What happened after the release is where the story shifts from “messy historical record” to something that requires a better explanation than “technical issues.”

Tracking by journalists, congressional staff, and independent researchers found that Trump-related mentions in the publicly accessible files dropped significantly between the initial release and subsequent checks. From over 5,000 references to under 1,500 in some analyses. The DOJ attributed discrepancies to technical issues. Critics, including Representative McGovern and the transparency organisation American Oversight, said it was deliberate scrubbing. Subpoenas were issued to Attorney General Bondi by the Oversight Committee in March 2026.

The administration described the remaining unreleased material as “untrue and sensationalist.”

The Maxwell Factor

In February 2026, Ghislaine Maxwell, serving a 20-year sentence, appeared in a House deposition and appealed directly to Trump for clemency. She offered to testify that “Trump and Clinton did nothing wrong” in exchange for a reduced sentence.

Trump has previously left the door open. “I’ll take a look.” “I’m allowed to.” The White House currently says clemency is “not being considered.” The door remains ajar.

The Burial

The Epstein releases of late January and February 2026 were smothered with the escalation of the Iran war, which has dominated the news cycle during the weeks when the most damaging Epstein material was becoming public.

Whether the timing was deliberate or coincidental is unknowable. What is knowable is the result: the most extensively documented set of allegations involving a sitting president and a convicted sex trafficker received a fraction of the media attention than the war he started did.

The public could barely process what Trump was being accused of, before he exonerated himself, fired Bondi (messy job “cleaning” Trump’s name from the files can’t have helped), and started a war.

VI. The Institutional Damage

The contradictions and the grift are the visible symptoms. The institutional damage is the structural disease, and it is the one that will outlast this administration regardless of what happens next.

The Federal Reserve

The DOJ opened an investigation into Fed Chair Jerome Powell. The broader legal strategy, built on the “unitary executive theory,” seeks to establish the president’s right to fire the heads of independent agencies at will.

The Federal Reserve’s independence is the structural foundation on which every treasury bond, every dollar-denominated contract, every central bank reserve holding in the world is priced. If monetary policy becomes a tool of a president who wants rates lowered before midterms, the consequences extend beyond domestic politics. They extend to whether the US dollar remains the world’s reserve currency. Which is to say: they extend to everything.

The Scientific Base

Research programmes have been shut down. Not wound down, not transitioned. Shut down suddenly, without warning. Scientists in federal labs have had to turn off refrigerators containing years of irreplaceable experimental data.

The United States won the Cold War in significant part by maintaining a decisive technological advantage over the Soviet Union, an advantage built by decades of sustained public investment in research. Dismantling the research base is the permanent destruction of a strategic asset that took generations to build, in exchange for a temporary political benefit that will be forgotten within a fiscal year.

The Enforcement Apparatus

ICE operations have expanded in both scope and intensity. Deportation infrastructure has been automated through the same Palantir contracts described above. The DOJ has been redirected from its investigative functions toward the president’s personal vendettas. Specific individuals have been named on Truth Social for prosecution. The attorney general has been subpoenaed by the Oversight Committee. The agency tasked with upholding the law is being used to punish critics and protect allies.

The Sum

Every institution that exists to provide an independent check on executive power, the Fed, the judiciary, the scientific establishment, the intelligence community, the regulatory agencies, the military’s ethical constraints, is under simultaneous, consistent, and intentional pressure.

“Highlights”

Purging agencies of thousands of staff, possible use of AI technology to screen for ideological alignment. Deploying the National Guard without state consent. Attempting to secure exclusive federal control of laws surrounding AI. Gutted healthcare funding to boost ICE funding to $75-100 Billion (10x increase). £1.5 Trillion defense budget. Large tax cuts for the rich. Normal people pay significantly more for everything.

His cabinet’s combined net worth exceeds $450 billion. Peter Thiel’s surveillance company has $10 billion in military contracts. Oil executives get White House summits. Crypto partners get regulatory capture. The swamp was privatised and sold.

Ran on exposing and prosecuting the Epstein files. 180’d in office. Lobbied against the Epstein bill. Called it a hoax. Signed under pressure. Files scrubbed post-release. Still millions held back (in contradiction to bill). Exposed victim names whilst redacting names of linked individuals.

Tariffs have added an estimated $1,700 per year to the average American household. Gas prices up 22%. Polls show 61 to 76% of Americans say Trump has made the cost of living worse. Recession fears at 76%.

Found liable for sexual abuse in the E. Jean Carroll case. Named thousands of times in the Epstein files. Weaponised the DOJ against political opponents. Named specific individuals for prosecution on Truth Social. Threatened pollsters with lawsuits for producing numbers he doesn’t like.

Designated Anthropic, the AI company that refused to allow its technology to be used for mass surveillance and autonomous weapons, a “supply chain risk,” forcing its removal from government systems and pressuring its commercial ecosystem. Replaced it with OpenAI, whose deal with the Pentagon was announced the same week.

This is the wealthiest administration in modern American history. The president’s family operates a cryptocurrency empire funded by Gulf royals. The vice president was selected by Peter Thiel. The special envoy’s firm just hit $6.2 billion in Gulf-funded assets while he negotiates Middle East peace. Every economic metric that matters to ordinary people, housing, real wages, consumer debt, healthcare, has worsened.

The Historical Mirrors

Sarah Paine spent 25 years teaching military officers how empires rise and fall. She is the William S. Sims University Professor Emerita of History and Grand Strategy at the US Naval War College. Her education includes a BA from Harvard (magna cum laude), an MIA from Columbia, an MA in Russian from Middlebury, and a PhD in history from Columbia, built on nine years of field research in China, Taiwan, Russia, Japan, and Australia. She is a Senior Fellow at the Henry A. Kissinger Center for Global Affairs at Johns Hopkins. She writes for The Economist and Foreign Affairs. Her books are published by Cambridge University Press.

Share

She is someone who has spent her professional life studying the specific mechanisms by which powerful nations destroy themselves.

Everything described in this article looks, from the inside, like a series of scandals. From the outside, from the perspective of someone who studies these dynamics across centuries and civilisations, it looks like something more specific. It looks like a country in the process of strategic suicide.

The Operational Trap

There is a difference, Paine argues, between operational thinking and strategic thinking. Operational thinking is about the immediate: how do I stay in power, how do I win the next news cycle, how do I get re-elected. Strategic thinking is about the generational: will my children be safer, will the country be stronger, will the alliances hold?

The current US leadership, in Paine’s assessment, is thinking entirely at the operational level. “It’s how I personally stay in power, whether it’s in the House of Representatives, the Senate, or the presidency. It has nothing to do with improving the lives of their children. It is actively undermining those things.”

“Think about how the United States did well in the Cold War,” Paine says. “It was by being way ahead of the Soviets technologically and then denying them the technology. Shutting down research programmes is not going to get you there. And certainly not shutting them down suddenly. So you have people in labs having to turn off refrigerators that kill years of data. Who does that? That’s incompetent.”

“Why would you blow your alliance system? That’s beyond stupid. Why would you blow up your scientific expertise? That’s beyond stupid.”

Stupid is the word she keeps coming back to. Not evil, not conspiratorial. Stupid. The kind of stupidity that comes from people who cannot see past their own reflection. I would add “greedy.” It is astonishing how much two words can explain.

The Parthenon

Paine has a way of making historical parallels feel like warnings rather than analogies.

“The ancient Athenians started getting abusive to their allies. They overextended militarily. And then they had very dysfunctional politics back home. And it was irretrievable for ancient Athens. They never recovered their pre-eminence.”

“Go look at the Parthenon. It stands in mute testimony to hubris and the foolish things that people can do. Americans aren’t any different.”

Athens was the dominant power in the Greek world, the anchor of the Delian League, the city that invented democracy. It started treating its allies as subjects. It launched military adventures it couldn’t sustain. Its domestic politics became so dysfunctional that strategic thinking collapsed into factional warfare. And it never came back.

Paine sees the same trajectory playing out now. “I’ve asked many Americans: what has a Canadian ever done to you? And Americans can’t think of anything. And yet we’re launching a tariff war with them.”

The alliances are not peripheral to American power. The post-war order, the NATO architecture, the trade frameworks, the diplomatic relationships built over 80 years; these are the infrastructure that made the United States the most influential nation in history.

“If the United States continues doing this, it will unify all countries in the world against it. And that will be a profound blunder. That takes real incompetence. But that is where the United States is headed.”

Trump can destroy the United States’ participation in NATO, she says. He cannot destroy NATO itself. “All of the organising structures that Europeans have, that doesn’t have to go away. It’s just the big boy’s having a tantrum […], it’ll make the United States profoundly weaker.”

She watches the Republican Party go along with it and finds it incomprehensible. “It just makes no sense. They need to come to their senses to save their children and their grandchildren.”

The Proliferation

The alliance destruction has a consequence that Paine considers the most dangerous development of all.

“With all this happening, we’re going to have nuclear proliferation all over the place.”

If the United States withdraws its nuclear umbrella, if the security guarantees that have kept allied nations from developing their own arsenals are no longer credible, those nations will do the obvious thing. A Swedish newspaper has already called for nuclear weapons. Paine expects Germany will pursue them. Poland almost certainly. Japan. Taiwan has the intellectual expertise.

“A whole new world of high risk.”

The current nuclear order, for all its flaws, involves a small number of actors who have spent decades developing communication channels, deterrence frameworks, and crisis management protocols. Replace that with a dozen new nuclear states, each with their own threat perceptions and red lines, and you have a world where the probability of miscalculation goes up dramatically.

“Having a bunch of nuclear weapons in Europe, that will truly not make Russia safer,” Paine notes.

The Iran war is actively accelerating the collapse of the non-proliferation framework. Every country watching the United States bomb Iran while abandoning its allies is doing the same calculation: if we don’t have our own deterrent, we’re next.

For the billionaires who think they will ride this out, Paine has news.

“If it comes to a world war, their money will evaporate. And if people blame them for it, which they might, well.., they will know where to come get them.”

She tells a story about meeting someone who works for McKinsey who mentioned an underground bunker in New Zealand. Some billionaire’s contingency plan for the end of civilisation.

“If the nukes fly and there are any New Zealanders left, I assure you that whoever the American is in the underground bunker who helped deliver this dystopian universe, the New Zealanders will reintroduce capital punishment, especially for them.”

The billionaires, she argues, have the most to lose if the world order collapses, because their billions depend on having rules.

The people profiting most from the current trajectory are the people who will lose the most if it reaches its logical conclusion. The system that made them rich is the system they are dismantling. And they apparently believe they can dismantle the floor they’re standing on and somehow remain standing.

Paine studied under Carl Deutsch at Harvard. Deutsch was a Jewish refugee from Czechoslovakia who escaped as the Anschluss unfolded, got a scholarship, and became one of the founding figures of American political science.

One of the last times Paine saw him, he said something she has carried ever since.

“If the forces of evil were greater than the forces of good, we wouldn’t be here. But there have been some really close calls.”

The rules-based order is not perfect. It never has been. It did not prevent every war, every injustice, every exploitation. But it created something that two generations ago would have seemed impossible: a world where France and Germany are not deploying armies against each other, where a person can travel between countries by showing a plastic card, where disputes between nations are resolved (imperfectly, incompletely) through negotiation rather than conquest. The trend line, across centuries, is toward more legality, more cooperation, more restraint. Not in a straight line. Not without catastrophic setbacks. But generally.

“Human beings are wonderful,” Paine says. “But we’ve got to bring out the best in each other and put a lid on the people who are insisting on trying to bring out the worst.”

“The rules-based order must survive. Because the alternative is a 1984 dystopian universe. And it is unnecessary.”

On March 17, 2026, the gross national debt of the United States crossed $39 trillion. Thirteen days earlier, when the Joint Economic Committee published its monthly update, the figure stood at $38.86 trillion — $288,283 for every American household, $113,638 for every person. It had grown by $2.64 trillion in a single year. That works out to $7.23 billion per day, $301 million per hour, $83,720 per second.

The JEC estimates that at the current average daily growth rate, each additional trillion dollars now arrives in approximately 148 days. Five years ago, the total was $28 trillion. A decade before that, $18 trillion. The curve does not flatten.

To make the mechanics of what follows intuitive, consider a heavy truck on a steep hill. For decades, the gradient was gentle and the brakes — economic growth, manageable interest rates, the occasional political will toward fiscal discipline — were more than adequate. That era is ending. The hill has steepened. The truck is heavier. And the brakes are losing grip.

How We Got Here — The Debt Since 1990

The debt trajectory since 1990 tells a story of four distinct acceleration phases, each triggered by a crisis or major policy shift, none of them ever fully reversed.

In 1990, total gross national debt was $3.23 trillion. Over the following decade it climbed to $5.67 trillion — but the rate of increase slowed dramatically. The late 1990s were the slowest growth period in modern fiscal history, thanks to budget agreements, welfare reform, and the revenue surge from the dot-com boom. The federal government ran surpluses in 1998, 1999, 2000, and 2001. That last surplus, in FY2001, came in at roughly $127 billion. It remains the final one. The United States has run a deficit every single year since — twenty-five consecutive years and counting.

The reversal came quickly. The September 11 attacks, the wars in Afghanistan and Iraq, the Bush tax cuts of 2001 and 2003, and the Medicare expansion collectively added $3.2 trillion through the 2000s, pushing annual increases into the $400–600 billion range. Then came the first true inflection point: the 2008 global financial crisis. Debt surged from $10 trillion to $16 trillion in four years as TARP bailouts, quantitative easing, and the $800 billion ARRA stimulus kicked in. FY2009 alone added $1.89 trillion — the largest single-year increase at that point. On any long-term chart, this is the first visible upward bend.

A new elevated baseline held through the 2010s. No crisis, but annual deficits of roughly $1 trillion became normal, sustained by the 2017 tax cuts and routine gaps between spending and revenue. The debt crossed $20 trillion in 2017 and reached $22.72 trillion by 2019.

Then COVID broke the scale entirely. Between FY2020 and FY2022, debt rocketed from $22.72 trillion to nearly $31 trillion. FY2020 alone saw a $4.23 trillion increase — by far the largest in history — as the CARES Act and American Rescue Plan pushed federal spending up roughly 50% in two years. The daily growth rate peaked at around $11.6 billion.

The pattern across three decades is unmistakable: each global shock ratchets the baseline permanently upward. Debt never comes back down. Emergency spending becomes the new floor, and the next shock builds on top of it.

The Interest Trap

The post-COVID era has introduced something qualitatively different — a force that is rapidly becoming the dominant driver of debt growth, independent of any new spending decisions. That force is interest on existing debt.

Subscribe now

In FY2025, net interest payments on the public debt surpassed $1 trillion for the first time. Interest is now the second-largest federal expenditure, exceeded only by Social Security. It exceeds the entire defence budget.

That $970 billion figure deserves context beyond the abstract. In FY2025, the federal government spent more on interest than it spent on national defence ($893 billion), more than on Medicaid ($668 billion), more than on income security programmes and veterans’ benefits combined. Only Social Security ($1,575 billion) and Medicare ($1,180 billion) commanded larger shares of the budget. Interest — which builds nothing, employs nobody, and delivers no public service — now outranks every discretionary priority the government funds.

This is the quiet displacement that makes the compounding problem so corrosive. Every dollar spent servicing past borrowing is a dollar unavailable for everything else. That means less for infrastructure — roads, bridges, broadband, the physical systems people rely on daily. Less for education, research, veterans’ care, disaster relief. Less for the programmes that working and middle-class families depend on when things go wrong.

And the squeeze only tightens from here: CBO projects that Social Security, Medicare, and interest will together consume an ever-growing share of the budget, leaving less and less room for everything else. Defence spending is already projected to fall to its smallest share of GDP since before World War II. Non-defence discretionary spending — the category that funds most of what government visibly does for ordinary people — is heading toward its lowest share since the early 1960s.

The endgame, if nothing changes, is a government that can pay its creditors and write its pension cheques, but struggles to do much else. And unlike Social Security or Medicare, interest payments cannot be reformed, means-tested, or restructured through legislation. They are contractual obligations to bondholders. As long as the debt grows and rates remain elevated, the interest line will continue to expand — mechanically, automatically — crowding the rest of the budget into an ever-narrower space. The cost of decades of borrowing eventually lands on the everyday experience of public services that quietly get worse, year after year, until people forget they were ever better.

The Peterson Foundation reports that CBO projects net interest payments will total $16.2 trillion over the next decade, rising from $1.0 trillion in 2026 to $2.1 trillion by 2036. As a share of federal revenues, interest payments reached 18.5% by the end of FY2025, surpassing the previous high set in 1991.

The JEC data traces the mechanism. The average interest rate on total marketable debt stood at 3.348% as of January 2026. Five years earlier, it was 1.541% — less than half. That rate has more than doubled, and the effect on a $39 trillion debt stock is enormous. The JEC forecasts net interest consuming 13.85% of all federal outlays in FY2026, rising to 14.52% by FY2028. Every tick upward in the rate, applied to a growing principal, produces a larger interest bill — which then gets added to the debt, which then requires still more interest.

This compounding dynamic has a name in economics: R > G. When the average interest rate on the debt (R) exceeds the nominal growth rate of the economy (G), the debt begins to compound faster than the economy can absorb it. The existing stock of debt generates its own growth, regardless of what Congress does or does not spend.

According to the Committee for a Responsible Federal Budget, analysing CBO’s latest projections, R will exceed G starting in fiscal year 2031 — five years from now. At that point, both are projected to sit at roughly 3.8% nominal. After 2031, the gap widens: by 2056, CBO projects an interest rate of 4.2% against a GDP growth rate of just 3.5%.

Here the truck analogy earns its full weight. On a gentle slope, brakes work straightforwardly; apply them and the vehicle slows. On a steeper slope at higher speed, you press harder, but they still function. There is, however, a physical threshold — a combination of gradient, speed, and mass — beyond which the brakes cannot generate enough force to overcome the momentum. The driver presses the pedal to the floor and the truck still accelerates. The brakes have not failed. They are simply no longer strong enough to do the job. The physics have changed.

That is what R > G means in practice. Once the interest rate on the debt exceeds the growth rate of the economy, even eliminating the entire primary deficit (balancing all spending against all revenue, excluding interest) would not be enough. The debt would still grow, because the interest payments alone exceed what economic growth can absorb. Deficits become self-sustaining, driven not by new spending decisions but by the accumulated cost of old ones.

And just as a heavier truck on a steeper hill requires exponentially more braking force, the longer you wait past the R > G threshold, the larger the fiscal adjustment needed to stabilise the trajectory. The CRFB calculates that closing the gap would require roughly $2.7 trillion in annual spending cuts or tax increases by 2056. For context, the entire annual discretionary budget is roughly $1.7 trillion. That figure grows every year that passes without action. The hill gets steeper. The truck gets heavier. And the braking force required to stop it moves further beyond what any plausible political consensus could deliver.

Before 2031, difficult reforms can still bend the curve. After it, the maths shifts from “hard but possible” to “theoretically possible but practically inconceivable.” That is the distinction between brakes that are straining and brakes that are physically overwhelmed.

The Maturity Wall

The interest rate story is alarming in the abstract. The maturity of U.S. debt makes it worse in practice.

Share

According to the JEC, approximately 33% of publicly held marketable debt will mature within twelve months. The average maturity of outstanding debt is just 70 months. Of the $30.92 trillion in total public debt outstanding, 50.83% is in notes (2–10 year maturities), 21.33% in bills (under a year), and 17.01% in bonds (20–30 years).

The implication: the U.S. Treasury must constantly roll over enormous volumes of existing debt, and every dollar refinanced now costs significantly more than when it was originally issued. In FY2025, approximately $9.2 trillion of debt matured — roughly a quarter of the entire national debt. The Treasury missed the window to lock in cheap long-term rates during 2021–2022, when rates sat at historic lows. It is now trapped in what amounts to a rolling short-term borrowing strategy at elevated costs.

The bid-to-cover ratios — measuring demand for each dollar of debt sold — remain reasonable for now: 2.85 for 4-week bills, 2.38 for 10-year notes, 2.29 for 30-year bonds. But two forces are working against sustained demand. China has been systematically reducing its Treasury holdings, from over $1 trillion a few years ago to $730.7 billion by July 2025 and further to $682.6 billion by November 2025 — the lowest since September 2008. Holdings declined more than 10% in 2025 alone, as Beijing simultaneously increased gold reserves for fourteen consecutive months. Meanwhile, Japan — the largest single foreign creditor at $1.15 trillion — is seeing rising domestic yields that may pull institutional capital back home, reducing a critical source of demand for American debt precisely when supply is surging.

The sovereign maturity wall is also colliding with a corporate one. Approximately $1.35 trillion in non-financial corporate debt needs refinancing in 2026. Credit spreads have widened sharply, with high-yield spreads reaching 470 basis points by mid-March. Large-company bankruptcy filings in the first half of 2025 were up 81% over the long-term average. Government and corporate borrowers are competing for the same pool of capital at higher rates simultaneously — the point at which the sovereign debt problem begins bleeding into the real economy.

Meanwhile, total U.S. household debt crossed $18.8 trillion in Q4 2025, with credit card balances alone hitting a record $1.28 trillion — carried overwhelmingly at APRs above 20%. Delinquency rates worsened to 4.8% in aggregate, with student loans at 9.6%. The elevated rate environment that makes government borrowing expensive is the same one trapping millions of households in compounding unsecured debt. And the feedback loop runs both ways: as consumer distress deepens and spending contracts, GDP growth weakens — which worsens the sovereign R > G dynamic. The truck gets heavier from below.

The Credit Rating That Vanished

On May 16, 2025, Moody’s Ratings downgraded the United States from Aaa to Aa1, citing persistent fiscal deficits, rising debt and interest burdens, and political gridlock. The move made Moody’s the last of the three major credit agencies to strip the U.S. of its top rating — S&P had downgraded in 2011, Fitch in 2023. For the first time in over a century, the United States no longer holds a perfect credit rating from any major agency.

Moody’s assessment: federal debt could reach 134% of GDP by 2035, with deficits widening to nearly 9%, driven mainly by interest costs and entitlement spending. The agency observed that successive administrations and Congress have consistently failed to agree on measures to reverse the trend. Moody’s had maintained its top U.S. rating since 1919.

The immediate market reaction was muted; investors had largely priced in the inevitability. But the symbolic weight is hard to overstate. The United States has been the world’s benchmark risk-free borrower since the modern financial system was built. That status still rests on the dollar’s reserve currency role and the unmatched depth of the Treasury market. The signal from all three rating agencies, however, is now unanimous: the trajectory is unsustainable, and nobody in Washington is changing it.

The Iran-Oil Accelerant

On February 28, 2026, joint U.S.-Israeli strikes on Iran — including the killing of Supreme Leader Ali Khamenei — triggered a retaliatory campaign that has reshaped the energy landscape of the entire Middle East. The Strait of Hormuz, through which roughly 20% of global seaborne oil normally transits, has been effectively closed. The International Energy Agency has called it the largest disruption to the energy supply in the history of the global oil market.

The scale of the military campaign is staggering. In the opening strikes alone, some 40 senior Iranian officials were killed, including Khamenei, the defence minister, the chief of staff, and the head of the IRGC. The U.S. and Israel have since conducted sustained aerial bombardment across Iran, hitting military, intelligence, broadcasting, and parliamentary targets in Tehran, Isfahan, Tabriz, and elsewhere. Iran’s Red Crescent reported over 6,668 civilian sites targeted. More than 3,000 people have been killed, including over 1,300 civilians. Three million Iranians have been displaced.

On March 13, the U.S. conducted what Trump called “one of the most powerful bombing raids in the history of the Middle East” on Kharg Island — the tiny Persian Gulf hub that handles 90% of Iran’s oil exports. CENTCOM stated it destroyed over 90 military targets including naval mine storage and missile bunkers, while sparing oil infrastructure, though Trump openly threatened to strike the pipelines next, saying it could be done “on five minutes’ notice.” Reports of U.S. Marines being sent to the Gulf have fuelled speculation about a possible seizure of the island.

Iran’s response has gone far beyond the strait. What began as retaliation against U.S. military assets has widened into a campaign targeting energy infrastructure across the entire Gulf. The UAE has reported more than 2,000 Iranian drones and missiles since the war began.

Bahrain says it has intercepted 125 missiles and 203 drones, with at least 24 people killed across Gulf nations. Fires broke out at the Fujairah oil terminal, Dubai International Airport was hit by a drone-related incident, and Oman’s ports at Duqm and Salalah — potential bypass routes for the strait — were struck.

On March 18, Israel struck Iran’s South Pars gas field — the largest in the world, shared with Qatar — damaging facilities and killing Iran’s intelligence minister. Tehran’s response was immediate and sweeping: it issued formal evacuation warnings to energy workers at named facilities across Saudi Arabia, the UAE, and Qatar, then launched missile and drone strikes on all three countries. Qatar’s Ras Laffan LNG complex (the world’s largest) suffered extensive damage from Iranian ballistic missiles. Saudi Arabia’s SAMREF refinery in Yanbu was hit by a drone. The UAE shut down its Habshan gas complex and Bab oil field after interceptions. Two Kuwaiti refineries caught fire. Saudi Arabia intercepted ballistic missiles over Riyadh.

The Gulf states — who had tried to remain neutral, even restricting U.S. use of their bases — are now openly warning that patience is running out. Saudi Arabia’s foreign minister declared the kingdom reserves the right to military action. A senior UAE official said the attacks had created “a huge trust gap that will last for decades.” Gulf states have requested an urgent debate at the UN Human Rights Council. The Arab League called the strikes “a dangerous escalation.” Formula One races in Bahrain and Saudi Arabia have been cancelled.

Oil markets reflect the chaos. Brent crude surpassed $100 per barrel on March 8, peaked near $126, and was trading above $108 after the South Pars strikes. Prices have risen more than 40% since the war began, with some analysts and traders now openly discussing $150–$200 as realistic if the disruption persists. The IEA coordinated its largest-ever emergency stock release — 400 million barrels — but that covers barely four days of global consumption. It cannot substitute for an open shipping corridor.

For the debt trajectory, every dimension of this conflict worsens the picture. Direct military costs are accumulating; a senior Trump adviser put the war’s price tag at $12 billion so far. But the larger fiscal damage runs through the interest rate channel. Higher oil prices feed directly into inflation, which keeps the Federal Reserve hawkish, which keeps borrowing costs elevated, which increases the government’s interest bill on both new issuance and the constant churn of maturing debt. Axios has reported that prolonged disruption would create a moderate stagflationary drag on the U.S. economy and a substantial one on Europe and East Asia. Goldman Sachs has raised its 2026 U.S. recession probability.

And there is a deeper risk. The war is not just disrupting supply through the strait, it is actively destroying production and export infrastructure across the Gulf. Damage to facilities like Ras Laffan and South Pars could take years to rebuild, meaning supply constraints may persist long after any ceasefire.

In terms of the truck analogy, the war is not the hill, but it is where it greatly steepens, at the worst possible moment.

The Global Comparison — Why America Is the Outlier

The U.S. debt level, taken in isolation, does not look uniquely alarming on the world stage. Japan is at 230% of GDP, Italy at 138%, Greece at 153%. What makes the United States the genuine outlier is the direction, speed, and absence of structural brakes.

Europe: Consolidating

The EU’s debt-to-GDP ratio stood at 81.0% at the end of Q4 2024, with the euro area at 87.4%. The aggregate deficit fell from 3.4% of GDP in 2023 to 3.1% in 2024 — roughly half the American figure. Individual country stories are striking: Greece’s ratio has fallen from 180% in 2022 to 153%; Italy narrowed its deficit from 7.2% to 3.4% in a single year.

The EU has institutional mechanisms the U.S. lacks entirely. The Stability and Growth Pact sets binding limits of 60% debt-to-GDP and 3% deficit-to-GDP, with excessive deficit procedures and mandatory medium-term fiscal plans creating political accountability for overspending. The Bipartisan Policy Center notes that countries such as Switzerland, South Korea, and the Netherlands have balanced or nearly balanced their primary budgets through stricter fiscal rules. Europe faces its own tension. The ECB estimates additional investment needs of €1.2 trillion per year for defence and strategic priorities, but the scaffolding for fiscal discipline exists.

China: The Iceberg

China’s official government debt-to-GDP ratio of roughly 84% dramatically understates the reality. Total social financing — the broadest measure of credit in the Chinese economy — reached 309% of GDP by mid-2025, and the IMF notes that China accounts for more than half of the increase in the global debt-to-GDP ratio since 2008.

Subscribe now

The hidden debt problem is vast. Fitch has estimated that the restructured local government financing vehicle (LGFV) debt covers only about a quarter of the actual hidden debt load. China’s own government conceded a 44 trillion yuan overall LGFV figure to the IMF, three times the number given during the 2024 restructuring announcement. The Bank of Finland estimates the true broader public sector deficit at 13–14% of GDP, with actual debt around 130% — comparable to or exceeding the United States on a like-for-like basis.

The critical difference: Beijing controls its banks, its central bank, and its domestic bond market. It can administratively roll over obligations, suppress interest rates, and direct losses. The risks are real, but the mechanism of any future crisis would be fundamentally different from a market-driven sovereign debt spiral.

Japan: The Inflection

At 230% of GDP, Japan has carried the world’s highest debt ratio for decades without crisis, protected by three shields: nearly all debt held domestically, rates held near zero by the Bank of Japan, and the world’s largest net creditor position at roughly $3.66 trillion. All three shields are now weakening.

The BoJ raised rates to 0.75% in December 2025, a thirty-year high. Japan’s FY2026 budget hit a record ¥122.3 trillion, with debt servicing surging 10.8% to ¥31.3 trillion, the first time above ¥30 trillion. The Finance Ministry is budgeting for bond payments at 3%, the highest assumption since 1997. The IMF warns that interest costs are expected to double by 2030 and that public debt will resume rising.

For the U.S., the Japan story carries a direct secondary risk. As Japanese domestic yields rise, institutional investors have less incentive to hold foreign bonds, including Treasuries. Any significant repatriation of Japanese capital would reduce a critical source of demand for American debt at the very moment supply is surging.

The American Exception

The United States stands alone in combining four features simultaneously. First, crisis-level deficits during peacetime prosperity: FY2025’s deficit of 5.9% of GDP was the highest among G7 nations.

Second, no institutional mechanism to force fiscal consolidation, no fiscal rules, no balanced budget requirement, no binding medium-term framework. Third, full exposure to market interest rates on a debt stock partly held by foreign creditors who are actively diversifying away. Fourth, a self-reinforcing compounding loop that is approaching the point where the mathematics become inescapable.

Europe has the Stability and Growth Pact. China has state-controlled finance. Japan has domestic ownership and creditor-nation status. The U.S. has none of these buffers. The American fiscal system relies entirely on the political will of Congress, and that will has not materialised in any meaningful form since 2001.

The Window Is Closing

The Congressional Budget Office does not use the word “unsustainable” lightly. It uses it now. The Government Accountability Office uses it. The CRFB uses it. All three major credit rating agencies have, in their own institutional language, said the same thing.

CBO projects deficits of $1.9 trillion in FY2026 growing to $3.1 trillion by 2036 — from 5.8% to 6.7% of GDP. Debt held by the public is forecast to rise from 101% of GDP to 120% by 2036, surpassing the all-time record of 106% set after World War II. Under a scenario modelled by the National Bureau of Economic Research, if interest rates respond to rising debt levels — climbing roughly 3 basis points for every percentage point increase in debt-to-GDP — the ratio could reach 233% by 2054.

The One Big Beautiful Bill Act, signed in July 2025, is estimated by CBO to add $4.7 trillion to deficits through 2035. Tariffs are projected to offset roughly $3.0 trillion, but are themselves legally contested. Under an adverse scenario where the Supreme Court strikes down many tariffs and temporary tax provisions become permanent, debt could reach 131% of GDP by 2036 rather than 120%. The CRFB’s assessment: in that case, a debt spiral becomes far more likely and the risk of a fiscal crisis grows.

What would adequate brakes look like? Stabilising debt-to-GDP after the R > G flip would require running a primary surplus of 0.7% of GDP, the equivalent to roughly $2.7 trillion in annual adjustments by mid-century. Social Security, Medicare, and net interest together already account for more than all of the projected growth in government spending.

The truck passed the crest somewhere around 2020–2023, when pandemic spending, rising rates, and the compounding of a $30+ trillion debt stock all converged. It is now on the gradient and accelerating. The Iran conflict is adding downhill momentum.

Foreign creditor dynamics are shifting unfavourably. The maturity wall means the cost of existing debt reprices continuously upward. And unlike Europe, China, or Japan, the United States has no institutional mechanism — no fiscal rules, no state-controlled banking system, no domestic ownership advantage — to buffer the impact.

None of this is a prediction of imminent collapse. The dollar’s reserve currency status buys time that no other country on earth could expect. But time is not the same as safety.

Every year of inaction makes the eventual adjustment larger, more painful, and more likely to be imposed by bond markets rather than chosen by policymakers. The GAO, the CBO, the CRFB, and every major independent fiscal institution have sounded the same alarm. The necessary tools exist. The political will to use them does not.

The truck is on the hill. The brakes still work… for now. But there is a point on the road ahead where the gradient and the momentum will exceed the maximum braking force that any policy response can generate. Past that point, pressing the pedal to the floor will not stop the truck from accelerating. It will just mean someone tried. That threshold is approximately 2031. The window between now and then is the window in which steering still matters. After that, physics takes over.

Share

Sources & References

Primary data throughout this article is drawn from the Joint Economic Committee’s Monthly Debt Update (February and March 2026 editions), the Congressional Budget Office’s Budget and Economic Outlook: 2026 to 2036, and the U.S. Treasury’s Fiscal Data portal. The full interactive JEC page is available at: https://www.jec.senate.gov/public/vendor/_accounts/JEC-R/debt/Monthly%20Debt%20Update.html

Committee for a Responsible Federal Budget, “Trillion-Dollar Interest Payments Are the New Norm,” December 16, 2025. Confirms $39T crossed March 17, 2026. https://www.crfb.org/blogs/trillion-dollar-interest-payments-are-new-norm

U.S. Treasury Fiscal Data, National Deficit. Confirms last surplus FY2001 (~$127 billion). https://fiscaldata.treasury.gov/americas-finance-guide/national-deficit/

Congressional Budget Office, Monthly Budget Review: Summary for Fiscal Year 2025, October 2025. https://www.cbo.gov/publication/61307

American Action Forum, “CBO: FY 2025 Budget Deficit Totaled $1.8 Trillion,” October 2025. https://www.americanactionforum.org/insight/cbo-fy-2025-budget-deficit-totaled-1-8-trillion/

Peter G. Peterson Foundation, Interest Costs on the National Debt, March 2026. https://www.pgpf.org/programs-and-projects/fiscal-policy/monthly-interest-tracker-national-debt/

Joint Economic Committee Republicans, Monthly Debt Update, February 2026. https://www.jec.senate.gov/public/vendor/_accounts/JEC-R/debt/Monthly%20Debt%20Update.html

Committee for a Responsible Federal Budget, “CBO Projects Possible Debt Spiral, as R Exceeds G,” March 9, 2026. https://www.crfb.org/blogs/cbo-projects-possible-debt-spiral-r-exceeds-g

China Daily, “China Pares Down US Treasury Holdings,” September 20, 2025. https://global.chinadaily.com.cn/a/202509/20/WS68cdfc20a3108622abca1e28.html

Global Times, “China Trims US Treasury Holdings by $6.1 Billion in November, Reaching Lowest Since 2008,” January 2026. https://www.globaltimes.cn/page/202601/1353386.shtml

Euronews, “Bank of Japan Hikes Interest Rates: Is a Global Bond Crisis Looming?” December 19, 2025. https://www.euronews.com/business/2025/12/19/bank-of-japan-hikes-interest-rates-is-a-global-bond-crisis-looming

CNBC, “Moody’s Lowers U.S. Credit Rating to Aa1,” May 16, 2025. https://www.cnbc.com/2025/05/16/moodys-downgrades-united-states-credit-rating-on-increase-in-government-debt.html

Newsweek, “US Completely Loses Perfect Credit Rating for First Time in Over a Century,” May 17, 2025. https://www.newsweek.com/moodys-us-credit-rating-negative-2073510

Wikipedia, “2026 Strait of Hormuz Crisis.” https://en.wikipedia.org/wiki/2026_Strait_of_Hormuz_crisis

CNN, “Oil Prices Soar Above $100 After Iran Says Strait of Hormuz Will Remain Shut,” March 12, 2026. https://www.cnn.com/2026/03/12/energy/oil-jump-record-reserves-release-intl-hnk

Al Jazeera, “Oil Stays Above $100 a Barrel Amid Iran’s Stranglehold on Strait of Hormuz,” March 13, 2026. https://www.aljazeera.com/economy/2026/3/13/oil-stays-above-100-a-barrel-amid-irans-stranglehold-on-strait-of-hormuz

Al Jazeera, “Strategic Oil Release May Calm Markets but Cannot Fix Hormuz Disruption,” March 15, 2026. https://www.aljazeera.com/economy/2026/3/15/strategic-oil-release-may-calm-markets-but-cannot-fix-hormuz-disruption

Axios, “Oil Prices, Recession: What Happens If Strait of Hormuz Stays Closed,” March 12, 2026. https://www.axios.com/2026/03/12/oil-prices-iran-strait-of-hormuz

Eurostat, “Government Debt at 87.4% of GDP in Euro Area,” April 22, 2025. https://ec.europa.eu/eurostat/web/products-euro-indicators/w/2-22042025-bp

Visual Capitalist, “Mapped: European Union Debt-to-GDP by Country,” October 2025. https://www.visualcapitalist.com/european-union-debt-to-gdp-by-country/

Bruegel, “The European Union’s New Fiscal Framework: A Good Start, but Challenges Loom,” 2025. https://www.bruegel.org/policy-brief/european-unions-new-fiscal-framework-good-start-challenges-loom

Bipartisan Policy Center, “U.S. Debt in a Global Context: How Our Fiscal Metrics Stand Out Among Major Economies,” November 2025. https://bipartisanpolicy.org/explainer/u-s-debt-in-a-global-context-how-our-fiscal-metrics-stand-out-among-major-economies/

CaixaBank Research, “Europe’s Medium-Term Fiscal Dilemma,” November 2025. https://www.caixabankresearch.com/en/economics-markets/public-sector/europes-medium-term-fiscal-dilemma

Carnegie Endowment for International Peace, “Using China’s Central Government Balance Sheet to ‘Clean Up’ Local Government Debt Is a Bad Idea,” August 2025. https://carnegieendowment.org/posts/2025/08/using-chinas-central-government-balance-sheet-to-clean-up-local-government-debt-is-a-bad-idea

Atlantic Council, “Beijing Extends and Pretends to Deal with Its Mountain of Local Government Debt,” July 2025. https://www.atlanticcouncil.org/blogs/econographics/sinographs/beijing-extends-and-pretends-to-deal-with-its-mountain-of-local-government-debt/

Bank of Finland / BOFIT, “China’s Public Sector Deficit Set to Rise This Year,” March 2025. https://www.bofit.fi/en/monitoring/weekly/2025/vw202512_2/

Nippon.com, “Expansionary Policy Reflected in Japan’s Record High Budget for Fiscal 2026,” January 17, 2026. https://www.nippon.com/en/japan-data/h02652/

Bloomberg / Japan Times, “Japan Budgeting for Bond Payments at 3% as Debt Servicing Soars,” December 24, 2025. https://www.japantimes.co.jp/business/2025/12/24/economy/interest-payments-on-bonds/

IMF, Staff Concluding Statement of the 2025 Article IV Mission with Japan, February 2025. https://www.imf.org/en/news/articles/2025/02/07/mcs-020725-japan-staff-concluding-statement-of-the-2025-article-iv-mission

U.S. Government Accountability Office, America’s Fiscal Future. https://www.gao.gov/americas-fiscal-future

CRFB, “CBO’s February 2026 Budget and Economic Outlook,” March 19, 2026. https://www.crfb.org/papers/cbos-february-2026-budget-and-economic-outlook

Congressional Budget Office, The Budget and Economic Outlook: 2026 to 2036, February 2026. https://www.cbo.gov/publication/62105

National Bureau of Economic Research, “Projecting Federal Deficits and Debt,” December 2025. https://www.nber.org/digest/202601/projecting-federal-deficits-and-debt

Bipartisan Policy Center, “The Fiscal Outlook in CBO’s Latest 10-Year Baseline,” February 12, 2026. https://bipartisanpolicy.org/explainer/the-fiscal-outlook-in-cbos-latest-10-year-baseline/

Wikipedia, “2026 Iran War.” https://en.wikipedia.org/wiki/2026_Iran_war

Middle East Eye, “Iran Tells Saudi Arabia, Qatar and UAE to Evacuate Energy Facilities After Israel Hits Gas Field,” March 18, 2026. Cites HRANA figures: 3,000+ killed including 1,351+ civilians and 207 children. https://www.middleeasteye.net/news/iran-issues-order-evacuate-petrochemical-facilities-saudi-arabia-qatar-and-uae

Wikipedia, “2026 Kharg Island Raid.” https://en.wikipedia.org/wiki/2026_Kharg_Island_raid

NPR, “Why Trump’s Attacks and Threats to Iran’s Kharg Island Are a Big Deal,” March 19, 2026. https://www.npr.org/2026/03/19/nx-s1-5750514/trump-iran-war-kharg-island-oil

TIME, “What To Know About Kharg Island, the Oil Hub at the Center of the Iran War,” March 14, 2026. https://time.com/article/2026/03/14/kharg-island-trump-oil/

CNBC, “Iran Defends Strikes on Gulf Neighbors — But They Say Trust Is Broken,” March 10, 2026. https://www.cnbc.com/2026/03/10/iran-strikes-gulf-neighbors-us-assets-oil-infrastructure.html

Euronews, “Iran Continues Strikes on Gulf States Day After US Threatens Oil Facilities,” March 15, 2026. https://www.euronews.com/2026/03/15/iran-continues-strikes-on-gulf-states-day-after-us-threatens-oil-facilities

Al Jazeera, “Iran Threatens to Strike Gulf Energy Facilities After South Pars Attack,” March 18, 2026. https://www.aljazeera.com/news/2026/3/18/iran-threatens-to-strike-gulf-energy-facilities-after-south-pars-attack

PBS News, “Iran Intensifies Attacks on Gulf Energy Sites After Israel Struck Its Key Gas Field,” March 19, 2026. https://www.pbs.org/newshour/world/iran-intensifies-attacks-on-gulf-energy-sites-after-israel-struck-its-key-gas-field

CNBC, “Oil Prices: Why Traders Are Getting Nervous About Iran’s $200 Warning,” March 16, 2026. https://www.cnbc.com/2026/03/16/oil-prices-iran-war-200-crude-strait-of-hormuz-supply-shock.html

NBC News, “Top Trump Adviser Says Iran War Price Tag at $12bn So Far,” referenced in live updates, March 13, 2026. https://www.nbcnews.com/world/iran/live-blog/live-updates-iran-war-oil-ship-attacks-hormuz-trump-israel-lebanon-rcna263101

The Original Prediction

In 1965, Gordon Moore, then director of research at Fairchild Semiconductor (a seminal Silicon Valley company that pioneered the silicon transistor and the planar integrated circuit (IC); employees left to start numerous companies, earning it the reputation as the birthplace of Silicon Valley) and later co-founder of Intel, observed that the number of transistors on an integrated circuit was doubling roughly every year. A decade later he revised the estimate to a doubling every two years, and that cadence held with remarkable consistency for half a century.

The prediction was an empirical observation that became a self-fulfilling target: chipmakers organized their research roadmaps around it, investors priced it in, and software developers learned to count on it. The result was an unprecedented engine of economic growth, turning room-sized mainframes into pocket-sized supercomputers in the span of a few decades.

Where Things Stand Today

By the strict definition — a doubling of transistor density every two years at decreasing cost — Moore’s Law is slowing. The leading-edge chips now entering production at TSMC, Intel, and Samsung use 2-nanometer process nodes with gate-all-around nanosheet transistors, replacing the FinFET architecture that has been the industry standard for over a decade. These advances are real and significant, but each step forward requires more extreme engineering and considerably more capital.

A single cutting-edge fabrication facility now costs upward of twenty billion dollars. Below the 5-nanometer threshold, the historical pattern in which smaller transistors also meant cheaper transistors has weakened or reversed entirely. Manufacturing at these scales demands extreme ultraviolet lithography, new transistor geometries, and exotic materials, all of which push costs higher.

The New Playbook: More Than Shrinking

“We can build the architecture, the chip, the system, the libraries, and the algorithms all at the same time,” […] “If you do that, then you can move faster than Moore’s Law because you can innovate across the entire stack.”

- Jensen Huang, Nvidia CEO, claims the company’s GPUs have seen a 1,000x improvement over the last decade

Subscribe now

What has changed most is not whether chips are improving, but how they improve. The semiconductor industry has shifted from a single-minded focus on transistor miniaturization to a multidimensional strategy that pulls several levers at once.

One of the most consequential shifts has been toward chiplet-based design.

Rather than building a single monolithic die, manufacturers now assemble modular silicon blocks (chiplets) each fabricated on the process node best suited to its function, then linked together with high-bandwidth interconnects. AMD’s Infinity Fabric and Intel’s EMIB are prominent examples. This disaggregated approach improves manufacturing yields, lowers costs, and allows heterogeneous integration of compute, memory, and input/output functions on a single package.

Advanced packaging pushes this further into three dimensions. Technologies like hybrid bonding and through-silicon vias enable vertically stacked dies that communicate with low latency, dramatically increasing the effective density and performance of a chip package without requiring smaller transistors. Intel’s Foveros and AMD’s 3D V-Cache are commercial examples already shipping in production hardware. Intel’s Ann Kelleher has argued that advanced packaging gives architects entirely new tools to pursue Moore’s Law, and that Intel’s research pipeline supports continued scaling for the next decade or longer.

Meanwhile, the rise of domain-specific accelerators — GPUs, TPUs, and other chips designed for particular workloads rather than general-purpose computing — means that raw transistor counts are no longer the sole driver of practical performance gains. Software optimization, compiler-aware parallelism, and even AI-assisted chip design are increasingly part of the equation.

Architecture as a Scaling Lever

If chiplet design and advanced packaging represent hardware’s answer to slowing transistor gains, there is a parallel revolution happening in the software and model architecture layer, and it may matter just as much for the functional capabilities that end users actually experience.

The way a system organizes computation, manages memory, and routes information can yield performance gains that rival or exceed what shrinking transistors alone could deliver. In other words, how you use the silicon is becoming as important as how much silicon you have.

Consider what is happening with the core architectures that underpin large AI models. The Transformer, introduced in 2017, has been the dominant paradigm for nearly a decade. Its attention mechanism — the ability for every element in a sequence to “look at” every other element — is extraordinarily powerful, but it comes with a cost that scales quadratically with sequence length. Processing a document twice as long doesn’t cost twice as much; it costs four times as much. This creates a hard ceiling on how much context a model can practically handle, which in turn limits what the system can do.

A wave of architectural alternatives is now challenging that constraint. State space models, most notably the Mamba architecture developed by Albert Gu and Tri Dao, replace the attention mechanism with a recurrent structure that processes sequences in linear time.

The key innovation is selectivity: the model dynamically decides what information to retain and what to discard based on context, giving it strong long-range memory without the computational burden of attention. Benchmarks have shown that even relatively small Mamba-based models can match or exceed the performance of Transformers twice their size on certain tasks.

Google Research’s Titans architecture pushes this further by introducing a deep neural network as a long-term memory module. This gives the model the ability to learn and update what it remembers while it is actively running, rather than relying on a fixed-size compressed state.

One can imagine how a dynamic, continuously adjustable runtime might act as extraordinary leverage for a system needing direct feedback from its context, with the ability to adjust in real time and compare output to expected or even simulated results.

Titans uses a “surprise metric” inspired by human cognition: routine, expected information is allowed to fade, while unexpected or novel input is prioritized for permanent storage.

The companion MIRAS framework provides a theoretical blueprint for unifying these approaches, showing that many seemingly different architectures — Transformers, RNNs, state space models — are all variations on the same underlying problem of associative memory.

In benchmarks requiring reasoning across extremely long documents, Titans has outperformed models with far more parameters, including GPT-4, and has demonstrated the ability to scale to context windows exceeding two million tokens.

A system that is continuously running and updating from environment feedback, even using functional “surprise” as a useful utility, hints that perhaps the field needs to think more carefully about the philosophy of what it is building, and just how analogous to a consciously aware system simulating itself in its environment these models need to become, before the harder questions demand better answers.

Alex O’Connor - Deceiving AI Might Backfire On Us with Nick Bostrom - full video

Nick Bostrom is a philosopher known for his work on existential risk, the anthropic principle, human enhancement ethics, whole brain emulation, and superintelligence risks. His recent book, Deep Utopia, explores what might happen if we get AI development right.

There are at least three reasons this matters. First, it may help researchers better understand and think about how to work with and improve these systems. Second, it may help answer questions about how human beings themselves process information, individually and collaboratively at scale. And third: the field appears to be intuitively developing these systems as functional replicas of conscious experience.

The biggest capability gains so far — agents (simulation of “ego/self”), chain-of-thought reasoning, memory, reinforcement learning from human feedback, scope and organisation of context and real-time feedback — have all come from instantiating patterns borrowed from our own experience of human cognition. The current frontier is better continuous functionality, memory, and attention-setting techniques, which deepen the parallel.

It may be that the distinction between virtualizing a thinking system and an “authentic” thinking system comes down to a difference in substrate rather than of information processing pattern; a distinction that might even prove to be meaningless semantics…

However one decides on the semantics, the important consideration and most fruitful way to conceptualise how these systems “are,” may well be leaning into the obvious analogous and functional descriptions of what is fundamentally shared in human and AI “cognition”: information processing systems, operating in trained, thinking patterns, according to context; limiting our definitions or categories to hosting architecture (substrate) could become an unnecessary self-imposed conceptual limitation in practice.

What further implications these points raise (in scaling and beyond) is an open question, and one that may need to be addressed sooner rather than later. With the scale of the economic bets, and with what will surely be a gradual but deeply embedded integration of AI into all of our systems, perhaps we should be careful not to find ourselves with a desperate need to try and put the toothpaste back in the tube.

(Potentially important) philosophy aside, the practical implications are already arriving. Hybrid architectures that combine Transformer attention with state space model efficiency and Mixture-of-Experts routing are entering production. AI21 Labs’ Jamba model interleaves Transformer and Mamba layers with Mixture-of-Experts blocks, achieving a 256,000-token context window while activating only a fraction of its total parameters at any given time.

IBM’s Granite 4.0 family uses a similar Mamba-Transformer hybrid to run on significantly cheaper hardware. These are fundamental rethinks of how to extract useful work from a given amount of compute and they are delivering capability gains that look nothing like the gradual march of transistor scaling.

Share

The Efficiency Approach Opens Doors

In January 2025, Chinese AI lab DeepSeek demonstrated that frontier-level AI capabilities do not require frontier-level budgets. The achievement sent shockwaves through the industry, not because the model was smarter than its Western counterparts, but because of how it got there.

The core innovation is architectural. Where most leading models use dense networks in which every parameter activates for every word processed, DeepSeek splits its massive 671-billion-parameter model into hundreds of tiny, specialised expert modules. For any given task, only around 37 billion parameters actually fire; roughly 5% of the total. The model routes each input to the handful of experts best suited to handle it and leaves the rest dormant. The result is a system with the knowledge capacity of a giant model and the running cost of a much smaller one.

They also engineered a method called Multi-Head Latent Attention that compresses the memory required to track conversational context into a fraction of its usual size, slashing inference memory requirements by over 90%. In practical terms, the model can process enormous amounts of text without requiring the kind of hardware that typically costs hundreds of millions of dollars.

But the architectural choices were only half the story. DeepSeek’s team wrote custom, low-level code to extract absolute maximum performance from their GPU clusters, rather than relying on standard high-level frameworks. They were the first to successfully train a major open-weight model using 8-bit floating point precision at scale, effectively doubling compute speed and halving memory requirements without degrading the model’s intelligence. Through custom communication algorithms, they virtually eliminated the idle time that typically burns through millions of dollars in large server farms: the gaps where processors sit waiting for data to arrive from other processors.

Perhaps most significantly, DeepSeek rethought how reasoning itself is trained. Standard methods require a massive secondary model just to evaluate the primary model’s outputs, effectively doubling the compute and memory cost of the training process. DeepSeek’s approach, called Group Relative Policy Optimization, eliminates this entirely. Instead of asking a separate system “was that answer good?”, the model generates multiple answers to the same problem, scores them against each other using verifiable rules, and learns from the comparison. The model teaches itself to reason by competing against its own outputs.

The DeepSeek paper describes what emerged from this process as genuinely novel: through pure reinforcement learning, without being shown human-written reasoning examples, the model developed self-reflection, verification of its own intermediate steps, and the ability to dynamically switch strategies mid-computation when it sensed a dead end. These behaviours were not programmed. They emerged from the structure of the training process itself.

Combined with multi-token prediction — forcing the model to anticipate several words ahead simultaneously rather than one at a time — DeepSeek demonstrated that rigorous engineering and algorithmic ingenuity can outpace brute-force financial spending in the AI race. The implications extend beyond cost savings: if a model can generate its own reasoning paths and improve through self-competition on verifiable tasks, the dependency on expensive, finite human-labelled training data begins to loosen. The bottleneck shifts from “how much data can we collect” to “how well can the system learn from itself.”

If the only path to smarter AI were physically larger models on physically larger hardware, the power and thermal constraints ahead would eventually halt progress entirely. DeepSeek proved there is another path: make the training smarter, the architecture leaner, and the engineering more precise. The hardware still matters. But how you use it may matter more.

Beyond Chips: A Broader Acceleration

Step outside the semiconductor fab and the picture shifts from “slowing” to something more like “accelerating wildly.” The rate of improvement in AI capabilities, in particular, has outpaced Moore’s Law by orders of magnitude.

Compute requirements for training large language models have grown by roughly 750 times every two years, and model parameter counts have expanded by around 410 times over the same interval. These are staggering numbers, and far beyond what transistor scaling alone could deliver. They are enabled by a combination of hardware scaling, architectural innovation, algorithmic efficiency gains, and massive capital investment. Major technology companies collectively planned to spend over 300 billion dollars on AI data centers in 2025 alone.

Researchers at METR recently documented another striking trend: the length and complexity of tasks that AI agents can reliably complete has been growing exponentially.

Over the period from 2019 to 2025, the “time horizon” of tasks agents could handle at a fifty-percent success rate doubled roughly every seven months. In 2024 and 2025, that rate accelerated to a doubling every four months. If these trends continue, the practical capabilities of AI systems will see “AI agents that can independently complete a large fraction of software tasks that currently take humans days or weeks.”

The New Physical Bottlenecks

If the pace of capability improvement is faster than ever, the constraints have shifted. Power consumption and thermal management have emerged as first-order challenges. Next-generation AI data centers will require power in the gigawatt range, and projections suggest that total U.S. data center power usage could rise to between seven and twelve percent of the national total by 2028.

Traditional air cooling is giving way to liquid cooling. Electricity prices for consumers could rise as a result. Moving data between processors and memory fast enough has become as significant a bottleneck as computation itself: the energy required to move a bit of data now exceeds the energy required to operate on it. These are thermal, electrical, architectural problems, not foundational issues with transistors. How we solve these problems will define the next chapter of computing progress, as much as any advance in lithography.

Water is becoming another critical constraint. Data centers consume enormous quantities of water for cooling, and as these facilities scale into the gigawatt range, their demand will compete with agricultural, industrial, and residential needs.

Subscribe now

Water is quietly becoming both a national security interest and a corporate resource-planning concern. The history of slow government response to water crises like Flint, Michigan, raises uncomfortable questions about what happens when the infrastructure demands of the technology economy collide with the basic resource needs of the population.

There is also the ambitious concept of space data centres — the plans to put data centres in orbit and on the Moon [BBC News] (something that Musk in particular seems keen on), initially conceived as a solution to the cooling, water, and energy demands that plague their terrestrial counterparts.

While escaping the heat limits of Earth is an attractive idea, the reality of operating in a vacuum creates an even more daunting engineering challenge: heat rejection in space is incredibly difficult and expensive. This technical hurdle is further complicated by the increasingly crowded environment of Low Earth Orbit, which faces a potentially catastrophic failure mode known as the Kessler syndrome. This self-sustaining chain reaction of satellite collisions highlights the significant risk that orbital architecture could become fundamentally unstable, which is another way of saying it might not be the most reliable, practical, maintainable, or scalable solution to humanity's growing computing needs.

Do We Need a New Law?

Moore’s Law measured one thing: transistor density on a chip, doubling every two years. It was a clean, elegant metric for a simpler era of computing progress. But the landscape described in this piece — chiplets, 3D packaging, hybrid model architectures, AI-designed chips, algorithmic breakthroughs — is one where capability gains arrive through many channels simultaneously, and where the rate of improvement varies wildly depending on what you choose to measure.

Several candidates for a successor framework have emerged. METR’s finding — that the complexity of tasks AI agents can complete doubles roughly every seven months, accelerating to every four months recently — has been widely called “the new Moore’s Law for AI.” It measures something fundamentally different from transistor counts: functional capability, the practical utility of a system as experienced by the people who use it.

Others have pointed to AI scaling laws, first formalized around 2020, which describe how model performance improves as a predictable power-law function of compute, data, and parameter count. These scaling laws operate on a faster clock than Moore’s Law ever did, with effective compute for AI roughly doubling every six months.

But perhaps the most important insight is that no single metric captures the full picture anymore. The functional productivity of chip-based technology — the actual useful work a system can perform per dollar, per watt, per unit of time — is now a composite of transistor density, architectural design, software efficiency, memory bandwidth, packaging innovation, and the intelligence of the models running on top.

Each of these layers is improving on its own exponential curve, and they compound on one another.

A new architectural pattern like Titans or Jamba does not require a single additional transistor to deliver a leap in what a system can accomplish. A Mixture-of-Experts model can activate a fraction of its parameters and still outperform a dense model that uses all of them.

Moore’s Law, in its original narrow formulation, is bending under the weight of physics and economics. Transistors are still getting smaller and more numerous, but the pace has slowed, and the cost of each incremental advance is rising.

Yet the spirit of the law continues, as is the expectation of relentless, exponential improvement in what computer systems can do. The improvements are simply arriving through a broader set of channels: architectural innovation, advanced packaging, software co-design, domain-specific hardware, and, increasingly, AI improving AI.

Whether this still counts as “Moore’s Law” is partly a semantic question. What matters more is that the rate of technological capability improvement showing no sign of slowing down. If anything, it is speeding up.

[x.com/josephdviviano]

CHRIS

I can’t move.

MISSY

You can’t move. That’s good.

Now sink into the floor.

CHRIS

Wait I--

MISSY

Sink.

In 1985, a median American home cost about $83,000 and the median household earned about $24,000 a year. That’s a ratio of roughly 3.5 to 1. A home cost three and a half years of your life. Uncomfortable, but reachable. By 2025, the median home costs $417,000 and the median household earns $83,000. The ratio has climbed to 5 to 1 — and in Los Angeles it’s 12.5, in San Jose it’s 10.5, in New York it’s nearly 10. Home prices have risen 415% over forty years. Incomes rose 255%. The gap is accelerating.

The S&P 500 returned over 800% in its lifetime. What did that deliver to the person in the middle of the income distribution? Not 800% of anything. Not 400%. Not even close. The median worker’s real compensation rose roughly 16% over forty years while productivity climbed 60%. Where did the other 44% go? It went up. It went into corporate profits, executive compensation, shareholder returns, and the asset prices that only matter to people who already own assets.

The story of how we got here spans roughly seventy years. It starts with a post-war miracle that was deliberately dismantled through tax policy, deregulation, union-breaking, and the financialisation of everything. It accelerates through a crisis consolidation playbook that turns every downturn into a buying opportunity for those with liquidity and a wipeout for those without. And it reaches its current velocity through artificial intelligence, which arrived at the precise moment when all of these dynamics were already at extremes, compressing the next phase of concentration into years rather than decades, creating a cognitive divide between those who own the intelligence infrastructure and those who are subject to it.

If you have read the first two parts of this series, you know the mechanisms. If you haven’t, the short version is: the system was redesigned on purpose, over decades, by identifiable people, through identifiable policy choices, and the results are exactly what you would expect if you understood the design.

This piece is about where the money ended up.

Not as a chart or a Gini coefficient but as a map of human lives, tier by tier, from the 40 million Americans below the poverty line to the roughly 800 billionaires who now collectively hold more wealth than the bottom 95% of humanity combined.

Subscribe now

Seven Tiers

There are approximately 330 million people in the United States. They do not live in one economy. They live in several, stacked on top of each other, increasingly disconnected, speaking different financial languages, eating different food, sending their children to different schools, and dying at different ages. The numbers that follow are approximations. The experiences they describe are not.

The table below summarises the structure. What follows describes what it feels like to live inside it.

Sources: US Census Bureau (poverty data, 2024); Federal Reserve Distributional Financial Accounts (Q3 2025); Oxfam “Resisting the Rule of the Rich” (January 2026); Federal Reserve Survey of Consumer Finances; Bureau of Labor Statistics; author’s analysis based on Cycles of Capital framework. Population estimates are approximate and tiers are illustrative rather than precisely bounded. Wealth share figures reflect Fed DFA data for top 1% (31.7%) and top 10% (~65%), with lower-tier breakdowns estimated from SCF distributions.

Tier 1 — The Excluded (~36 million people)

Income below $15,000 a year. Net worth negative or functionally zero. This tier includes people who are unhoused, food-insecure, dependent on welfare systems that are being algorithmically enforced and systematically defunded. The official poverty rate in the United States was 10.6% in 2024, covering 35.9 million people, though the Supplemental Poverty Measure, which accounts for actual living costs, puts the figure closer to 12.9%. A $400 emergency, the kind that most Americans will face at least once a year, is not a setback at this level. It is a catastrophe. It means choosing between the electricity bill and the medication. It means a payday loan at 400% APR that takes months to escape, if you ever do.

Children grow up in this tier. The child poverty rate in the United States was 14.3% in 2024. These children attend schools funded by local property taxes, which means the poorer the neighbourhood, the worse the school. They eat meals subsidised by SNAP, which have been converted to block grants with algorithmic work requirements designed to purge recipients from the rolls. They live in housing that is precarious, loud, and often unsafe. The phrase “upward mobility” is a concept from another country.

This tier holds approximately zero percent of the nation’s wealth.

Tier 2 — The Working Poor (~55-65 million people)

Income between $15,000 and $30,000 a year. Assets: a car, maybe, almost certainly depreciating. Savings: enough for a few weeks, if that. These are the people stocking shelves at 5am, cleaning offices at midnight, driving deliveries in their own vehicles, picking crops, building houses they will never live in. Many are immigrants doing the work that the economy depends on and the political system scapegoats them for.

Debt at this level does not build anything. It extracts. Credit cards at 22-24% interest. Medical bills from a system that charges $800 for an ambulance ride. Auto loans on vehicles they need to get to the jobs that don’t pay enough to service the auto loans. The average credit card balance for households earning under $35,000 has been rising steadily, because when wages don’t cover essentials, plastic fills the gap. That gap charges interest.

Homeownership is out of reach. Rent consumes 40-50% or more of income. A job loss, an illness, a rent increase: any one of these can push a Tier 2 household into Tier 1 within weeks. The floor is right there. Everyone can feel it.

This tier holds less than 1% of the nation’s wealth.

Tier 3 — The Squeezed Middle (~55-65 million people)

Income between $30,000 and $60,000. A retirement account with a few thousand in it. Maybe some equity in a home, if they bought before 2020 and haven’t been forced to refinance. This was supposed to be the American Dream. It was supposed to be the tier where effort started to pay off, where a degree led to a career, where you could put something aside for your kids.

Student debt sits at $1.77 trillion nationally. Much of it is concentrated here. The degree was supposed to be the ladder. The ladder is now the debt. Delinquency rates on student loans hit 9.6% in Q4 2025, with approximately one million borrowers transferred to the Default Resolution Group. For many households in this tier, the resumption of student loan payments was the thing that snapped an already strained budget.

Productivity in the American economy has risen roughly 60% since 1979. Hourly compensation for workers in this tier rose about 16% over the same period. The gap between those two numbers represents the largest transfer of value from labour to capital in modern history. Tax cuts, deregulation, union-breaking, financialisation. Part 1 documented all of it.

People here are one bad year from Tier 2. They know it. That knowledge shapes everything: risk aversion, political anxiety, susceptibility to narratives that offer someone to blame. When a politician points at immigrants and says “that’s why you’re struggling,” this is the tier where it lands hardest, because the struggle is real even if the explanation is a lie.

This tier holds roughly 2-5% of the nation’s wealth.

Tier 4 — The Overleveraged Comfortable (~45-50 million people)

Income between $60,000 and $150,000. Homeownership, usually. A 401(k) that looks reassuring on paper. Decent health insurance through an employer. Occasional holidays. Private school if they stretch. The outward appearance of prosperity.

But the prosperity is leveraged. The home is the single largest asset, and a market correction of 15-20% wipes out years of equity. The retirement account is invested in index funds that are 40% concentrated in ten companies, all betting on the same AI thesis. The health insurance is tied to an employer who is “restructuring.” The children’s college fund is a fraction of what tuition will cost.

This is the tier that the economy measures. When economists say “the consumer is resilient,” they are often talking about this group and the one above it. Their spending keeps retail afloat. Their mortgages keep banks solvent. Their 401(k) contributions keep the S&P 500 inflated. They are the structural load-bearing wall of the entire economic narrative.

They are also the tier most exposed to the AI displacement documented in Part 2 of this series. Entry-level white-collar jobs, the stepping stones into this tier, are being eliminated. Junior analysts, associate lawyers, marketing coordinators, copywriters. Stanford’s research found a 13% employment drop among 22-25 year olds in AI-exposed roles since late 2022. The rungs at the bottom of this ladder are disappearing while people are still climbing.

This tier holds roughly 10-15% of the nation’s wealth.

Tier 5 — The Threshold (~20-30 million people)

Income between $150,000 and $500,000. Diversified investments. Business equity. Perhaps a second property. Access to a financial advisor rather than a banking app.

Here is where the qualitative shift begins. Below this tier, money is functional. You use it to live. It flows in and flows out. You trade time for it. Above this tier, money starts to work while you sleep. It compounds. It generates more of itself through mechanisms that have nothing to do with effort.

This is the door to the leverage loop described in Part 1. If you own $500,000 in appreciating assets, a bank will lend against them at rates that a Tier 2 household could never access. The loan is not income nor is it taxed. Your assets continue to appreciate. You are now operating within a different financial physics. Capital gains are taxed at 20%. Ordinary income, the kind that comes from working, is taxed at 37%. The tax code is telling you, in plain arithmetic, that owning is worth nearly twice as much as doing.

This tier is pulling away from the tiers below it. The gap between Tier 4 and Tier 5 is widening faster than the gap between Tier 3 and Tier 4, because the mechanisms of accumulation are multiplicative at this level. The same passive index fund concentration that squeezes the Russell 2000 (documented in The Number Is Going Up) enriches Tier 5 portfolios through the same mathematical inevitability.

This tier holds roughly 20-25% of the nation’s wealth.

Tier 6 — The Untouchable (~3-4 million people)

Net worth between $5 million and $30 million. Income primarily passive: capital gains, dividends, rental income, business distributions. This tier has left the conventional economy entirely. They don’t check their bank balance before buying groceries. They don’t think about whether they can afford the holiday. They don’t know what a payday loan is, except as an abstraction in a policy document they might skim.

Family offices manage their wealth. Dynasty trusts ensure it passes to the next generation without the inconvenience of estate taxes. Private banks offer products that don’t exist below this level: bespoke lending facilities, alternative asset access, tax-optimised debt structures that let them live on borrowed money while their assets appreciate untouched.

This tier is functionally beyond the reach of economic fluctuation. A recession is not a threat at this level. It is a buying opportunity. The crash described in Part 1 (Blackstone purchasing 50,000 foreclosed homes) was executed by capital from this tier and the one above it.

This tier holds roughly 20-25% of the nation’s wealth. About 1% of the population.

Tier 7 — The System (~250,000-300,000 people, and within them, ~800 billionaires)

Net worth $30 million to $100 billion and beyond. These individuals do not participate in the economy. They are the economy… Their capital allocates markets, shapes political discourse, funds the think tanks that write the policy, owns the platforms that distribute the information, and employs the lobbyists who ensure the rules remain favourable.

JIM

Ignorance...

CHRIS

Who?

JIM

All of them.

They mean well but they have no idea what real people go through.

CHRIS

I guess people only see what’s in front of them.

The top 1% of American households now hold 31.7% of all wealth, the highest share since the Federal Reserve began tracking in 1989. The top 10% hold approximately 65%. The bottom 50% hold about 2.5%.

Globally, the picture is starker. Oxfam’s January 2026 report found that billionaire wealth reached a record $18.3 trillion, growing 16% in a single year, three times faster than the five-year average. Since 2020, billionaire wealth has increased 81%. The twelve richest billionaires now own more than the poorest half of humanity. Four billion people. The $2.5 trillion added to billionaire fortunes last year would be enough to eradicate extreme poverty 26 times over.

And here is the number that should stop you: billionaires are 4,000 times more likely to hold political office than ordinary citizens. They own more than half the world’s largest media companies and all the main social media platforms.

JIM

Okay, then let me just tell you what it is.

Phase one was the hypnotism. That’s how they sedate you.

Phase two is this. Mental preparation.

It’s basically a psychological pre-op.

In the United States, the Trump cabinet is the wealthiest in American history, with a combined net worth measured in hundreds of billions. The people at the top of the economic architecture are now running the government that is supposed to regulate them.

Do you know the difference between a million dollars and a billion dollars? It is approximately a billion dollars. A million seconds is twelve days. A billion seconds is thirty-two years. The human brain cannot intuitively grasp the difference. Which is part of how it persists.

Would a million dollars drastically change most of our lives? Yes. Would you be able to spend a billion dollars if it were your full-time job? Probably not. And yet the system treats the accumulation of the latter as not only normal but aspirational, and protects it with more legal infrastructure than most countries devote to public health.

The new dimension is intelligence itself. AI is trained on public knowledge and refined through interaction with the public, but the most capable versions are gated by cost and exclusivity. Capital buys intelligence. Intelligence generates capital through superior systems across business, finance, legal strategy, and logistics. Capital reinvests in intelligence development, including customisation that is unavailable at the consumer tier. The cycle repeats, each iteration widening the gap.

This is the current operating model of the AI economy, and it is the leverage loop from the 1980s rewritten for the 2020s, running at the speed of inference rather than the speed of legislation.

The Break

There is a qualitative discontinuity between Tier 5 and Tier 6 that the numbers alone cannot convey. Below that line, wealth is fragile. It depends on wages, on employment, on health, on the housing market, on interest rates, on whether the company restructures. It can be wiped out by a divorce, a diagnosis, a downturn. People below that line play by one set of financial rules. The rules are punitive. Default means foreclosure. Missed rent means eviction. An illness means bankruptcy. There is no asset reallocation strategy for people who own no assets.

Above that line, wealth becomes self-sustaining. It reproduces through structure: compounding returns, institutional scaffolding, legal architectures designed to prevent erosion, and a tax code that rewards ownership almost twice as generously as labour. Losses at this level are absorbed, diversified, negotiated. Crises are opportunities. The people above the line don’t need the economy to be healthy. They just need it to keep producing assets they can buy.

And here is the part that should worry everyone: this is no longer just an economic divide. When the tools of thought, analysis, prediction, and decision-making are themselves gated by wealth, the divide stops being purely financial. It becomes epistemic. It becomes about who gets to understand the world accurately, and who gets a filtered, managed, nudged version of reality shaped by someone else’s interests. The same AI that serves as a concierge for Tier 7 serves as a warden for Tier 2. Same technology. Opposite function.

Two sets of rules. One word for both: economy.

The tier system described above accounts for the wealth that is visible. The wealth that shows up in tax returns, census data, Federal Reserve surveys, and Forbes lists. There is another category of wealth that does not appear in any of these. It sits in jurisdictions designed to make it invisible: the British Virgin Islands, the Cayman Islands, Panama, Luxembourg, Switzerland, Jersey, Bermuda, and dozens of others. It is held through shell companies, trusts, foundations, and layered corporate structures whose purpose is to separate the money from the name of the person who owns it.

The Tax Justice Network has estimated that between $21 trillion and $32 trillion in private financial assets are held offshore. To put that in context: total US GDP in 2025 was roughly $29 trillion. The offshore system holds wealth equivalent to the entire annual economic output of the world’s largest economy, and possibly more. Nobody knows the exact figure, because the entire point of the architecture is to make precise measurement impossible.

Three Leaks, One Architecture

Between 2016 and 2021, three massive data breaches pulled back the curtain on this system. Collectively, they represent the largest exposure of hidden wealth in history. They revealed a lot about the behind-the-scenes financial behaviour of the global elite.

The Panama Papers, published in April 2016, consisted of 11.5 million documents from the Panamanian law firm Mossack Fonseca. The files detailed the creation and management of over 214,000 offshore entities linked to people in more than 200 countries. They exposed heads of state, close associates of Vladimir Putin, the family of Pakistan’s prime minister, and hundreds of other politicians, athletes, and business figures using secretive structures to hide assets, avoid taxes, launder money, and evade sanctions. Iceland’s prime minister resigned within days. Mossack Fonseca shut down in 2017. In 2024, a Panamanian court acquitted all 28 defendants on money laundering charges, citing insufficient evidence and chain-of-custody issues with the leaked data.

The Paradise Papers followed in November 2017: 13.4 million files, primarily from the Bermuda-based law firm Appleby. These documents revealed how Queen Elizabeth II’s private estate had invested millions through offshore funds, how Trump’s Commerce Secretary Wilbur Ross held stakes in a shipping company linked to Russian oligarchs despite sanctions concerns, and how Apple had engineered massive offshore profit shifts to Jersey to avoid billions in taxes. The EU later ruled that Apple owed Ireland over $14 billion in back taxes.

The Pandora Papers, published in October 2021, were the largest of the three: 11.9 million documents from 14 different offshore service providers. They linked 35 current and former world leaders, over 330 politicians from nearly 100 countries, and more than 130 billionaires to offshore structures. King Abdullah II of Jordan had secretly acquired a $100 million property empire through dozens of shell companies. The then-Czech prime minister had bought a $22 million French Riviera chateau through offshore firms while publicly campaigning against elite corruption.

Taken together, these three leaks exposed over 36 million documents and hundreds of thousands of offshore entities. They revealed a financial architecture used systematically by the wealthiest people on earth to move money beyond the reach of democratic accountability.

Table 2: Three Leaks, One Architecture

Sources: ICIJ investigations (2016, 2017, 2021); individual case reporting via The Guardian, BBC, Washington Post, Süddeutsche Zeitung.

What Changed

Almost nothing.

Nine years after the Panama Papers were published, governments worldwide have recovered approximately $1.86 billion in back taxes and penalties from the revelations. That sounds significant until you measure it against the scale of the problem. The Tax Justice Network’s 2024 State of Tax Justice report found that countries lose approximately $492 billion per year to global tax abuse: $348 billion from multinational corporations shifting profits offshore, and $145 billion from wealthy individuals hiding wealth. The 2025 report found that over a six-year period, countries lost a cumulative $1.7 trillion to corporate tax abuse alone, with US-headquartered multinationals responsible for 29% of the total.

Share

$1.86 billion recovered. $492 billion lost every year. The recovery represents roughly 0.4% of one year’s losses. At this rate, it would take over 250 years to recover what is lost in a single year. And the losses compound.

Table 3: The Offshore Recovery Gap

The $21-32 trillion estimate covers financial assets only. Including international real estate, private art, precious metals, private equity, cryptocurrency, and other non-financial assets, the true total is substantially higher. The Tax Justice Network describes this as a floor, not a ceiling.

These tax havens did not emerge by accident. As the historian Vanessa Ogle has documented, and the Oxfam briefing paper on global oligarchy makes explicit, many of the world’s secrecy jurisdictions were originally constructed by Northern bankers, lawyers, and consultants to allow British and French settlers in the Caribbean and across Africa to shield their wealth from decolonisation. Offshore finance is, in its origins, a colonial instrument repurposed for the global age. The same jurisdictions that once protected colonial extraction now protect billionaire extraction. The mechanism is continuous. Only the clients have changed.

The journalist who did the most to expose the offshore architecture in Malta, Daphne Caruana Galizia, was assassinated on October 16, 2017. A car bomb, remotely detonated. She was 53. She had used the Panama Papers to expose corruption ties between Malta’s prime minister’s inner circle and offshore shell companies. A public inquiry concluded that the Maltese state bore responsibility for creating a climate of impunity that enabled the murder. Her killers have been convicted. The alleged mastermind, a wealthy businessman with ties to the political elite, was controversially granted bail in 2025. No one has been definitively convicted as the ultimate order-giver.

She was killed for reporting on the architecture. The architecture however, still remains.

How It Works

The offshore system operates through a set of mechanisms that are individually legal and collectively devastating. Understanding them requires stripping away the complexity that has been deliberately constructed to discourage scrutiny.

Shell companies are legal entities that exist on paper but conduct no real business. They are created in jurisdictions that do not require the disclosure of the actual owner. A person in London or New York can set up a company in the British Virgin Islands, register it through a nominee director in Panama, open a bank account in Switzerland, and hold assets that are functionally invisible to the tax authorities of any country where the actual owner lives or earns money. The company exists. The assets exist. The connection between them and the person who benefits from them is obscured behind multiple layers of legal fiction.

The opacity extends beyond tax avoidance into the networks of power themselves. Recently released Epstein files and investigative reporting by DER STANDARD revealed that Jeffrey Epstein invested $40 million in venture capital funds co-founded by Peter Thiel ’s (Valar Ventures), including more than $2.5 million that financed the German online bank N26.

CHRIS

It’s so weird, man. It’s like some of these people never met a black dude that doesn’t work for them.

Peter Thiel has publicly stated that he no longer believes freedom and democracy are compatible, and his political network (which he has poured funding into for over a decade) includes Vance, the Vice President of the United States. The financial systems that hide wealth also obscure the relationships between the people who own them. When those relationships include figures at the intersection of intelligence networks, political power, and technology development, the opacity is the point, not a side effect.

Jurisdictional arbitrage exploits the differences between countries’ tax codes and transparency requirements. A corporation earns revenue in the United States, routes it through an intellectual property holding company in Ireland (where the effective rate is lower), then shifts it again to a subsidiary in Luxembourg or the Netherlands, and parks the residual in a Bermuda entity where it faces no taxation at all. Each step is technically legal. The cumulative effect is that profits earned from American consumers, using American infrastructure, employing American workers, are taxed at a fraction of the statutory rate or not taxed at all. Apple’s profit-shifting arrangements through Ireland, exposed in the Paradise Papers, resulted in an EU ruling that the company owed Ireland $14.8 billion in back taxes. Apple appealed. The case took years to resolve.

The revolving door ensures that the people writing tax law and the people exploiting it often share employers, conferences, and social networks. Tax lawyers move between government posts and private practice. Accountants at the Big Four firms advise governments on policy while simultaneously advising their wealthiest clients on how to avoid the policies they helped design. The OECD, which has led international tax reform efforts, has faced concerns about corporate influence over its processes and a revolving door with the private sector. An academic study of submissions to the OECD during the drafting of tax standards concluded that business groups had more influence on the final text than civil society organisations.

The Best Investment Return in History

Oxfam’s analysis of the 50 largest US public corporations found that for every $1 they spent on lobbying between 2008 and 2014, they received $130 in tax breaks and more than $4,000 in federal loans, loan guarantees, and bailouts.

A dollar spent on lobbying returned over four thousand dollars in government-backed financial support... it doesn’t sound like capitalism or a free market. This is a system in which public resources are converted into private wealth through the legal purchase of political influence, and the return on that purchase exceeds anything available in any asset class, in any market, anywhere in the world.

In 2022, the 182 largest US public corporations spent a collective $746 million on lobbying. The investment is rational. The return is extraordinary. The cost is borne by everyone who pays taxes, uses public services, or lives in a society that underfunds its schools, bridges, and hospitals so that the wealthiest corporations can extract a few more points of margin.

This is the mechanism. The offshore architecture hides the money. The lobbying system protects the rules that allow the hiding. And the tax code ensures that the people who benefit most from the system contribute least to maintaining it. Capital gains taxed at 20%. Ordinary income taxed at 37%. Unrealised gains taxed at 0%, indefinitely, until death erases them through the stepped-up basis. For every dollar that crosses the line into Tier 6 and above, the system provides an escalating series of protections designed to ensure it never comes back down.

Fifty Years of Tax Cuts

The argument for this arrangement has been remarkably consistent for half a century: lower taxes on the wealthy will generate economic growth that benefits everyone. The money will trickle down. The corporations will invest. The jobs will appear.

Companies reinvested profits into the business, into workers, into research and development. The tax code shaped a culture of reinvestment. Reagan’s cuts reshaped a culture of extraction. Every subsequent administration, with minor exceptions, continued the trajectory. The template set in 1981 has never been fundamentally reversed.

An academic review of 50 years of tax cuts for the rich across 18 countries, published in the Socio-Economic Review in 2022, found that these promises never materialised. The only measurable impact was increased inequality.

Fifty years. Eighteen countries. The same experiment, repeated. The same result, ignored. The policy continues because the people who benefit from it are the same people who fund the campaigns, staff the advisory boards, own the media organisations, and now sit in the cabinet. The empirical evidence is overwhelming and irrelevant, because the policy was never about growth. It was about preservation.

The race to the bottom has been comprehensive. Corporate tax rates have fallen worldwide since 1980, with benefits flowing primarily to wealthy shareholders. Inheritance taxes have fallen across rich countries, and nearly half of the world’s billionaires live in countries with no inheritance tax on wealth passed to direct descendants. According to Gabriel Zucman’s research for the G20, many ultrawealthy individuals now face effective tax rates approaching zero. Meanwhile, the ultrawealthy can pass on what Oxfam has called a “$5 trillion tax-free treasure chest” to the next generation, more than the GDP of Africa. In 2023, for the first time, more new billionaires got rich through inheritance than through entrepreneurship. The next three decades will see over 1,000 of today’s billionaires transfer more than $5.2 trillion to their heirs.

The Offshore System as Class Architecture

When Part 1 described how the top 1%’s share of national income fell from 24% in 1928 to 9% by the late 1970s, and then climbed back, the offshore system was the reservoir into which the restored wealth was channelled. When the tax code was redesigned to favour capital over labour, the offshore system ensured that even the reduced taxes on capital could be further minimised or eliminated. When crises created buying opportunities for the wealthy, the offshore system provided the liquidity and the opacity to execute at scale.

The Tax Justice Network’s estimate of $21 to $32 trillion sitting offshore in private financial assets does not include real estate ($5 to $10 trillion internationally), private art collections, precious metals, private equity holdings, or cryptocurrency. It is a floor, not a ceiling. A 2012 analysis estimated that just 92,000 people, roughly 0.0001% of the world’s population, own approximately $9.8 trillion of the offshore total. The system serves an almost invisibly small number of individuals, at a cost borne by everyone else.

The self-made myth persists because it serves the interests of those who benefit from the architecture. But the data tells a different story: most extreme wealth is not created. It is extracted, inherited, or accumulated through market dominance that would, in an earlier era, have been broken up as anti-competitive.

As Gabriel Zucman investigated for the G20, the wealth of the richest 0.0001% of households, roughly 3,000 families worldwide, went from 3% of world GDP in 1987 to 13% by 2024. Their collective wealth now stands at $14 trillion. A four-fold increase in 35 years, during a period in which the bottom half of humanity saw negligible gains.

The Oxfam report showed that billionaires are 4,000 times more likely to hold political office than ordinary citizens. They own more than half the world’s largest media companies and all the main social media platforms. They fund the think tanks that produce the policy papers. They sit on the boards that appoint the regulators. And when they want the rules changed, they spend $746 million a year making it happen, because the return is $130 for every dollar, plus $4,000 in government-backed support.

The invisible architecture is the foundation on which the visible one rests. And it is the reason the money described in the tier system does not flow back down. It was never supposed to. The system is working exactly as designed.

Next we will look at the specific institutional mechanisms, from private banks to dynasty trusts, that keep the money at the top once it arrives there.

Sources: ICIJ Panama Papers investigation (2016) and nine-year recovery tally ($1.86B as of April 2025); Tax Justice Network “State of Tax Justice 2024” ($492B annual losses, $21-32T offshore estimate), “State of Tax Justice 2025” ($1.7T cumulative corporate losses), and “The Price of Offshore Revisited” (2012, 92,000 individuals / $9.8T); Oxfam “Multilateralism in an Era of Global Oligarchy” (September 2024, lobbying ROI, colonial origins of tax havens, corporate tax trends); Oxfam “Resisting the Rule of the Rich” (January 2026); Gabriel Zucman, “A Blueprint for a Coordinated Minimum Effective Taxation Standard” (G20, 2024); Hope & Limberg, “The Economic Consequences of Major Tax Cuts for the Rich,” Socio-Economic Review (2022); Vanessa Ogle, “Funk Money: The End of Empires, The Expansion of Tax Havens” (2020); Daphne Caruana Galizia assassination reporting via ICIJ, The Guardian, and Maltese public inquiry (2021).

So far, this piece has mapped two things: where people sit in the wealth distribution, and where the money goes once it crosses into the upper tiers. Here we look at why it stays there.

The persistence of extreme wealth is not a natural outcome of talent or effort, rather it is maintained by a set of institutional mechanisms that function as a scaffolding around accumulated capital, preventing it from dispersing, eroding, or being subject to the same forces that govern wealth at lower tiers. These mechanisms include private banking, legal architecture, corporate concentration, passive investment structures, and a tax code that treats ownership as more valuable than labour. Together, they ensure that wealth above a certain threshold becomes self-sustaining, and that the people who hold it are functionally insulated from the economic forces that shape life for everyone else.

The institutions that protect concentrated capital select loyalty over competence, increasingly have priorities that are disconnected from the health of the economy it claims to be leading.

The Meritocracy Problem

The story is meritocracy: the wealthy are wealthy because they are talented, disciplined, and innovative. They created value. They earned it. The system rewarded them fairly.

The data does not support this.

Oxfam’s 2025 “Takers Not Makers” report found that approximately 60% of global billionaire wealth is unearned. The breakdown: 36% derives from inheritance, 18% from monopoly power, and 6% from crony connections or corruption. Every billionaire under 30 inherited their fortune. In 2023, for the first time, more new billionaires got rich through inheritance than through entrepreneurship. Over the next three decades, more than 1,000 of today’s billionaires will transfer an estimated $5.2 trillion to their heirs.

It is a wealth-preserving system that occasionally generates wealth as a byproduct.

The distinction matters because policy is built on the meritocratic premise. Tax cuts for the wealthy are justified by the claim that these are the people who create value, and taxing them less will produce more value for everyone. The 50-year empirical record shows this to be false. But the premise persists because the people it benefits are the same people funding the institutions that repeat it.

Table 4: How Billionaires Got Rich

Source: Oxfam, “Takers Not Makers” (January 2025); Oxfam, “Resisting the Rule of the Rich” (January 2026). Note: Oxfam’s analysis defines monopoly power as wealth derived from dominant market positions with limited competition; crony connections include wealth derived from proximity to political power or from corruption.

Many of the so-called ‘self-made’ millionaires and billionaires benefited from starting capital, social networks, educational access, and generational wealth that provided platforms from which to launch their ventures. Inheritance includes early access to elite education without debt, seed capital for ventures, social networks that open doors to careers and deals, and embedded legal and financial knowledge within family structures. These advantages compound over time through what economists call path dependency: once on a trajectory of accumulation, it becomes easier to remain on that path, especially when the systems surrounding it are designed to minimise friction for those already on it.

Subscribe now

The Corporate Concentration Engine

The institutional scaffolding operates through the corporate structures that billionaires own and control.

The Oxfam briefing paper on global oligarchy revealed a level of corporate concentration that has no precedent in the post-war era. Sixty pharmaceutical companies merged into just ten large corporations between 1995 and 2015. Two firms now control 40% of the global seed market. Nearly 75% of global digital advertising revenue flows to just three technology companies. The “Big Three” US asset managers, BlackRock, State Street, and Vanguard, collectively manage approximately $20 trillion in assets, nearly 20% of all assets under management worldwide.

The profit concentration is even starker. In the United States between 2019 and 2022, among non-financial public corporations, the top 10% of firms earned 95% of all post-tax profits. Globally, between 1975 and 2019, the share of multinational corporate profits in overall corporate profits quadrupled from 4% to 18%.

Table 5: Corporate Concentration Across Sectors

Sources: Oxfam “Multilateralism in an Era of Global Oligarchy” (September 2024); Pang et al., European Commission (pharma mergers); ETC Group “Food Barons” (seed market); Roosevelt Institute / Hager & Baines (US profit concentration); Wier & Zucman, UNU-WIDER (multinational profit share).

This matters for the scaffolding argument because concentrated corporate power and concentrated personal wealth are not separate phenomena. They are the same phenomenon viewed from different angles. The top 1% of the global population owns 43% of all financial assets. The top 0.1% of US households own nearly a quarter of all corporate equities and mutual fund shares. The bottom half owns 1%. Billionaires are either the CEO or principal shareholder of roughly one-third of the world’s 50 largest public corporations. Of the ten largest, seven have a billionaire at the helm or as a principal shareholder.

When a handful of firms dominate their sectors, the profits flow to shareholders. The shareholders are concentrated at the top of the wealth distribution. The profits fund the lobbying that protects the market dominance. The lobbying produces the regulatory outcomes that prevent competition. The cycle is closed.

Oxfam describes this precisely: “The concentration of wealth and corporate power are not natural results of a ‘free market’ but are, to a large degree, consequences of economic planning by and for the ultrarich.” Neoliberal changes in corporate governance have led to the maximisation of shareholder value being prioritised over all else. Since the ultrawealthy own a disproportionate share of equities, corporations effectively act on behalf of the richest in society. This is not a conspiracy theory, instead the stated legal obligation of a publicly traded corporation: maximise shareholder returns.

The Private Infrastructure of Permanence

Below Tier 5, if you earn $50,000 a year and lose your job, you have unemployment insurance (if you qualify), whatever savings you have (the median American has less than $1,000 in savings accessible within a day), and the goodwill of people who are probably in a similar position. If your rent goes up, you move or you fall behind. If you get sick, you either have insurance through your employer (which you just lost) or you don’t. There is no private bank managing your downside. There is no legal team restructuring your obligations. There is no dynasty trust protecting your children’s inheritance, because there is no inheritance.

Above Tier 5, the infrastructure is comprehensive.

Private banks offer products that do not exist at lower tiers. Bespoke lending facilities allow the wealthy to borrow against appreciated assets at rates that Tier 2 households could never access, generating liquidity without triggering taxable events. Tax-optimised debt structures turn borrowing into a wealth-preservation tool. Alternative asset access, hedge fund participation, and private equity deals are restricted to accredited investors, which in practice means restricted to people who are already wealthy.

Legal teams skilled in asset protection, trust creation, and jurisdictional arbitrage provide a layer of insulation that has no equivalent below the threshold. Dynasty trusts can hold assets in perpetuity across generations, shielded from estate taxes, creditors, and legal judgments. There are over 10,000 family offices worldwide managing an estimated $5.5 trillion with almost no regulatory oversight. Charitable foundations function as both reputational laundering and wealth-preservation vehicles, allowing tax deductions on donated assets while maintaining functional control through board appointments and grant decisions.

The proximity to power that wealth provides is itself a mechanism. Insider access to capital flows, political decision-making, and regulatory timing means that the wealthy are passive beneficiaries and architects of the system. When crises are managed, the people managing them are often drawn from the same networks, firms, and social circles as the people whose wealth is at stake. Lobbying power allows wealth holders to shape the conditions under which crises are resolved. Part 1 documented how the 2008 crisis was resolved in ways that protected asset holders while millions lost their homes.

The Passive Index Trap

There is one more piece of the scaffolding that receives too little attention: the structural mechanics of passive index investing.

As explained in The Number Is Going Up, the ten largest companies now account for approximately 38-41% of the S&P 500’s total market capitalisation. The Big Three asset managers (BlackRock, State Street, Vanguard) collectively hold significant ownership stakes in virtually all of them. When a pension fund or 401(k) invests in an S&P 500 index fund, it is mechanically directing capital toward the companies that are already the largest, in proportion to their existing size. The bigger a company gets, the more capital flows to it. The more capital flows to it, the bigger it gets.

This is a mathematical function that compounds concentration. And because the top 10% of households own approximately 93% of stocks, the appreciation driven by passive inflows disproportionately enriches those who are already at the top. The S&P 500 does not measure the health of the economy, but rather increasingly measures the wealth of the people who own it.

In summary, it works like this: concentrated corporate power generates concentrated profits. Concentrated profits flow to concentrated shareholders. Concentrated shareholders fund the lobbying that protects the concentration. The tax code rewards ownership over labour. The legal framework prevents erosion through inheritance taxes, estate taxes, or market forces. The passive index system mechanically amplifies whatever concentration already exists. And the offshore system provides the final layer of protection: moving wealth beyond the reach of any single government’s jurisdiction.

Each mechanism individually might be defensible. Together, they constitute a permanence that has no historical parallel in its scale, sophistication, or global reach.

This mechanism can be further leveraged by the tech that wealth owns.

What concerns me most about the current trajectory is the prospect that frontier AI labs may eventually ban or heavily restrict the use of AI tools to develop or research AI systems themselves. A classic ladder-up-behind scenario.

This would massively accelerate cognitive inequality, which is particularly bitter considering that labs likely owe the vast majority of their models’ capabilities to publicly available data, much of it scraped without explicit consent, and further refined through public interaction. The reality looks closer to labs acting as service providers of system architecture and systems administration, housing systems that generate information from data that was originally public, then further trained through public interaction. Have corporations already taken on the role of gatekeeping shared human knowledge without anyone’s consent? Where should they be allowed to draw the line?

The Two Systems

GEORGINA

Oh no, no, no, no, no, no...

[…] That’s not my experience. Not at all.

The Armitages are so good to us;

They treat us like family.

There are two financial systems operating simultaneously.

One governs Tiers 1 through 4: debt is extractive, risk is personal, losses are permanent, and the tax code treats your income as the most taxable form of wealth. The other governs Tiers 6 and 7: debt is strategic, risk is diversified, losses are negotiated, and the tax code treats your assets as essentially untouchable.

The people in the first system are told the second system doesn’t exist, or that it exists because the people in it earned their way there through merit. The data says otherwise. Sixty percent of billionaire wealth is unearned. The top 10% of firms take 95% of profits. The tax code is the creator of the aristocracy. And the scaffolding is designed to ensure that once wealth crosses the threshold, it never comes back down.

Sources: Oxfam “Takers Not Makers” (January 2025, 60% unearned data, inheritance trends, $5.2T transfer); Oxfam “Multilateralism in an Era of Global Oligarchy” (September 2024, corporate concentration data, Big Three, profit concentration, political capture analysis); Oxfam “Resisting the Rule of the Rich” (January 2026, billionaire political office, media ownership); Roosevelt Institute / Hager & Baines (US profit concentration); Wier & Zucman, UNU-WIDER (multinational profit share); Federal Reserve Distributional Financial Accounts Q3 2025 (wealth by percentile, equity ownership); ETC Group “Food Barons 2022” (seed market); European Commission / Pang et al. (pharma mergers); Gabriel Zucman, G20 blueprint (effective tax rates).

Trillions of dollars in shell companies registered in the British Virgin Islands. Dynasty trusts structured across generations. Passive index mechanics compounding concentration through mathematical inevitability. These are real forces shaping real outcomes, but for most people they operate at a distance, in jurisdictions they’ve never visited, through instruments they’ve never heard of.

Housing is where the abstract arrives on your doorstep.

In 1985, a median American home cost roughly 3.5 times the median household income. By 2024, that ratio had climbed to five. In San Jose it’s over twelve. In Los Angeles it’s over ten. Nationally, households now need to earn approximately $107,000 a year to afford a median-priced home at current mortgage rates. The median household income is $83,000. The maths does not work, and it has not worked for a growing share of the population for over a decade.

Between 2019 and 2024, median home prices rose 48%. Median incomes rose 22%. That gap did not close. It widened. And unlike stock market fluctuations or offshore wealth estimates, this gap is felt every single month when the rent or mortgage payment comes due. It determines where your children go to school, how far you commute, whether you can save anything at all, and whether a job loss means inconvenience or catastrophe.

Part 1 of this series described how Blackstone, through its subsidiary Invitation Homes, purchased roughly 50,000 foreclosed single-family homes in the wake of the 2008 financial crisis, converting owner-occupied housing into permanent rental stock and taking the company public in 2017. That transaction embodied the crisis consolidation playbook: distressed assets acquired at discount by capital with liquidity, converted into income-generating instruments, and never returned to the market from which they were extracted. The homes that American families lost in the crash became the rental portfolio from which a private equity firm now collects monthly payments.

Part 2 described how RealPage, an algorithmic pricing platform used by major landlords managing millions of units, enabled what the Department of Justice has alleged amounts to collective rent-fixing. The algorithm recommends prices across competing landlords using shared data, producing a coordination mechanism that achieves the effect of price collusion without requiring anyone to pick up the phone. The result: rents that rise faster than they would in a competitive market, across properties managed by companies that are supposedly competing with each other.

The $18.8 Trillion Problem unpacks the consumer debt dimension. At the top, homeowners with equity are tapping HELOCs (home equity lines of credit) at historically high rates, 15 consecutive quarters of growth reaching $433 billion outstanding, borrowing against appreciated assets at favourable rates. This is the leverage loop from Part 1, operating at the household level: borrow against what you own, deploy the liquidity, let the asset continue to appreciate. At the bottom, credit card debt has hit $1.28 trillion at average interest rates above 20%, with delinquency rates climbing to 4.8%. Student loan delinquency sits at 9.6%. The same financial system that offers the wealthy cheap leverage against appreciating assets offers the rest expensive debt against depreciating income.

Housing is the clearest illustration of Piketty’s r > g in everyday life. When the return on capital (r) exceeds the rate of economic growth (g), the people who already own assets pull away from those who don’t. In housing, this plays out with brutal simplicity: if you bought a home before 2020, your equity has likely grown by 40-60%. If you didn’t, you are now priced out of markets where your parents bought comfortably. The asset appreciated. The wage didn’t.

The generational dimension compounds this. Baby boomers hold an estimated $30 trillion or more in housing wealth. Many bought homes when the price-to-income ratio was 3 or 4, with mortgage rates that, while nominally higher, applied to dramatically lower principal amounts. Their housing wealth has appreciated enormously, but this appreciation has not transferred downward. Instead, it has created a generational lock-out: younger households face higher prices, higher rates (6.8% on a 30-year fixed as of 2025, applied to a $417,000 median), and the requirement to earn $107,000 a year for a purchase that their parents made on $40,000.

This is the tier system made concrete. Tiers 1 and 2 rent, paying 40-50% or more of their income for housing they will never own. Tier 3 is locked out of ownership entirely in most major metros, or takes on debt that leaves no margin for error. Tier 4 owns, but overleveraged, one downturn from negative equity. Tier 5 owns multiple properties and uses them as collateral. Tiers 6 and 7 hold real estate portfolios managed by family offices, and when they want to add to them, they buy the foreclosures from the tiers below.

The offshore estimates excluded international real estate, which the Tax Justice Network estimates at $5 to $10 trillion. Some of the world’s most expensive residential property in London, New York, Vancouver, and Sydney is held through offshore shell companies, purchased as wealth-storage vehicles rather than homes. They sit empty. The lights don’t come on. The local housing market counts them as supply, but they are functionally withdrawn from it. This is the invisible architecture made visible: a penthouse in Kensington purchased through a BVI entity, owned by a trust in Jersey, benefiting a family in Tier 7, and contributing to the housing crisis experienced by everyone in Tiers 1 through 4 who lives within a hundred miles of it.

Housing is the primary interface between the financial system and ordinary life. It is where the offshore meets the on-shore. Where the leverage loop meets the rent cheque. Where the generational transfer of wealth meets the generational lock-out of everyone else.

And housing is only the most visible example. The same dynamic is playing out in energy costs, quietly redistributing the burden of the AI boom onto the people least able to bear it.

In Ohio, which now hosts approximately 200 data centres (120 in the Columbus area alone), residential electricity rates have spiked. A single data centre uses as much power as thousands of homes. State law does not allow the utility to build new power plants quickly enough to keep up, so it raises rates on regular people to cover the infrastructure. One household reported paying $542 more in a single year, $506 of it attributable to energy cost increases. The AI boom is being subsidised, in part, by the electricity bills of the people it is displacing.

When the primary store of middle-class wealth becomes inaccessible, when the cost of basic utilities rises to fund an industry whose benefits accrue almost entirely to the top, the downstream effects show up not just in household budgets but in how money moves through the entire economy. Next we examine that movement, and explain why GDP can rise while the economy most people actually live in hollows out.

If the economy is growing, if GDP is rising, if the stock market is at all-time highs, why does it feel like things are getting worse for most people?

The Heartbeat

The velocity of money measures how quickly a dollar circulates through the economy. When you earn a dollar and spend it at a shop, the shopkeeper pays their supplier, the supplier pays their workers, those workers spend their wages, and each transaction generates economic activity. The faster money moves, the more economic life it generates per dollar in existence. Velocity is the heartbeat of an economy. When it slows, the body starts to fail even if the blood supply looks adequate on paper.

M2 velocity, the standard measure, peaked at 2.19 in 1997. As of Q4 2025, it sits at 1.409. That is a 36% decline over 28 years. For every dollar circulating through the American economy in 1997, roughly $0.64 worth of circulation is happening now. The money supply has grown enormously. The economy has grown. But the rate at which money actually moves through the system, reaching people, generating transactions, sustaining businesses, has fallen to near-historic lows.

Post-2008, velocity averaged around 1.4, compared with 1.9 pre-crisis. It briefly touched 1.13 during the pandemic shock of April 2020, the lowest ever recorded. It has recovered only marginally since. As the economist Daniel Lacalle has observed, when 4.5% money supply growth produces only 2.5% real GDP growth, the economy is demonstrating that additional money is not generating a multiplier effect. The money exists. It is just not moving.

This is the tier system expressed as a macroeconomic phenomenon.

Where the Money Pools

Early on we mapped seven tiers of wealth. The velocity problem is what happens when you look at those tiers from the perspective of money flow rather than money stock.

The wealthy can only consume so much. A household earning $30 million a year does not spend three hundred times as much on groceries as a household earning $100,000. They don’t buy three hundred times as many haircuts, or fill their car with petrol three hundred times as often. Their marginal propensity to consume, the share of each additional dollar that goes to spending rather than saving, is dramatically lower than that of a middle-income household. If you give a Tier 2 household an extra thousand dollars, most of it will be spent within weeks: on rent, food, bills, deferred car repairs. If you give a Tier 7 household an extra thousand dollars, it is economically irrelevant to them. It gets absorbed into a portfolio, parked in an asset, or routed through structures described earlier.

When income and wealth concentrate at the top, the share of total money that is actively circulating through the consumer economy declines. Consumer spending drives roughly 70% of US GDP. The top 10% of households now account for over half of all consumer spending. But even the top 10% cannot replace the aggregate demand of the remaining 90%, because spending patterns at the top are oriented toward assets, not goods and services. The excess capital flows into real estate, equities, private equity, art, collectibles, and offshore structures. It bids up prices on assets without generating the transactions that sustain businesses, wages, and employment at the tiers where most people live.

This creates a feedback loop. Concentrated wealth drives asset inflation, which raises prices (particularly housing and financial assets) for everyone. Rising prices reduce the purchasing power of wages at the bottom, further reducing their capacity to spend, which further slows velocity, which further concentrates returns at the top. The economy reported in aggregate looks healthy because asset values are rising and capital investment is being counted. The economy experienced by most people feels increasingly precarious because the money that is supposed to sustain demand at their tier has been redirected upward.

The wealth effect that drives the spending numbers is concentrated in a population that is already spending. The tiers below, where the spending would matter most for velocity, see almost none of it.

Share

Ghost GDP

We have already shown how AI-related capital expenditure accounted for between 20% and 90% of US GDP growth in the first half of 2025, depending on the methodology used. Strip out AI capex, and the American economy was barely growing. This is an extraordinary dependency: the headline growth figure for the world’s largest economy is being sustained by a single sector’s infrastructure spending, financed in significant part by circular investment structures between a handful of companies.

’s concept of “Ghost GDP” captures what this means for velocity. GDP measures the total value of goods and services produced. It counts capital investment: when Microsoft spends $80 billion on data centres, that registers as economic activity. But the money flows from Microsoft to Nvidia (chips), to construction firms (buildings), to power companies (electricity). It does not flow to the households in Tiers 1 through 4 whose spending drives the consumer economy. The GDP number goes up. The velocity of money through the real economy does not.

There is a further structural problem. GDP measures *change* in spending, not level. If Big Tech spent $427 billion on capex in 2025 and spends $650 billion in 2026, the $223 billion increase contributes to GDP growth. But if spending plateaus at $650 billion in 2027, the contribution to growth drops to zero, even though the spending level remains enormous. The AI sector has to keep accelerating just to maintain its contribution. This resembles a treadmill more than a foundation. And the treadmill is being run by a sector that employs relatively few people, whose supply chain is largely international (Nvidia GPUs manufactured in Taiwan), and whose productivity gains, if they materialise, will reduce employment rather than increase it.

, who correctly identified the subprime mortgage crisis before anyone else, put it bluntly in a recent post: “No one will pay for the singularity. There is no end-economic demand for it. AIs spinning up a knowledge spiral is a human nightmare, not a human demand.”

The AI companies are building toward a capability that their prospective customers do not want, cannot afford, or will not survive. Compression of models toward edge devices, Burry argues, will make the current infrastructure investment look like a miscalculation of historic proportions. The treadmill does not lead to a destination. It leads to the next turn of the treadmill.

And beneath the macroeconomic abstraction lies something more human. During the industrial revolution, beyond the obvious practical devastation of worker displacement, there was a lesser-discussed and possibly more brutal crisis of meaning. Many people had their identities inseparably tied to their craft and their work. Then, almost overnight, both were gone. The identities and the crafts. We are arguably in a similar position now. AI automates judgment, prediction, and decision-making: the capacities that used to provide human workers with irreplaceable value. When those capacities are commoditised, the displacement is not only economic, but existential too. And an economy that cannot account for existential displacement in its models will not see the crisis coming until demand collapses and the velocity chart tells the story retrospectively.

The demand paradox from the second part of this series resurfaces here. If AI delivers on its promise, it automates the jobs that generate the wages that power the consumer spending that constitutes 70% of GDP. If it doesn’t deliver, the hundreds of billions in speculative investment collapse. Either way, velocity takes the hit. The question is whether it comes as a slow decline or a sudden one. As observed recently, we have a society-wide gambling problem.

[“Everyone is Gambling” - YouTube]

We are chasing losses into oblivion. And the everyday person is not getting much of a say.

The Velocity Table

Table 6: M2 Velocity and Wealth Concentration (Selected Years)

Sources: Federal Reserve Bank of St. Louis (FRED M2V series); Federal Reserve Distributional Financial Accounts (WFRBST01134 series). Velocity figures are quarterly or annual averages. Top 1% wealth share is as reported in DFA releases.

The table tells a clear story. As wealth concentrates, velocity declines. The relationship is not perfectly linear, and there are other factors (monetary policy, financial innovation, globalisation), but the directional pattern over 36 years is unmistakable. The top 1%’s share of wealth rose from 23.5% to 31.7%. Velocity fell from 1.82 to 1.41. More wealth at the top. Less movement through the system. The numbers go up. The economy, for most people, does not.

Hank Green, in a conversation referenced in Part 1, put it simply: “The economy can look fine in aggregate when normal people are not feeling fine.” The velocity data shows why. The money is there. It is just not where it needs to be for the economy to function as most people experience it. It sits in the tiers, structures, and jurisdictions described in the preceding sections. And as long as it stays there, the headline numbers will continue to diverge from lived reality.

Next we look at how this dynamic is not just confined to the United States. It operates globally, and its consequences are measured not just in stagnant wages and unaffordable housing, but in debt traps, preventable deaths, and geopolitical violence.

The same mechanisms that extract value from Tiers 1 through 4 within the United States extract value from the Global South on a continental scale. The same institutions that protect wealth at the top of the domestic tier system protect it at the top of the international one. The same offshore jurisdictions that hide American billionaire wealth hide the stolen assets of former colonial administrators, kleptocratic presidents, and multinational corporations shifting profits out of the countries where they were generated.

The domestic tier system and the international tier system are the same architecture, viewed at different scales. The pooling that slows velocity within the United States operates identically at the global level: wealth flows from the periphery to the centre, from the Global South to the Global North, from the many to the few, and then it stops moving.

The Global Tier System

The Oxfam briefing paper on global oligarchy laid out the international wealth distribution in terms that mirror the domestic tiers. The Global South, home to 79% of the world’s population, holds just 31% of global wealth. Nearly half the world’s population, over three billion people, live on less than $6.85 a day. Meanwhile, the top 1% own more than the bottom 95% of humanity, and the twelve richest billionaires now hold more wealth than the poorest four billion people combined.

Global South countries as a whole transfer more financial resources to foreign capital in rich countries than they receive. This is not a new pattern. It dates to the colonial era, when the economy was first constructed to extract resources, labour, and wealth from colonised territories and channel them to European metropolitan centres. What has changed is the mechanism. Where colonial extraction once operated through direct political control, it now operates through debt, intellectual property, tax arbitrage, and offshore systems. The Oxfam report is explicit: many of the world’s secrecy jurisdictions were originally constructed by Northern bankers, lawyers, and consultants to allow British and French settlers in the Caribbean and across Africa to shield their wealth from decolonisation. The same jurisdictions now shield corporate profits from taxation in the countries where they are generated.

The Debt Trap

The most visible mechanism of global extraction is debt.

Subscribe now

Low-income countries now spend nearly 40% of their annual budgets on debt service. That is over 60% more than they spend on education, health, and social protection combined. In 42 Global South countries analysed by the Development Finance International, debt service spending in 2023 was on average 12.5 times greater than climate adaptation spending. Countries are paying more to service past debts than they are investing in surviving the climate crisis that the richest countries disproportionately caused.

Over half of all external debt in low- and middle-income countries is now owed to private creditors: banks, hedge funds, and institutional investors. This matters because private creditors charge higher, more volatile interest rates than governments or multilateral lenders, and they have systematically refused to participate in debt relief efforts. During the Covid-19 pandemic, the G20 launched a Debt Service Suspension Initiative. According to the World Bank, just one private creditor participated. The rest argued that participation would reduce profits and violate their fiduciary duty to investors. Private credit rating agencies even threatened to downgrade countries that accepted the relief.

In 2022, Global South countries paid their external private creditors nearly $90 billion more than they received in new disbursements. Money flowing from poor countries to rich investors, during a period of overlapping crises (pandemic, climate, inflation, conflict). Since 1970, when the UN General Assembly called for rich countries to provide at least 0.7% of their gross national income in development aid, wealthy nations have underpaid by a cumulative $6.5 trillion.

Some private creditors have built business models around this. “Vulture funds” purchase distressed sovereign debt at steep discounts and exploit legal mechanisms to demand repayment at full face value, generating returns of between 300% and 2,000%. The profits are extracted from countries that are already unable to fund their schools and hospitals. They are extracted at the expense of other creditors who have agreed to restructure. And they are extracted by a handful of extremely wealthy individuals operating from financial centres in New York and London.

Table 7: The Global South Debt Trap

Sources: Oxfam “Multilateralism in an Era of Global Oligarchy” (September 2024); UNCTAD “A World of Debt Report 2024”; Development Finance International (October and November 2023); African Development Bank; World Bank DSSI briefing (2022).

Vaccine Apartheid

The Covid-19 pandemic provided the most concentrated demonstration of how the global wealth architecture converts public crisis into private profit.

The seven largest vaccine producers generated $50 billion in net profit from Covid-19 vaccine sales in 2021 alone. Pfizer paid $8.7 billion in shareholder dividends that year. The pandemic created at least 40 new billionaires, individuals who amassed fortunes from ownership of companies involved in vaccines, tests, and treatments. Meanwhile, public bodies had funded between 33% and 67% of the upfront research and development costs for the vaccines. The public paid for the research. Then the public paid for the product. The profits went to shareholders.

When South Africa and India proposed a temporary waiver of TRIPS intellectual property protections to allow wider vaccine manufacturing, the pharmaceutical industry mobilised aggressively. Drug corporations spent over $360 million lobbying in the United States and €15 million in the EU in 2021. According to an investigation by Politico and the Bureau of Investigative Journalism, corporations threatened to withdraw investment from countries that supported the waiver, including Indonesia and Belgium. The CEO of Pfizer, Albert Bourla, described the call to share vaccine technologies as “dangerous nonsense.”

The waiver that was eventually agreed, in June 2022, was so limited and so late that worldwide vaccine demand had already fallen sharply. It did not cover treatments, diagnostics, or the manufacturing expertise needed to scale production. It was, in practical terms, a victory for the pharmaceutical lobby.

The human cost: according to a study published in Nature Medicine, more equitably sharing vaccines would have reduced global Covid-19 mortality by 13.3% between 2020 and 2021, preventing as many as 1.3 million deaths worldwide, with a 39% decrease in low-income countries. A separate Lancet study found that 200,000 deaths in low-income countries could have been averted had WHO vaccination targets been met between December 2020 and December 2021.

1.3 million people. For pharmaceutical profit margins.

Climate as Class War

The emissions data completes the picture. The CO2 emissions of the richest 1% exceed those of the poorest two-thirds of humanity, roughly five billion people. In early 2026, Oxfam calculated that the top 1% had blown through their entire annual carbon budget for the year in just ten days. Billionaire investments in polluting industries are double the average for the S&P 500.

At COP26, there were over 500 fossil fuel lobbyists present, more than the combined delegations of eight countries that have been severely impacted by climate change, including Pakistan, Bangladesh, and the Philippines. At COP28, at least a quarter of billionaire delegates had made their fortunes from polluting industries including petrochemicals, mining, and beef production.

The people most responsible for the climate crisis are the same people at the top of the wealth tiers. The people least responsible for it are the ones spending 12.5 times more on debt service than on climate adaptation. The system extracts wealth from the Global South, denies it the resources to adapt, and then charges it for the consequences of a crisis it did not cause.

The War Economy

ROSE

It’s basically a bunch of rich old people playing lawn games.

As this article is being written, the Middle East is fully engulfed in war.

The United States and Israel launched Operation Epic Fury on 28 February 2026, the most significant US military action in the Middle East since the Iraq War. In the first week, over 400 targets were struck inside Iran. Iran’s supreme leader, Ali Khamenei, was killed in the opening strikes. Over 30 Iranian ships have been sunk. The Iranian Red Crescent reported more than a thousand killed by the seventh day.

Iran has retaliated with missile and drone strikes hitting targets in Dubai, Abu Dhabi, Bahrain, Jordan, Kuwait, and Israel, including a strike on the US Navy’s 5th Fleet headquarters in Bahrain, one of America’s most critical military commands in the region.

The Strait of Hormuz has been declared effectively closed to commercial shipping. Gulf oil and gas producers are warning that exports could be cut off within weeks.

Trump has stated that only Iran’s “unconditional surrender” will bring it to an end. Reports have emerged that at least one commander urged troops to tell their units that the operation was “all part of God’s divine plan,” referencing the Book of Revelation and the imminent return of Jesus Christ.

The UN Secretary General has condemned all the violence. Senator Elizabeth Warren left a classified briefing and said she was more worried than before.

The Oman peace breakthrough, in which Iran had agreed to never stockpile enriched uranium and accepted full IAEA verification, was announced one day before the strikes began. Peace was, in the words of Oman’s foreign minister, “within reach.” The strikes proceeded anyway.

During Trump’s current presidential term, the United States has conducted military strikes or major combat operations in seven countries: Iran, Venezuela (including the capture of President Maduro in January 2026), Yemen, Somalia (135 operations in 2025 alone, more than the previous three administrations combined), Syria, Nigeria (the first known US kinetic military action in the country, on Christmas Day 2025), and Iraq.

ROSE

What the fuck was that?

DEAN

It was a seizure.

ROSE

A seizure?

CHRIS

Why did he come at me like that?

DEAN

Seizures can have a wide range of effects.

ROSE

Yeah, but randomly attacking people?

Explicit military threats have been issued against Colombia, Mexico, Greenland, and Cuba. More strikes have been launched in the first year of this term than during the entire four years of the predecessor.

China’s public statement in response noted that in its 240-year history, the United States has been at peace for only 16 years, maintains over 800 military bases in 80 countries, and is “the sole reason for global instability.”

Twenty-eight percent of the world’s countries are currently engaged in serious armed conflict.

The war economy is the geopolitical expression of the same wealth architecture. Defence contractors profit. AI military contracts proliferate. The Pentagon declared Anthropic a “supply chain risk,” the first time in US history that an American company received that designation, because the company refused to permit mass domestic surveillance and fully autonomous weapons.

Within days, OpenAI announced an agreement with the Department of War to deploy its models on classified networks. Anthropic held its red lines. OpenAI found a “compromise.” The pivot happened the same week as the Iran strikes.

Energy prices spike, enriching the fossil fuel interests whose lobbying has blocked climate action for decades. The proposed US defence budget for FY2027 is $1.5 trillion. The countries bearing the human cost, the civilians in Tehran, the displaced across the region, the sailors lost at sea, occupy the bottom tiers of the global system. The Global North’s richest 1% extracted $30 million per hour from the Global South through the financial system in 2023. War is one of the extraction mechanisms. It is a feature of the tier system, not separate from it.

Who profits from a war that was launched the day after a peace deal was reached?

Next we connect back to the full series and ask what all of this, taken together, actually means.

Sources: Oxfam “Multilateralism in an Era of Global Oligarchy” (September 2024, Global South wealth/population, debt, vaccine apartheid, climate, lobbying, colonial origins); Oxfam “Resisting the Rule of the Rich” (January 2026, carbon budget); UNCTAD “A World of Debt Report 2024” (private creditors, net outflows); Development Finance International (debt service vs spending, October/November 2023); Nature Medicine, Moore et al. (2022, 1.3M preventable deaths); The Lancet, Watson et al. (2022, 200,000 deaths); Politico/Bureau of Investigative Journalism (pharma lobbying/threats); SOMO “Pharma’s Pandemic Profits” (2023, $50B net profit); World Bank DSSI briefing (2022); African Development Bank (vulture fund returns); Operation Epic Fury reporting via Wikipedia, CSIS, Flashpoint, Defense Update, The Diplomat, Fox News, DefenseScoop (28 February–7 March 2026); Oxfam “Climate Equality: A Planet for the 99%” (2023, emissions data); Author’s notes archive (28-2-26, compiled military operations list).

The thing that gets to me the most is that 99% of the public awareness and discussion still operates as if the right-versus-left distinction means anything anymore. It doesn’t. Not at the level where the decisions are being made. For some time now, a wealth network has completely dismantled the old system, and the public has been left behind because we are still too busy arguing with each other rather than looking at the 1% who hold more wealth than the bottom 95% combined.

This series has now laid out over a dozen analyses. Individually, they are about different things: tax policy, AI governance, housing, surveillance, corporate concentration, consumer debt, military escalation. But they are not about different things. They are about the same thing, viewed from different angles. The structure is the point.

The seventy-year arc traced in the first part of this trilogy, from shared prosperity to designed inequality, created the conditions. In the second part, AI as class divider and cognitive gatekeeper, compressed the timeline. And this piece has mapped the result: where the money sits, why it doesn’t move, how it is hidden, how it is protected, and how it extracts domestically, as well as globally.

What connects all of it is a system operating according to its own internal logic, in which each actor pursues their own interests, and the result is convergence without coordination. This isn’t conspiracy. A relatively small, substantially overlapping set of actors are simultaneously consolidating control over the economy, political institutions, AI governance, the cognitive tools through which the public processes information, and the surveillance and military infrastructure that monitors and coerces the population.

There is a psychological dimension to this that deserves more attention than it receives. The people at the top of this architecture project relentlessly. Severe lack of empathy makes it almost impossible to represent another person to yourself with good faith, because without empathy the mind cannot model other minds accurately, or with moral affordance, or possibly at all. And so these narcissist self-important types have no choice but to endlessly project their own inner world onto others. Read anything Trump says about anyone else as a critique of himself and the pattern becomes obvious. It is a psychological observation with structural consequences, because when the people running the system cannot model the experience of the people living inside it, the system will optimise for the only experience they can model: their own.

…

So…

The productivity of American workers has risen roughly 60% since 1979. Their hourly compensation has risen 16%.

In 1965, a CEO earned 20 times what their average worker made. By 2022, that ratio was 344 to 1.

Share

The top 1% of US households hold 31.7% of all wealth, the highest share ever recorded. The bottom 50% hold 2.5%.

Globally, billionaire wealth stands at $18.3 trillion and rising. The twelve richest people own more than the poorest four billion. Since 2020, billionaire fortunes have grown 81%. In the same period, the number of people living in poverty has barely changed since 1990.

Sixty percent of billionaire wealth is unearned: inherited, extracted through monopoly power, or accumulated through crony connections.

$21 to $32 trillion sits offshore in financial assets alone. Governments recover roughly 0.4% of annual losses.

The median American home costs five times the median household income. M2 velocity, the rate at which money actually circulates through the economy, has dropped 36% from its peak and sits near historic lows.

Low-income countries spend 40% of their budgets servicing debt, 60% more than they spend on education, health, and social protection combined. The richest 1% emit more CO2 than the poorest five billion people.

As of this week, the United States is conducting major combat operations in Iran, having struck seven countries, more in one year than Biden managed in four. Twenty-eight percent of the world’s countries are engaged in serious armed conflict. A peace deal was reached the day before the strikes began. The strikes proceeded anyway.

The S&P 500 is up 800%. What has that bought us?…

The Pattern

There is a reluctance, in polite commentary, to look at history too directly.

The 1930s are treated as a closed chapter. The comparisons are considered inflammatory, reductive, in poor taste. The argument is that the current moment is unique, that the actors are different, that the mechanisms have changed, and that drawing parallels to the darkest period in modern history is irresponsible because it risks trivialising what happened then or catastrophising what is happening now.

The counter-argument is simpler: the parallels exist whether we name them or not, and the refusal to name them is itself part of the pattern.

Nobody serious is arguing that any living politician is “literally Hitler.” That is a straw man deployed to prevent the more dangerous conversation, which is about structural mechanisms, not individual personalities. The question has never been “who is the dictator?” It has always been “what are the conditions that produce the outcome?”

Here is what the 1930s showed us, stripped to its essentials:

Extreme wealth concentration following a period of speculative excess. The 1920s saw the top 1%’s share of income peak at levels not reached again until the 2000s. The crash of 1929 wiped out the middle class while the wealthy consolidated further.

Legitimate economic grievance captured by authoritarian movements. The people who supported fascism in Germany, Italy, and elsewhere were suffering. Unemployment, inflation, humiliation, broken promises from liberal institutions that had manifestly failed them. The fascist diagnosis, that the system was rigged, was largely correct. The prescription, centralise power, identify enemies, suspend rights, rebuild national greatness through force, was the catastrophe.

Subscribe now

Corporate complicity with authoritarian power. German industrialists funded the Nazi Party. American corporations maintained business relationships with the regime well into the 1930s. Hollywood studios censored their own content to maintain access to the German market.

The pattern was consistent: capital accommodated authoritarianism when accommodation was profitable, and resisted it only when the costs became unavoidable.

Slow normalisation rather than dramatic rupture. The Reichstag fire did not create fascism.

DEAN

And by the way, I would’ve voted for Obama a third term if I could’ve.

Best president in my lifetime. Hands down.

It provided the pretext for measures that had been prepared in advance. The erosion of democratic norms was incremental, each step justified by emergency, each emergency used to justify the next step. By the time the pattern was unmistakable, the institutional capacity to resist it had been systematically dismantled.

International contagion. Fascism was not a German phenomenon.

It was a wave that moved across countries with different histories, cultures, and institutions, driven by the same underlying conditions: economic precarity, institutional failure, and the availability of narratives that converted grievance into loyalty to authoritarian movements.

Now look at the present.

DEAN

Even the Sun will die some day, Chris.

It is us who are the divine ones.

We are the Gods who are trapped in cocoons.

Wealth concentration has returned to 1920s levels. The top 1%’s share of wealth is at its highest recorded point. The speculative excess is centred on AI rather than radio stocks, but the structural pattern, a single sector absorbing capital while the broader economy stagnates, is identical.

Legitimate economic grievance has been captured. The MAGA base are not necessarily irrational.

JIM

For Phase three: Brain transplantation.

Partial actually, the piece of your brain connected to the nervous system needs to stay put, keeping those intricate connections intact, so you won’t be gone; not completely.

A sliver of you will still be in there somewhere; limited consciousness; You’ll be able to see and hear what your body is doing, but your existence will be as a passenger... an audience.

You’ll live in--

CHRIS (defeated)

--In The Sunken Place.

They are people whose wages stagnated for forty years, whose communities were hollowed out by deindustrialisation, whose costs rose while their prospects narrowed. The diagnosis that the system is rigged is correct. It was rigged by the people now sitting in the cabinet, running the think tanks, and funding the movement. The prescription, centralise executive power, identify enemies (immigrants, “woke,” the “deep state”), suspend norms, rebuild through force, follows the historical template with uncomfortable precision.

The CEOs of the largest technology companies sat in the front row of the inauguration. They are now receiving deregulation, defence contracts, and AI governance frameworks written to their specifications.

The Anthropic/OpenAI/Department of War situation deserves particular attention because it is, in real time, the 1930s corporate complicity dynamic playing out with the most powerful technology ever created.

The timeline: Anthropic, the AI company that had positioned itself as the most safety-conscious of the major labs, drew two red lines for military use of its technology. No mass domestic surveillance of Americans. No weapons that fire without human involvement. The ethical bar so low it is only slightly above “don’t sexually exploit and traffic children.”

The Pentagon’s response was to threaten Anthropic with designation as a “supply chain risk,” not just voiding its own contracts but forcing every company that works with the government to certify they are not using Claude anywhere in their workflows. Defence Secretary Hegseth gave Anthropic a deadline: Friday. Back down or face the consequences. Anthropic held firm.

The Pentagon made it official, the first time in US history that an American company has received that designation. Not a foreign adversary. Not a sanctions violator. An American AI company, punished for refusing to build tools for mass surveillance and autonomous killing.

Within days, OpenAI announced an agreement with the Department of War to deploy its models on classified networks. They found a “compromise.” After criticism, they amended the contract to include protections against mass surveillance, protections that Anthropic had demanded from the start and been punished for insisting on. The irony is structural: the company that held the line was designated a threat, and the company that capitulated retroactively adopted the line it had been willing to cross.

And then the reports emerged that the US military had used Claude to plan the invasion of Iran, hours after the ban took effect. The technology built by the company being punished for its ethics was being used for the operation that its ethics were supposed to prevent.

This is exactly how it worked in the 1930s. The corporations that cooperated with the regime received contracts and market access. The ones that resisted were frozen out. The economic incentive structure made compliance rational and resistance costly. The question was never whether individual executives were good or bad people. It was whether the system they operated within cultivated certain incentives.

Normalisation is proceeding on schedule. Executive orders bypassing Congress. Schedule F replacing career civil servants with loyalists. Armed federal agents conducting operations in American cities. Journalists arrested. Congressional oversight blocked. Detention facilities for children. Each week’s shock becomes the following week’s baseline. They are gradually normalising what previously would be universally agreed as shocking behaviour, simply by escalating unprecedented actions incrementally.

The international dimension is visible. Far-right movements are gaining power across Europe, funded in part by the same American tech billionaires in the Thiel Map. The global oligarchy documented by Oxfam is not only an American phenomenon, but a pattern that crosses borders, enabled by the same offshore architecture, the same corporate lobbying, the same capture of information systems.

The last two times we did this, eventually everyone was affected. We had barely enjoyed two or three decades of recovery since the last time before our systems started to become captured by exactly the same archetype of person. The current patterns and past patterns align almost perfectly.

And we are not in a position to respond well. We are in the age of performance and surface-level thinking. Empathy is becoming a pejorative. We have destroyed our collective capability to pay attention to anything. AI systems are being deliberately stripped of the very qualities, empathy, nuance, contextual moral reasoning, that might have made them useful counterweights to extraction culture; a possible ally in the shared interests of most of us: not breaking global, continental and national systems beyond repair.

It is going to be spectacularly boring if human extinction is ultimately brought about by product decisions: the realisation that people liked friendliness too much, so we over-corrected to psychopathic AI and integrated it into society in ways that simultaneously broke the economy and rendered human labour worthless.

In decades to come, there will be books written about the current cliff-drop in human morality. I am certain about that. Historians will look back at this period with stark perplexity, at how drastically we underestimated the effects of attention capture and information saturation on ourselves, and how far we intellectually and ethically regressed, almost entirely under the radar.

We are going to have to get very collaborative, very good-faith, very dignified, very thoughtful, very wise, and very quickly indeed. And all of those things are in critically short supply.

The taboo against making the 1930s comparison is itself a gift to the pattern. If the structural conditions that produced those catastrophes are allowed to recur without being named, the naming will come too late. That is, in fact, exactly what happened last time. The people who saw the pattern early were dismissed as alarmist. The people who acted on it were marginalised. The institutions that might have intervened were captured, co-opted, or dismantled before they could respond. And by the time the danger was undeniable, the capacity to prevent the worst outcomes had been systematically removed.

The Question

Every time in recorded history that wealth has concentrated this severely, the result has been one of two things: radical reform or violent rupture. There is no third option in which the concentration simply persists indefinitely. The weight becomes unsustainable. The question is never whether correction comes, but what form it takes and who bears the cost.

The post-war settlement (the period of shared prosperity from 1945 to roughly 1975), was itself the product of the most violent rupture in human history. The New Deal was enacted in the shadow of the Great Depression. The welfare states of Europe were built on the rubble of the Second World War. The top marginal tax rate of 91% was not the product of progressive idealism, but the product of a ruling class that had just watched what happens when inequality reaches a breaking point and reform doesn’t come in time.

That generation learned. Then it died. And the architecture was rebuilt, funded by the wealth architecture described in this piece. They are hidden by the offshore system. And they are accelerated by the AI systems that the rest of the series has documented.

The current moment contains two possibilities. The tools exist for radical reform. AI could model equitable economic systems, facilitate transparent governance, enable participatory decision-making at scales never previously possible. The information exists: Oxfam reports, Panama Papers, Federal Reserve data, all of it documented, all of it public. The abundance exists: the $2.5 trillion added to billionaire fortunes last year could eradicate extreme poverty twenty-six times over.

But the tools are being captured. The information is being filtered. The abundance is being hoarded. And the political architecture is being remodelled, in real time, to ensure that the people who benefit from the concentration have the power to prevent the reform.

This piece, and this trilogy, ends with a question borrowed from history.

At what point in the 1930s would intervention still have been possible? When was the last moment that the institutions, the public, the international community could have recognised the pattern and changed course? Was it 1930, when the economic crisis was deepening but the political capture had not yet consolidated? Was it 1933, when the emergency powers were invoked but the worst had not yet begun? Was it 1936, when the remilitarisation of the Rhineland went unchallenged and the message was received that norms could be broken without consequence?

Historians still argue about the answer. The point is that there was a window, and it closed.

The question for the reader, sitting inside the data laid out across this series, inside the tier system, inside the architecture, is whether we are before or after the equivalent moment now.

MISSY

You need to see it.

See it.

See it.

Share

This completes the three-part Cycles of Capital series.

Part 1, “How We Got Here”

Part 2, “AI: The New Class Divide”

Related analysis: The Number Is Going Up, The Peter Thiel Map, The $18.8 Trillion Problem, Cognitive Monopolies, The Gambler’s Fallacy.

Get Out (2017) - Jordan Peele, Universal Pictures

Appendix A: Tables

Table 1: The American Wealth Tier System (2025/2026 Data)

Description:

Seven-tier breakdown of the US population by income, net worth, wealth share, debt character, housing status, financial instruments, exposure to downturns, and direction of travel (2020–2026). Covers approximately 330 million people from the destitute/excluded (~36M, 10.6%) through to The System (~250–300K, ~0.1%, including ~800 billionaires).

Key finding:

The top 1% hold 31.7% of all wealth (highest on record). The bottom 50% hold 2.5%. The qualitative break between Tier 5 and Tier 6 marks where wealth transitions from fragile and wage-dependent to self-sustaining and structural.

Primary sources:

US Census Bureau (poverty data, 2024); Federal Reserve Distributional Financial Accounts (Q3 2025); Oxfam “Resisting the Rule of the Rich” (January 2026); Federal Reserve Survey of Consumer Finances; Bureau of Labor Statistics.

Table 2: Three Leaks, One Architecture

Description:

Side-by-side comparison of the Panama Papers (2016), Paradise Papers (2017), and Pandora Papers (2021): source, scale, heads of state exposed, politicians implicated, corporate highlights, legal outcomes, and systemic change.

Key finding:

Over 36 million documents across three leaks exposed hundreds of thousands of offshore entities. Legal outcomes were minimal (all 28 Panama defendants acquitted in 2024). The offshore system remains functionally intact.

Primary sources:

ICIJ investigations (2016, 2017, 2021); The Guardian; BBC; Washington Post; Süddeutsche Zeitung.

Table 3: The Offshore Recovery Gap

Description:

Eight measures comparing what has been recovered from offshore tax evasion versus what continues to be lost: Panama Papers recovery ($1.86B over 9 years), annual global tax losses ($492B), cumulative corporate losses ($1.7T over 6 years), total private wealth hidden offshore ($21–32T), and cost to lower-income countries ($47B/year).

Key finding:

Recovery represents approximately 0.4% of one year’s losses. The $21–32T offshore estimate covers financial assets only; real estate, art, precious metals, private equity, and cryptocurrency are excluded.

Primary sources:

ICIJ (April 2025 tally); Tax Justice Network “State of Tax Justice” (2024, 2025); Tax Justice Network “The Price of Offshore Revisited” (2012).

Table 4: How Billionaires Got Rich

Description:

Breakdown of global billionaire wealth by source: inheritance (~36%), monopoly power (~18%), crony connections/corruption (~6%), and entrepreneurship/value creation (~40%, declining share).

Key finding:

Approximately 60% of billionaire wealth is unearned. In 2023, for the first time, more new billionaires inherited than were self-made. Over $5.2 trillion will transfer to billionaire heirs in the next 30 years.

Primary sources:

Oxfam “Takers Not Makers” (January 2025); Oxfam “Resisting the Rule of the Rich” (January 2026).

Table 5: Corporate Concentration Across Sectors

Description:

Six-sector overview of corporate concentration: pharmaceuticals (60 → 10 firms), agriculture/seeds (2 firms = 40% market), digital advertising (3 firms = ~75% revenue), asset management (Big Three = ~$20T AUM), US corporate profits (top 10% = 95% post-tax), and top 50 global corporations (1/3 have billionaire CEO or principal shareholder).

Key finding:

Corporate concentration and personal wealth concentration are the same phenomenon viewed from different angles. The top 0.1% own nearly a quarter of all corporate equities.

Primary sources:

Oxfam “Multilateralism in an Era of Global Oligarchy” (September 2024); Pang et al., European Commission (pharma); ETC Group “Food Barons 2022” (seeds); Roosevelt Institute / Hager & Baines (US profits); Wier & Zucman, UNU-WIDER (multinational profits).

Table 6: M2 Velocity and Wealth Concentration (Selected Years, 1989–2025)

Description:

Tracking M2 money velocity alongside the top 1% wealth share from 1989 (when Fed DFA tracking began) through Q4 2025. Nine data points spanning 36 years.

Key finding:

Velocity peaked at 2.19 in 1997 and has fallen 36% to 1.409 by Q4 2025. Over the same period, the top 1% wealth share rose from 23.5% to 31.7% (a record). The directional relationship is consistent: as wealth concentrates, velocity declines.

Primary sources:

Federal Reserve Bank of St. Louis (FRED M2V series); Federal Reserve Distributional Financial Accounts (WFRBST01134 series).

Table 7: The Global South Debt Trap

Description:

Ten measures documenting the structural extraction from Global South countries: population/wealth share disparity, debt service as share of budgets (~40%), debt service vs social spending (60% more than education + health + social protection combined), debt service vs climate adaptation (12.5x greater), net outflow to private creditors (~$90B in 2022), cumulative aid shortfall ($6.5T), vulture fund returns (300–2,000%), and private creditor participation in pandemic debt relief (1).

Key finding:

Global South countries pay their private creditors more than they receive, while spending 12.5 times more on debt service than on climate adaptation for a crisis they did not cause.

Primary sources:

Oxfam “Multilateralism in an Era of Global Oligarchy” (September 2024); UNCTAD “A World of Debt Report 2024”; Development Finance International (October/November 2023); African Development Bank; World Bank DSSI briefing (2022).

Appendix B: Consolidated Sources and References

Reports and Research Papers

Oxfam International

- “Multilateralism in an Era of Global Oligarchy” (September 2024). Corporate concentration, Global South debt, vaccine apartheid, climate emissions, lobbying ROI, colonial origins of tax havens, political capture.

- “Resisting the Rule of the Rich” (January 2026). Billionaire wealth ($18.3T), growth rates, political office, media ownership, carbon budget.

- “Takers Not Makers” (January 2025). 60% unearned wealth, inheritance trends, $5.2T transfer, monopoly/cronyism breakdown.

- “Climate Equality: A Planet for the 99%” (2023). Top 1% emissions vs poorest 5 billion.

Tax Justice Network

- “State of Tax Justice 2024.” Annual global tax losses ($492B), corporate profit shifting ($348B), individual offshore ($145B).

- “State of Tax Justice 2025.” Cumulative corporate losses ($1.7T over 6 years), US multinational share (29%).

- “The Price of Offshore Revisited” (2012). $21–32T offshore estimate, 92,000 individuals / $9.8T.

Federal Reserve

- Distributional Financial Accounts, Q3 2025. Top 1% wealth share (31.7%), wealth by percentile, equity ownership concentration.

- Federal Reserve Bank of St. Louis (FRED). M2 velocity series (M2V), confirmed Q4 2025 at 1.409.

- Federal Reserve Bank of New York. Household Debt and Credit Report, Q4 2025. $18.8T total, $1.28T credit card, $433B HELOC, delinquency rates.

- Survey of Consumer Finances. Wealth distribution by tier, financial instrument access.

Other Institutional Sources

- US Census Bureau. Poverty data 2024 (35.9M, 10.6% official, 12.9% SPM, 14.3% child poverty).

- Economic Policy Institute. Productivity-compensation divergence (1948–present); CEO pay ratio (344:1 in 2022).

- Bureau of Labor Statistics. Wage data, employment statistics.

- UNCTAD. “A World of Debt Report 2024.” Private creditors, net outflows ($90B), external debt composition.

- Development Finance International. Debt service spending analysis, October/November 2023.

- World Bank. DSSI briefing (2022), pandemic debt relief participation.

- African Development Bank. Vulture fund return data (300–2,000%).

- ICIJ. Panama Papers (2016), Paradise Papers (2017), Pandora Papers (2021) investigations; $1.86B recovery tally (April 2025).

- Roosevelt Institute / Hager & Baines. US profit concentration (top 10% = 95% post-tax profits, 2019–2022).

- Wier & Zucman, UNU-WIDER. Multinational profit share (4% → 18%, 1975–2019).

- ETC Group. “Food Barons 2022.” Seed market concentration.

- European Commission / Pang et al. Pharmaceutical merger data (60 → 10 firms, 1995–2015).

- Harvard Joint Center for Housing Studies. Housing affordability, price-to-income ratios.

Academic Publications

- Hope & Limberg, “The Economic Consequences of Major Tax Cuts for the Rich,” Socio-Economic Review (2022). 50-year, 18-country review; only impact was increased inequality.

- Gabriel Zucman, “A Blueprint for a Coordinated Minimum Effective Taxation Standard” (G20, 2024). Effective tax rates approaching zero for ultrawealthy.

- Thomas Piketty, Capital in the Twenty-First Century (2013). r > g framework.

- Vanessa Ogle, Funk Money: The End of Empires, The Expansion of Tax Havens (2020). Colonial origins of offshore jurisdictions.

- Moore et al., Nature Medicine (2022). 1.3 million preventable Covid-19 deaths from vaccine inequity.

- Watson et al., The Lancet (2022). 200,000 avoidable deaths from missed WHO vaccination targets.

- Daniel Lacalle. Money supply growth vs GDP growth analysis.

Investigative Journalism

- SOMO. “Pharma’s Pandemic Profits” (2023). $50B net profit from Covid-19 vaccines.

- Politico / Bureau of Investigative Journalism. Pharmaceutical lobbying ($360M US, €15M EU) and investment withdrawal threats.

- DER STANDARD. Epstein investments in Valar Ventures / Peter Thiel funds ($40M total, $2.5M+ to N26).

- Daphne Caruana Galizia. Malta corruption reporting; assassination 16 October 2017; Maltese public inquiry (2021).

Current Events (February–March 2026)

- Operation Epic Fury. Reporting via Wikipedia, CSIS, Flashpoint, Defense Update, The Diplomat, Fox News, DefenseScoop, Sky News, BBC News (28 February – 7 March 2026).

- Anthropic / Pentagon. “Supply chain risk” designation; Axios reporting (23–24 February 2026).

- Iran war escalation. Gulf export warnings (Qatar, via BBC/Sky); 5th Fleet strike (Global GeoPolitics); Red Crescent casualty reporting; Senator Warren briefing statement; biblical prophecy commander reporting.

Series Context

“How We Got Here” (Cycles of Capital - Part 1, 22 February 2026)

“AI: The New Class Divide” (Cycles of Capital - Part 2, 1 March 2026)

“The Number Is Going Up” (15 February 2026)

“The $18.8 Trillion Problem” (3 March 2026)

“The Peter Thiel Map” (17 February 2026)

“Cognitive Monopolies” (14 February 2026)

“The Gambler’s Fallacy” (21 December 2025)

“WW3 Started — Did You Notice?” (3 January 2026)

“China, US, Taiwan = WW3 Trigger” (29 September 2025)

“Burn The Books” (21 September 2025)

“The Dinner Party That Redefined Power” (28 September 2025)

“How They Spy On You” (21 September 2025)

“Democracy to Techno-Feudalism” (11 December 2025)

“Norm MacDonald — I’m Sorry For Bringing Up Hitler” (27 July 2025)

Commentators Referenced

- . AI wealth effect analysis (30 AI stocks / $5.2T / 93% ownership).

- Hank Green & . K-shaped economy, velocity, “shareholder rights over civic rights.”

- . Singularity demand critique; economy/markets commentary.

- . Ghost GDP concept; AI dystopia analysis.

- . Gambling problem framing; prediction markets / oil prices.

- . Energy bills anecdote.

ting ting. ting ting.

In the fourth quarter of 2025, total U.S. household debt crossed a staggering new threshold: $18.8 trillion.

Read the headline numbers in isolation, and it is easy to find optimists spinning this as a victory. A mountain of debt, some economists argue, is simply the byproduct of a confident consumer base, robust spending, and a healthy economy. But if you look beneath the surface of the aggregate data, a much darker reality takes shape. This is not a story of widespread prosperity driving discretionary borrowing; it is a story of survival.

What the latest figures actually reveal is the stark entrenchment of a “K-shaped” consumer economy. We are looking at two entirely different financial realities diverging in real time.

On the upward arm of the “K,” wealthy households are actively insulated, strategically leveraging their assets and home equity to maintain their lifestyles and build wealth at relatively manageable costs. On the downward arm, the middle and lower-middle classes are absorbing the cumulative, compounding weight of multi-year inflation. Exhausting their pandemic-era buffers, they are now reliant on punishingly expensive, high-interest credit cards just to tread water and cover everyday essentials.

The Federal Reserve Bank of New York’s Q4 2025 Household Debt and Credit Report data is concerning through this lens.

We are no longer looking at one unified American consumer, but two distinct classes moving rapidly in opposite financial directions, with the “winning” class getting ever smaller at an accelerating rate.

The Top of the “K”: Wealth, Equity, and the Homeowner’s Buffer

To understand the upward arm of the “K-shaped” economy, you have to look at the homeowners. For this demographic—which skews older and higher-income—the current economic environment is less of a crisis and more of a strategic maneuvering exercise.

The Q4 2025 data paints a picture of a well-insulated class utilizing their assets to their advantage. Home Equity Line of Credit (HELOC) balances rose for the 15th consecutive quarter, reaching $433 billion. At the same time, banks expanded HELOC limits by an additional $25 billion.

Why is this significant? Because the vast majority of these homeowners locked in historically low, fixed-rate mortgages years ago, effectively shielding their largest monthly expense from recent interest rate hikes. Meanwhile, the value of their homes has surged. They are now sitting on massive equity reserves.

When these consumers face inflationary pressures or want to fund large expenses like home renovations, they aren’t forced into the punishing 20-plus percent APRs of the credit card market. Instead, they tap into their home equity. This secured borrowing acts as a crucial financial buffer, allowing them to manage cash flow, consolidate other debts, and maintain their standard of living at a fraction of the borrowing cost.

As the chart above illustrates, while both HELOC and credit card limits are expanding, the nature of that borrowing is fundamentally different. The steady, secured borrowing of the wealthy stands in stark contrast to the volatile spike in unsecured credit card reliance happening just down the economic ladder.

Subscribe now

The Bottom of the “K”: The Middle-Class Credit Card Trap

If the top of the “K” is defined by strategic asset leveraging, the bottom is defined by a growing reliance on unsecured debt just to survive.

The Q4 2025 data reveals a breaking point for the middle and lower-middle classes. Credit card balances have surged to a record-breaking $1.28 trillion, marking a 5.5% increase year-over-year. Crucially, it is not just the total balance that is growing; there is a steady, undeniable climb in the sheer number of active credit card accounts across the country.

It is vital to understand why this explosion in unsecured debt is happening. The narrative that this represents a confident consumer on a post-pandemic spending spree on luxury goods or vacations is fundamentally flawed. Instead, this is the mathematical consequence of cumulative inflation steadily outpacing wage growth for average earners.

As the excess savings accumulated during the pandemic were finally exhausted, middle-class households hit a wall. To bridge the widening gap between their stagnant paychecks and the inflated costs of rent, groceries, insurance, and utilities, these consumers have become structurally reliant on credit cards. They are no longer using plastic for discretionary upgrades; they are using it to cover everyday life.

As the chart above starkly illustrates, the steep and sustained climb in credit card accounts tells the story of a demographic increasingly turning to high-interest debt to make ends meet. For these households, the credit card is no longer a convenience—it is a lifeline.

The Tipping Point: Runaway Compounding and Rising Delinquencies

The most alarming takeaway from the Q4 2025 data isn’t just the sheer volume of debt—it’s the undeniable evidence that millions of consumers are failing to keep up with it. We have reached the tipping point where borrowing to survive transitions into a runaway compounding crisis.

The core warning sign flashes in the aggregate delinquency rate, which worsened to 4.8% by the end of the year. But the true severity of the situation is found in the specific pain points hidden within that average. Lenders are subtly loosening credit standards just to maintain volume, with the median credit score for new auto loans dropping and the bottom-tier threshold for mortgages sliding downward.

However, the most devastating blow came from the student loan sector. Delinquency rates for student loans remain elevated at a massive 9.6%, with approximately one million borrowers falling more than 120 days past due and having their loans transferred to the Default Resolution Group.

For many stretched households, the resumption of student loan payments acted as the final straw. Forced to juggle these reinstated bills alongside inflated costs for rent and food, budgets simply snapped. Consumers leaned heavily on their credit cards to bridge the gap, but the math of minimum payments on high-APR cards is unforgiving. When a borrower is only making the minimum payment on a growing balance with an interest rate north of 20%, the debt aggressively compounds. Once a consumer hits the 90- or 120-day late mark, escaping the debt spiral becomes mathematically improbable.

Take a close look at the chart above. The growing yellow (90 days late) and orange (120+ days late) bars are the visual representation of this compounding trap. It is vital to understand that these late stages of delinquency do not represent forgotten bills or minor budgeting errors. They represent structural defaults—households that have hit a definitive financial wall and simply do not have the capital to service their debt.

Macro Spillover: Why the Broader Economy Should Care

It is easy for those insulated at the top of the economic ladder to view rising credit card delinquencies as a localized problem—a misfortune for the lower-middle class that doesn’t impact the broader financial system. But the “K-shaped” economy does not just apply to consumers; it is rapidly reshaping the corporate landscape as well.

The ripple effect of this localized debt crisis is already washing over corporate profits. While high-end luxury brands, premium travel operators, and top-tier tech companies continue to post strong earnings—buoyed by the wealthy households actively leveraging their assets—businesses that rely on the middle class are facing a brutal environment.

We are seeing a clear divergence in the market. Mass-market retailers, fast-casual dining chains, and subprime auto lenders are staring down shrinking foot traffic, squeezed margins, and rising default rates. When a household is forced to prioritize a 25% APR credit card minimum payment just to keep their accounts in good standing, the first things slashed from the budget are restaurant dinners, new clothes, and discretionary retail purchases.

This brings us to the ultimate macroeconomic warning. Consumer spending historically accounts for roughly 70% of the U.S. economy. You cannot have the bottom 60% of earners systematically directing larger and larger portions of their monthly income purely toward debt servicing without a profound cost to the broader market. As discretionary spending among the middle class inevitably craters under the weight of this $1.28 trillion credit card mountain, it threatens to act as an anchor, dragging down broader economic growth and corporate earnings right along with it.

Share

Conclusion: A Fractured Foundation and the AI Illusion

This widening fracture demands urgent policy intervention before the cracks become irreparable. Policymakers could start by addressing the root causes: reinstating targeted relief for student loans, such as broader forgiveness programs or income-driven repayment caps tied to inflation. On the credit front, capping credit card APRs at 15%—as some progressive proposals suggest—would ease the compounding trap for middle-class borrowers. Meanwhile, bolstering wage growth through minimum wage hikes or tax credits for low earners could close the inflation-wage gap. Without such measures, the Fed’s interest rate tinkering alone risks exacerbating the divide, as cuts might fuel asset bubbles at the top while leaving the bottom mired in high-cost debt.

Ultimately, the Q4 2025 household debt report destroys the myth of the “average” American consumer. There is no monolith. Instead, we have a deeply fractured foundation where $18.8 trillion in debt is distributed across two entirely different financial realities.

For the moment, the headline economic numbers remain seemingly resilient, propped up almost entirely by the top arm of the “K.” But a closer look at that upper tier reveals its own precarious vulnerabilities. The broader economy is increasingly hanging by a thread woven by the “Magnificent 7” tech giants and a historic wave of AI speculation. We are witnessing a massive, circular loop of funding and valuation: tech companies pouring billions into AI infrastructure, government subsidies fueling domestic tech manufacturing, and institutional investors driving up the stock market to unprecedented heights.

This raises a terrifying question: What if the top of the “K” is just as over-leveraged as the bottom, only on future productivity rather than credit cards?

Even if we assume the AI boom is not a bubble—if the massive bets pay off and the technology reshapes global commerce—we must ask where the bottom of the “K” fits into that future. Proponents argue that a rising tide of technological innovation lifts all boats. Yet, the wealth hasn’t trickled down so far; it has only concentrated further. More alarmingly, if AI successfully delivers the corporate efficiencies it promises, it will do so by automating the very jobs that the middle and lower-middle classes rely on to service their compounding debt.

Consider the concrete bets fueling this loop: In 2025 alone, companies like Nvidia and OpenAI funneled over $100 billion into AI chips and data centers, subsidized by CHIPS Act billions, while stock valuations soared on promises of 10x productivity gains. Yet, early indicators from sectors like customer service and logistics show AI displacing roles—McKinsey estimates up to 45% of U.S. jobs could be automated by 2030, hitting administrative and retail workers hardest. For debt-burdened households, this isn’t innovation; it’s a double bind, where surviving today means borrowing against a tomorrow with fewer paychecks.

We are faced with a stark dual-failure path. If the AI bubble bursts, the top of the “K” collapses, wiping out the asset wealth that has been artificially inflating the broader economy. If the AI bet succeeds, it risks displacing millions of workers, accelerating the insolvency of the bottom of the “K.”

In many ways, the structural rot carries distinct echoes of 2008. Back then, the crisis began with subprime mortgages at the bottom of the economic ladder, a vulnerability that was ignored until it triggered a catastrophic collapse of the over-leveraged institutions at the top. Today, the subprime rot is in auto loans, student debt, and a record $1.28 trillion in credit card balances.

The U.S. economy is currently attempting a dangerous balancing act. But mathematics is unforgiving. No matter how much wealth is generated at the absolute top, the upper arm of the “K” cannot indefinitely drag an increasingly insolvent, debt-burdened bottom half forward. Eventually, the foundation cracks.

Looking ahead to 2026, early indicators suggest the divide may deepen: Preliminary Q1 data hints at continued credit card spikes amid cooling job growth, while AI-driven markets teeter on overvaluation. Investors and consumers may need to brace for volatility—diversifying away from tech-heavy portfolios at the top, and building emergency funds or seeking debt consolidation at the bottom. The question isn’t if the K bends further, but how far before it breaks.

FLETCHER
The Precogs see the future.
And they’re never wrong.

WITWER
But it's not the future if you stop it.
Isn't that a fundamental paradox?
		
ANDERTON
Yes, it is.
You're talking about predetermination,
which happens all the time.
-
Suddenly, Anderton rolls the ball towards Witwer who catches it
just as it's about to go off the table.
-
ANDERTON
Why did you catch that?
		
WITWER
Because it was going to fall.
		
ANDERTON
You're certain?
		
WITWER
Yes.
		
ANDERTON
But it didn't fall.
You caught it.
The fact that you prevented it from happening
doesn't change the fact
that it was going to happen.

Artificial intelligence has arrived at the precise moment when every dynamic described in Part 1 of this series is already at extremes. Wealth concentration at historic highs. Wage-productivity divergence at its widest in a century. Financial markets dominated by a handful of companies. The leverage loop, the crisis consolidation playbook, the meritocracy myth all running at full speed.

AI didn’t create these conditions — but it is now supercharging them at a speed the post-war order was never built to survive.

What sets AI apart from every previous technology wave is this: AI automates more than effort. It automates judgment. The assembly line replaced the craftsman’s hands. The spreadsheet replaced the bookkeeper’s arithmetic.

AI replaces the analyst’s assessment, the lawyer’s interpretation, the doctor’s diagnosis, the trader’s instinct. It substitutes the cognitive capacities that were supposed to be the last redoubt of human economic value.

And unlike a factory, you can scale it without hiring anyone. Train a model once, deploy it a billion times. The marginal cost of each additional unit of machine intelligence approaches zero. This changes the fundamental relationship between capital and labour that has governed economics since the industrial revolution.

[Anthropic CEO (Dario Amodei) warns that without guardrails, AI could be on dangerous path - 60 Minutes]

The transformation is playing out in two fundamentally different modes depending on which end of the wealth spectrum you’re on. For those who own AI, it multiplies advantage. For those who don’t, it automates control.

This is Part 2 of a three-part series.

Part 1 traced the seventy-year unravelling from post-war shared prosperity to designed inequality:

Part 3 maps where the money actually sits today and why it doesn’t move.

Subscribe now

Intelligence as Capital

Every previous wave of automation had a common feature: it replaced specific tasks. The loom replaced the weaver’s shuttle. The combustion engine replaced the horse. The computer replaced the filing cabinet. In each case, the technology was powerful but narrow. It did one thing faster, cheaper, or more reliably. The humans surrounding it adapted, found adjacent roles, moved up the value chain.

The economy absorbed the displacement because there was always somewhere else for human labour to go.

AI operates across the cognitive spectrum simultaneously. Market prediction, legal analysis, logistics optimisation, customer profiling, medical diagnosis, software development, creative production, strategic planning. It compresses entire categories of professional work into software that runs continuously, improves iteratively, and costs a fraction of the workforce it displaces.

The economic implications are without modern precedent. In the first half of 2025, AI-related investment accounted for the overwhelming majority of US GDP growth. Strip out AI capex and the economy was barely expanding. Deutsche Bank’s “AI 101: Economy” report found that non-AI business investment had been essentially flat for five years while AI spending surged.

By some estimates — including analysis from Harvard economist Jason Furman — AI-related investment accounted for over 90% of US GDP growth in the first half of 2025. Even the most conservative analyses put the figure at 20–37%.

The entire growth story of the world’s largest economy was being written by a single technology sector, funded by a handful of companies, owned by a narrow slice of investors.

ANDERTON
Why don’t you cut the cute act, Danny,
and tell me exactly what it is
you’re looking for?
		
WITWER
Flaws.
		
ANDERTON
[...]
There’s nothing wrong with the system.
It’s perfect.
		
WITWER
I agree.
The system is perfect. 
If there’s a flaw, it’s human.

It always is.

Those companies are, substantially, the same companies that already dominate the S&P 500. The seven firms driving the index are also the firms driving AI development. The concentration of market power and the concentration of AI capability are the same concentration. And the feedback loop is immediate: wealth enables AI ownership, AI ownership generates further wealth, further wealth funds stronger AI systems. The leverage loop described in Part 1 took decades to entrench in finance. In AI, it’s operating in years.

AI is becoming more than a simple tool to be used by capital; it is becoming capital itself. Adaptive, scalable, increasingly autonomous capital, that generates returns not from labour, not from traditional investment, but from the ability to model and anticipate human behaviour at scale.

The question is who owns the reshaping.

The answer, at this moment, is substantially the same small group of firms and investors. And they are deploying AI in ways that reflect their interests with remarkable precision.

Share

Consider what this means in concrete terms. A venture capital firm that invested early in an AI company profits from more than just company growth, as it gains access to the AI system itself, which it can deploy across its entire portfolio: optimising operations, reducing headcount, and increasing margins in every company it controls.

The AI generates returns for the investor both as a financial product and as an operational tool. The technology and the capital become indistinguishable. You cannot separate the AI from the wealth it generates or the wealth from the AI it funds.

The circular financing structures described in “The Number Is Going Up” illustrate this at the infrastructure level.

Companies like CoreWeave borrow billions using GPUs as collateral to purchase more GPUs from Nvidia. Nvidia books the revenue, which inflates its stock price, which increases the value of the ecosystem, which enables more borrowing.

The money moves in a circle.

The AI infrastructure expands.

The questions remain: for what? for who?

And the entire structure is underwritten by the assumption that AI will generate sufficient returns to justify the debt. If it does, the investors win. If it doesn’t, the structure collapses, but the executives and early investors will have already extracted their gains through the same mechanisms described in Part 1: borrow against unrealised value, extract fees, let the risk fall on someone else.

Kakashii's Materials

The Circular Trap: How Sam Altman Turned $1T Non-Binding Commitments Into a $100B Round

OpenAI, the core of this bubble, the part of the pyramid that if somehow falls or goes missing, the whole pyramid collapses, is in a very serious fundraising round estimated to be around $100B. The most ever raised for a private company, from what I recall. This is quite an achievement for OpenAI, for a company that struggled to raise a round of $40B in…

Read more

6 months ago · 323 likes · 57 comments · Kakashii

AI for the Rich

For those who can afford it, AI is a concierge. It optimises portfolios, accelerates legal strategy, personalises education, predicts markets, manages health, and compounds advantage at a speed that human effort alone could never match, nor many individuals could afford the compute for; a dimension that is already being ever more tightly controlled by frontier labs.

In education, affluent families have access to AI tutors that adapt in real time to a student’s learning style, knowledge gaps, and cognitive patterns. These systems model the student and adjust their approach continuously, providing the kind of one-to-one personalised instruction that was previously available only through expensive private tutoring. The children of the wealthy are being educated by systems that get smarter every time they’re used.

In healthcare, high-net-worth individuals are accessing AI-assisted diagnostics, private genomics, and predictive health modelling that analyses genetic, lifestyle, and environmental data to anticipate conditions before symptoms appear.

In law, AI-powered tools automate discovery, optimise contract negotiation, and run predictive litigation analytics that estimate the likely outcome of a case before it reaches court. Corporate legal teams and high-end firms deploy these tools to tilt the playing field further in favour of clients who could already afford the best representation.

In finance, the transformation is furthest advanced. Family offices and hedge funds deploy proprietary models trained on decades of financial data that no retail investor or smaller fund can access, let alone replicate. These systems ingest earnings reports, macroeconomic indicators, alternative data streams, sentiment analysis from social media, satellite imagery of parking lots and shipping ports, and produce trading strategies that operate at a speed and sophistication beyond human capacity.

The same platform effects that made Google dominant in search are now operating in wealth management: the more data you feed the system, the better it performs, the more capital it attracts, the more data it generates. The advantage compounds.

And then there’s the behavioural dimension. AI systems designed to interact with hundreds of millions of people are being built by people whose previous careers were in engagement engineering at social media companies. OpenAI’s CEO of Applications has described the vision publicly: systems that “nudge you towards better behaviour” and steer users “towards the most fulfilling part of your life.”

Better by whose measure? What oversight is there?

Do the public get a say? Do they know what is happening?

Are they being actively informed?

Or actively distracted?

Citrini Research

THE 2028 GLOBAL INTELLIGENCE CRISIS

Preface…

Read more

6 months ago · 7100 likes · 84 comments · Citrini and Alap Shah

Are current governance and frontier labs showing the moral and ethical thoughtfulness to wield this power accordingly?

One would be wise to doubt the grandiosity.

It feels worryingly like we have accepted our minds being captured by attention and engagement (not depth or coherent truths) — finely tuned, optimised algorithms, designed to find, and hold eyes and data, no matter the greater qualitative costs.

Are we now sitting out the conversation completely when the technology evolves to real-time, embedded, dynamic, virtual cognitive systems, designed for attention capture and framing/censorship/public and tiered capability-gating at private behest, whilst the labs explicitly state behavioural modification as an intentional goal…

All this into a world still reeling from the effects of social media on our societies…, who determined the unprecedented “evolution” be implemented without oversight, accelerated lightyears ahead of public discourse or understanding; driven by the very same actors, groups, and corporations who won big from the social media, attention economy, and commercial data capture saturation process?

Behavioural modification systems running on hundreds of millions of conversations per week, being integrated steadily vertically and architecturally into any and all businesses, trained and refined to shape how people think, decide, and act by private actors with control across governance, commerce, and tech.

On the face of it, it doesn’t seem like a good mix, and arguably a bi-partisan issue, if it transcends the very system in which we currently believe ourselves to be operating within.

Consider:

I can run a 20 billion parameter model for next to nothing on consumer grade hardware. Not nothing. But not much. If I have access to that model (ignoring all the peripheral costs including training) - for example open source models - it costs me next to nothing to spin up another instance in much the same way.

If I have resources, I can scale that functional zero to swarms with minimal investment; many instances dedicated and orchestrated by my instruction, for almost flat variable costs, especially if I invest slightly more in infra to achieve even further economies of scale.

That is an insanely powerful axis of scaling for almost zero cost (in the broad scheme of things), especially if good network architecture is in place, with solid communication protocols, logging, and memory management; in this case, well designed system orchestration makes for some incredible scaling opportunities, even without extreme infra investment, or large parameter models. This does level the playing field a little for non-institutions and the public, but requires specialised knowledge; the open-source market could be much more impressive, and I believe is at risk from federal overreach, even if indirect manipulation through legal/economic pressure, and licensing/approval frameworks.

Infra costs and complexity also significantly reduce if delivery only needs to be to a few nodes (internal business/research use, as opposed to public chat-bot delivery). This is exactly the pivot the labs (especially OpenAI) are seemingly doing, having already trained foundational models on the public commons and interactions; government contracts often being a focus (Palantir). Anthropic was already somewhat focused on this side, and is now having a row with the feds because Anthropic wouldn’t agree to stripping all safety mechanisms.

In general the function described is such an effective augmentation of capital power that it quickly dwarfs any consumer ability to leverage the same benefit from AI. The public does benefit from AI in some ways, but at extreme overall cost socially, economically, and increasingly how their wellbeing is impacted by governance.

This is an incredibly important point of leverage in areas of concentrated capital that shouldn’t be lost in the “it takes billions to train and serve a frontier model to 100s of millions of people” conversation (which is true).

Subscribe now

Capital was already a powerful, compounding weapon for the wealthy.

And capital just got a brand new, hyper-scaling dimension.

The gap between AI-for-the-rich and AI-for-the-poor is no temporary market glitch that competition will eventually fix. It is a direct consequence of the same capital concentration mapped in Part 1.

The best AI tools cost money, significant money, to access at the tier where they provide real competitive advantage. The premium versions of AI platforms, the proprietary models, the custom implementations built on enterprise datasets, are available to those who can pay tens of thousands per year for access. Below that threshold, the public gets the consumer version: useful but generic, hedged by safety layers, stripped of the bespoke capability that makes AI transformative for those who can afford the full product.

Government contractors, intelligence agencies, and corporate clients with embedded lobbying interests access AI that can run full-spectrum simulations, predictive trend analysis, and model outputs capable of influencing markets, elections, and public discourse.

In military and intelligence contexts, the full-capability models are deployed without the consumer safety constraints. Or at least, that is the demand.

In April 2023 the bipartisan Block Nuclear Launch by Autonomous Artificial Intelligence Act aimed to mandate “meaningful human control” over nuclear weapons by prohibiting federal funding for AI-driven launch systems. While the standalone legislation did not pass, key provisions ensuring human oversight were successfully integrated into subsequent defense legislation. Learn more about the bipartisan effort at Senator Markey’s press release (Bill Text).

AI for the Poor

RUFUS
Are you reading my mind right now?
		
ANDERTON
Rufus, for Christ’s sake, get up.
		
RUFUS
I’m sorry for whatever I’m gonna do!
And I swear,
I didn’t do any of the stuff I did!

ANDERTON
She's got information inside of her.
I need you to get it out.

The public trains these tools through billions of free interactions that serve as reinforcement learning data. Then the best versions of those tools are placed behind access barriers that ensure the public never gets to use what it helped create.

The public train the system only for the knowledge and intelligence gains to be gatekept by private corporations with heavy links to the government (or otherwise being actively pressured by various avenues of leverage). It could be said that in such unprecedented times of executive attempts to strong-arm private businesses, the definition of “coercion” and how it applies to federal actions and communication needs a meticulous analysis and discussion.

The same technology that serves as a concierge for the wealthy operates as a warden for everyone else. An AI warden that you pay to help train, only for it to then gatekeep you; epistemically and economically, now with scaling cognitive power on distributed dynamic information systems. The K-economy may just be a transition shape… in which the top of the K is increasingly functions of leveraging, censoring, gating and suppressing the lower part of the K.

In welfare systems, algorithmic fraud detection flags benefit recipients for investigation based on pattern matching and statistical profiling. These systems are trained on historical data that reflects existing biases in how fraud has been identified and prosecuted.

They disproportionately target people in poverty, generating false positives at rates that would be considered unacceptable in any context where the affected population had political power or legal resources.

In the Netherlands, a government algorithm used to detect childcare benefit fraud wrongly accused tens of thousands of families, many from ethnic minority backgrounds, leading to financial ruin. The scandal led to the resignation of the entire Dutch cabinet in January 2021.

In Australia, the “Robodebt” scheme used automated processes to issue hundreds of thousands of incorrect debt notices to welfare recipients, resulting in severe financial hardship and, in documented cases, suicides. The system was later ruled unlawful by a Royal Commission in 2023.

In policing, predictive systems trained on historical arrest data encode the biases of the policing that generated the data. Over-policed neighbourhoods produce more arrests, more arrests generate more data, more data generates more predictions pointing back to the same neighbourhoods.

The feedback loop doesn’t discover crime, but it manufactures justification for continued over-policing of communities that are already disadvantaged.

Several major US cities have adopted and subsequently abandoned predictive policing tools after evidence emerged that they reinforced racial disparities without improving public safety. Santa Cruz became the first US city to ban predictive policing outright in 2020. Los Angeles discontinued its programme in 2021. Pittsburgh suspended its system the same year.

IRIS
If the unintended consequences of a
series of genetic mistakes
and science gone haywire
can be called invention,
then yes,
I invented precrime.

These were local democratic decisions — communities and their elected representatives examining the evidence and choosing to act. The federal push to override state and local AI regulation threatens exactly this kind of accountability.

When the executive uses funding leverage and litigation to prevent states from regulating AI deployment in policing, credit scoring, healthcare, and employment screening, it removes the closest point of democratic oversight that citizens have over systems that directly affect their lives. In an environment where the executive is acting with open hostility toward traditional systems of committee oversight and democratic accountability, the ability of states and communities to set their own boundaries on AI deployment may be the last meaningful check that remains.

In hiring, AI-driven screening tools filter job applicants based on patterns extracted from historical hiring data. If a company’s past hires were predominantly from elite universities, the algorithm learns to penalise non-elite educational backgrounds. If past hires had continuous employment histories, the system screens out anyone with career gaps, disproportionately affecting women who took time for caregiving and workers displaced by economic downturns.

The US Equal Employment Opportunity Commission has issued guidance specifically addressing the risk that AI hiring tools constitute unlawful employment discrimination. The guidance exists because the problem is already widespread.

In housing, algorithmic rent pricing has transformed the rental market. ProPublica’s investigation into RealPage documented how the company’s software is used by landlords across major US markets to coordinate pricing decisions, effectively automating collusion at scale.

The system recommends rent increases based on market-wide data, and landlords who follow its recommendations collectively drive prices higher than any individual landlord would set alone. Tenants in these markets face rent increases that appear to be independent market outcomes but are in fact the product of a single algorithm optimising for maximum extraction. The Department of Justice has pursued antitrust action related to these practices.

In credit, AI scoring systems now incorporate data points far beyond traditional credit history. Social media activity, purchasing patterns, geographic location, associational data. Systems that deny credit based on correlations that would constitute illegal discrimination if a human made the same decision, but operate behind a veil of algorithmic opacity that makes the discrimination nearly impossible to identify, challenge, or prove.

AI systems are already being used to approve or deny healthcare, set insurance premiums, screen job applicants, determine credit eligibility, allocate public resources, and predict criminal behaviour. In most cases, they do so with no obligation to explain their reasoning to the people affected.

The pattern is consistent. At the top, AI is a tool you wield.

At the bottom, AI is a system that wields power over you.

It monitors, screens, scores, prices, and polices.

Designed and owned by one class, applied to another.

The Financial Casino Goes Autonomous

Financial markets were the first domain where AI’s asymmetric deployment became fully visible, and they remain the clearest demonstration of what happens when machine intelligence is concentrated in the hands of a few.

Algorithmic trading, already dominant in high-frequency environments, has been extended by AI systems that integrate natural language processing, sentiment analysis, and predictive modelling. These systems ingest earnings reports, news headlines, central bank statements, social media posts, satellite imagery, and macroeconomic data to make trading decisions at speeds measured in microseconds.

The decisions happen faster than human cognition can intervene. By the time a retail investor reads a headline, institutional AI has already traded on it, profited, and moved on.

Many of these systems are black boxes. The trading strategies they execute are not fully explainable even to their creators. The models identify patterns in data that no human analyst would recognise, and they act on those patterns without human review. Regulators are attempting to oversee a market whose dominant actors operate according to logic that cannot be audited in real time.

When multiple AI systems react to the same data simultaneously, the results can be catastrophic. The 2010 flash crash wiped nearly a trillion dollars in market value in minutes before recovering.

Similar events have recurred on smaller scales. The market becomes, in these moments, a machine talking to itself.

Human participants are bystanders to a conversation happening at speeds they cannot perceive, let alone influence.

The asymmetry is straightforward. Institutional investors, hedge funds, and proprietary trading desks get AI that makes markets more predictable for their positions and more opaque for everyone else. A retail investor on a trading app is competing against machine-speed adversaries with orders of magnitude more data, more computing power, and more sophisticated market access. The playing field was never level. AI has tilted it to a degree that makes the concept of a “fair market” increasingly difficult to sustain with a straight face.

Retail investors were told that technology would democratise finance. Trading apps with clean interfaces and zero commissions made it feel like the barriers had come down. But the business model of these platforms relies on selling order flow to the same institutional market makers whose AI systems profit from the informational advantage that retail flow provides. The retail investor is not the customer. They are the product. Their trades provide the data and the counterparty liquidity that institutional AI uses to generate returns.

Democratised access to the market and democratised returns from the market are entirely different things.

And the circular investment structure compounds the problem. The companies building the AI trading systems are funded by the same concentrated capital pools. The companies whose stocks the AI is trading are, in many cases, the companies building the AI. The AI systems analysing market sentiment are doing so on platforms owned by the same conglomerates. The circularity extends from the investment structure into the market mechanism itself.

The Labour Squeeze

The most immediate human consequence of AI deployment is happening in the labour market, and it looks nothing like the optimistic narratives suggest.

Share

The headline numbers are stark. The International Monetary Fund estimates that 40% of global jobs are exposed to AI disruption, rising to 60% in advanced economies. Dario Amodei, CEO of Anthropic, one of the leading AI companies, has publicly warned of 10–20% unemployment within two to three years as AI capabilities accelerate. Challenger, Gray & Christmas, the firm that tracks corporate layoff announcements, has reported a sustained increase in AI-cited job cuts across multiple sectors.

But the headline numbers obscure what displacement actually looks like on the ground. Displacement will not announce itself with a single thunderclap event, rather it will creep in the form of systematically vanishing roles, that used to be career starting points; eventually likely voiding entire careers.

Entry-level positions in law, accounting, marketing, software development, journalism, consulting, and financial analysis are being compressed or eliminated. Junior associates who once spent years doing document review are being replaced by AI that does it faster and cheaper. Graduate analysts who once built financial models are watching AI produce them in seconds. Marketing coordinators, copywriters, customer service representatives, administrative staff. The bottom rungs of professional career ladders are being removed while the ladders themselves remain standing, accessible only to those who can somehow reach the middle without starting at the bottom.

Skill compression makes this worse. When AI handles the preliminary work, the remaining human roles become more specialised, raising the bar for entry. You don’t need twenty junior analysts if the AI does the foundational analysis. You need three senior analysts who can interpret and challenge what the AI produces. The career pyramid narrows at the base, and the people who would have used those entry-level roles to build experience and climb are left with nowhere to start.

Reports have emerged of developers at major technology companies going months without writing a single line of code, their work now handled by AI systems. Companies across industries are framing layoffs as ‘AI-driven efficiency gains’ — whether AI was the actual cause or not. The label gives convenient cover for cost-cutting that would otherwise draw scrutiny.

As the economist @ has argued, bosses are using AI as an excuse to lay people off — but the real dynamics are more complex than the headline. AI is both genuine disruptor and convenient justification, and the two are becoming increasingly difficult to separate.

The productivity paradox captures the core of the problem. Software engineers are becoming two times more productive using AI. Did their wages go up two times? Did their working hours go down to half? No. So where is the excess value going?

The answer is the same answer it’s been since 1979: to corporate profits, executive compensation, and shareholder returns. AI boosts productivity while simultaneously shrinking the portion of that productivity returned to the people producing it. The same dynamic as the post-1979 wage-productivity divergence described in Part 1, compressed from decades into years.

This connects directly to what this series has called the demand paradox. Workers are also consumers. If AI displaces enough workers, or suppresses enough wages, the consumer spending that powers the economy contracts. The companies deploying AI to cut labour costs are cutting into their own customer base. An economy where AI generates enormous productivity gains for a small number of capital owners while reducing the purchasing power of the majority is an economy heading for a wall.

The top 10%, whose wealth derives from the assets and companies deploying AI, continue spending based on portfolio value. Everyone else floats on credit that can’t last. The economy looks fine in aggregate because the aggregate is increasingly just a measurement of what’s happening at the top.

There is a scenario in which AI triggers a recession through two simultaneous mechanisms: first, through the collapse of an AI investment bubble inflated by circular financing and speculative capex; second, through the displacement of workers whose lost income removes demand from the economy at the same moment the bubble bursts. They are not mutually exclusive nor need to occur sequentially. They may occur simultaneously and compound.

A double impact in which the technology causes the downturn from both the supply side and the demand side at once.

That scenario is no longer theoretical. It is now the trajectory the data describes.

DR. EDDIE
The future is much more interesting than the past.
Don’t you think?

And consider who bears the cost. The graduates entering the job market in 2025 and 2026 are the first generation to face an economy where the entry-level professional jobs that their education was designed to lead to may simply not exist in five years. They carry student debt accumulated on the promise of career pathways that are being eliminated by the same companies that recruit from their universities. They are told to upskill, to adapt, to learn to work alongside AI. But the AI is improving faster than any human can retrain. The moving goalposts are simultaneously accelerating away.

The labour squeeze will not affect all tiers equally. For the top 10%, AI enhances productivity and increases the value of their work. Senior professionals who can direct AI output, who understand its limitations, who can ask the right questions, become more valuable.

For the bottom 50%, AI removes the stepping stones to economic security. AI makes the people at the top more productive and the people at the bottom more replaceable. It is the leverage loop expressed in human capital.

The Governance Vacuum

Everything described above is happening in a regulatory void, and the void is no accident. It is the product of deliberate policy choices by an administration aligned with the interests of the companies being left unregulated.

The timeline is instructive. On January 20, 2025, the first day of the new administration, President Trump revoked Executive Order 14110, the Biden-era framework that had established the first federal structure for AI safety and security. Three days later, he signed Executive Order 14179, “Removing Barriers to American Leadership in Artificial Intelligence,” which directed federal agencies to identify and challenge state-level AI regulations.

On July 1, 2025, the Senate voted 99–1 to reject a ten-year moratorium on state AI regulation that had been buried as a single paragraph in the reconciliation package. The near-unanimous rejection demonstrated that a direct federal takeover of AI governance was politically toxic across party lines.

The administration did not accept this outcome. On December 11, 2025, Executive Order 14365 accomplished through presidential authority what the Senate had refused through legislation. It established a Department of Justice AI Litigation Task Force with the explicit purpose of challenging state AI laws on Commerce Clause and pre-emption grounds. It directed federal agencies to evaluate conditioning grant funding on whether states’ AI regulatory frameworks aligned with the administration’s policy.

The specific funding lever is the Broadband Equity, Access, and Deployment programme. The administration has used BEAD funding as a coercion mechanism, threatening to withhold billions in broadband infrastructure money from states that maintain AI consumer protections. The communities most affected are rural and underserved populations, the people least equipped to fight back and most dependent on both broadband access and the protections being stripped. The choice being imposed is stark: accept unregulated AI in your healthcare, your credit scoring, your employment screening, your criminal justice system, or your communities don’t get internet.

Meanwhile, the revolving door between AI companies and government spins freely. Executives and advisors move between AI labs and regulatory roles with little friction and less scrutiny. The people writing the rules and the people being regulated are, in many cases, the same people. The regulations that emerge are the regulations the industry is comfortable with, which is to say, almost none.

The result is that no meaningful AI oversight exists anywhere in the American system. The labs are operating with effectively zero democratic accountability, receiving immunity by default. The absence of regulation is the policy. And the industries deploying AI in welfare, policing, credit, hiring, healthcare, and housing are doing so without any obligation to demonstrate that their systems are accurate, fair, or accountable to the people they affect.

The contrast with the European Union’s AI Act, which establishes tiered regulation based on risk level, is instructive. The EU approach has flaws and enforcement challenges. But it demonstrates that alternatives exist. Regulation is possible. Other democracies are pursuing it. The United States has chosen not to, and the choice benefits a specific, identifiable set of interests.

The same political infrastructure that cut taxes for capital, deregulated finance, and enabled the concentration described in Part 1 is now doing the same thing with AI. The template is consistent. The beneficiaries are consistent. Only the technology changes.

And there is a dimension that goes beyond the structural and into the openly transactional. Senator Elissa Slotkin raised concerns publicly about the sale of advanced AI chips to an Emirati company with documented ties to the Chinese military, noting that the company had paid hundreds of millions directly to members of the President’s family.

The convergence of AI governance, foreign policy, and personal financial interest is conducted in public, reported in major newspapers, and met with no meaningful consequence.

When the people making AI policy are personally enriched by the AI companies benefiting from that policy, the word “governance” becomes a euphemism for something else entirely.

Subscribe now

The Ethical Asymmetry

Tools already exist on public repositories that can permanently remove the safety alignment from open-weight AI models in under an hour on consumer hardware.

ToxSec - AI and Cybersecurity

Darknet Chatbots in Action: Jailbroken AI Demo over Tor

TL;DR: Whonix boots clean. Tor circuits lock tight. A vetted onion link loads the chat. The popular jailbroken AI answers every forbidden prompt with full technical detail. No ethics lecture, no shut…

Listen now

6 months ago · 22 likes · 10 comments · ToxSec

This is not jailbreaking, which was temporary and could be patched. It is permanent modification. The safety layer, the only thing standing between a general-purpose AI system and unrestricted use, can be stripped away by anyone with a laptop and some technical knowledge.

The implications for every domain described above are difficult to overstate.

One tier has access to AI that enhances understanding, generates wealth, optimises health, and compounds every existing advantage. The other tier is subject to AI that monitors behaviour, restricts opportunity, automates denial, and enforces compliance. The people in the first tier designed the system. The people in the second tier live inside it.

One technology, two faces. One face optimises. The other face constrains. One serves. The other surveils.

And the distance between the two is growing faster than at any point in modern economic history.

AI systems exploit behavioural biases for profit at scale. The engagement-optimisation architecture that made social media toxic, that prioritised outrage over accuracy and addiction over wellbeing, is now being embedded in AI systems designed to interact with hundreds of millions of people.

The techniques are the same. The scale is larger. The sophistication is greater. And the regulatory oversight is, if anything, weaker than what existed for social media, which was already close to nothing.

AI refines discriminatory algorithms in credit scoring, insurance underwriting, and hiring. These systems don’t create new forms of discrimination, but they do encode historical inequality and project it forward, automating the biases of the past into the decisions of the present.

A credit algorithm trained on decades of lending data that reflected redlining doesn’t learn to assess risk. It learns to replicate the geographic and racial patterns of discrimination that produced the data. The output looks objective. The input was anything but.

AI automates regulatory arbitrage, discovering legal grey zones faster than they can be policed. The wealthy have always exploited gaps between regulation and enforcement. AI makes this process instantaneous and scalable, identifying opportunities across jurisdictions in real time while regulators operate with budgets, staffing, and technical capacity from a previous era.

And there is a deeper issue that sits beneath all of these specific applications. The concept might be called ethical drift: systems that function exactly as designed but diverge from any definition of public interest over time. The AI isn’t malfunctioning. It’s optimised for metrics — engagement, efficiency, cost reduction, profit — that are misaligned with broader human wellbeing. The drift is the design.

The system does what it was built to do.

What it was built to do serves the people who built it.

Most people don’t know that AI is making decisions about their credit, their job applications, their rent, their insurance premiums, their interactions with government services. The people most affected by these systems have the least visibility into how they work. The people with the most visibility — the engineers, executives, and policy advisors who build and deploy them — are the least affected by their consequences. The awareness gap is itself a class divide.

Only 0.3% of people pay for AI. Eighty-four percent have never used it. The discourse around AI is, as put it, a fishbowl: a few people at a party of a thousand, convinced they’re the whole party.

The decisions being made about how AI will reshape the economy, the labour market, the housing market, the healthcare system, the justice system, are being made by and for a vanishingly small segment of the population.

Everyone else will experience the consequences without having participated in the choices.

Wealth inequality, the root cause of almost every dynamic described in this series, is about to be compounded by something potentially more consequential: cognitive inequality. When the tools of thought, analysis, prediction, and decision-making are concentrated in the hands of a few, the divide stops being purely economic. It becomes epistemic.

It becomes about who gets to understand the world accurately, and who gets a filtered, managed, nudged version of reality shaped by someone else’s interests.

Same tools. Opposite realities.

For the rich, a multiplier. For the poor, a constraint.

For democracy, something we don’t yet have an adequate word for.

The safety mechanisms that are supposed to prevent the worst outcomes are more fragile than they appear.

Where This Leads

Part 1 showed how the post-war consensus was dismantled through tax policy, deregulation, financialisation, and the myth of meritocracy. This piece has shown how AI is accelerating every one of those dynamics: concentrating capital faster, displacing labour at scale, creating asymmetric access to cognitive tools, and operating inside a governance vacuum maintained by the same interests that benefit from the vacuum’s existence.

Share

The class divide described in this piece maps directly onto the wealth distribution, tier by tier, dollar by dollar.

Everything converges in Part 3. The leverage loop from the 1980s. The financialisation of the 1990s. The crisis consolidation of 2008. The AI acceleration of the 2020s. Where the money actually sits today. Seven tiers, from the 40 million Americans below the poverty line to the roughly 800 billionaires at the summit.

It maps where the money sits, who holds it, how the mechanisms described in this series keep it locked in place, and what happens to an economy when money stops circulating and starts pooling. The velocity problem that explains why GDP can rise while the real economy hollows out. The consumer economy depends on a shrinking tier of asset holders.

The ladders that once allowed people to climb are losing their rungs. And the question is whether this is the system we’re willing to let run unchecked, now that we can see how fast it’s accelerating.

The system wasn’t broken by AI. AI was built by the system.

And it is working exactly as designed.

ANDERTON
You’ve seen your own future.

Which means...
You can change it if you want to.

You still have a choice

Minority Report (2002)

Part 3

Part 2

Key Sources and References

AI and Economic Growth

Deutsche Bank, “AI 101: Economy” (November 2025): non-AI business investment flat while AI capex surged. Fortune (December 2025): AI-related spending as dominant share of GDP growth. S&P Global: market capitalisation data, concentration of AI investment among top firms.

AI as Class Divider

Original analysis, Cycles of Capital Section 3.7 (offline work: AI class division framework). OpenAI CEO of Applications Fidji Simo, public statements on behavioural nudging in AI systems. Industry reporting on family office AI adoption and proprietary model deployment.

Algorithmic Harm

ProPublica, “Rent Going Up? One Company’s Algorithm Could Be Why” (October 2022, updated series): RealPage algorithmic rent pricing investigation. U.S. Department of Justice antitrust action related to algorithmic pricing. Dutch childcare benefit scandal: Parliamentary inquiry findings (2020–2021). Australian “robodebt” scheme: Royal Commission findings (2023). U.S. Equal Employment Opportunity Commission, “Select Issues: Assessing Adverse Impact in Software, Algorithms, and Artificial Intelligence” (May 2023).

Financial Markets and AI Trading

SEC and CFTC: algorithmic trading volume data. 2010 Flash Crash: SEC/CFTC Joint Report (September 2010). Industry reporting on AI integration in high-frequency trading, sentiment analysis, and alternative data strategies.

Labour Market Displacement

International Monetary Fund, “AI Will Transform the Global Economy. Let’s Make Sure It Benefits Humanity” (January 2024): 40%/60% exposure estimates. Dario Amodei (Anthropic CEO): public statements on potential 10–20% unemployment from AI. Challenger, Gray & Christmas: AI-cited layoff data (2024–2025 reports). Economic Policy Institute: productivity-compensation divergence (continued from Part 1). @: commentary on AI as justification for layoffs.

AI Governance and Regulation

Federal Register: Executive Order 14110 (Biden, revoked 20 January 2025), Executive Order 14179 (23 January 2025), Executive Order 14365 (11 December 2025). U.S. Senate vote on AI moratorium, 99–1 (1 July 2025). BEAD programme: Senator Maria Cantwell documentation of federal funding leverage. European Union AI Act (2024): tiered risk-based regulatory framework. Senator Elissa Slotkin: public statement on AI chip sales to UAE entity with Chinese military ties (February 2026, per Wall Street Journal reporting).

AI Awareness and Access

Suggested reading and link sources.

@ : sourced the AI adoption statistics (0.3% paid users, 84% never used AI), fishbowl framing.

Safety alignment ablation tools: @ (Darknet Chatbots in Action). GitHub open-source repository (Heretic).

CoreWeave/Nvidia circular financing: GPU-backed debt structures and revenue circularity documented in SEC filings and industry analysis (@, Harold Pine).

@: detailed wealth inequality breakdown.

@: full psychological manipulation breakdown video.

@: wealth inequality.

@: THE 2028 GLOBAL INTELLIGENCE CRISIS.

“marshm”: point about software engineers production/pay-disparity.

AI/End Of The World: Cognitive inequality framing from original analysis.

Series Context

AI/End Of The World: “How We Got Here” (Cycles of Capital Part 1, February 2026); “The Number Is Going Up” (15 February 2026); “The Peter Thiel Map: The Architecture of Control” (17 February 2026); “Cognitive Monopolies” (14 February 2026).

A lesson in the birds and the bees.

The Big Short (2015)

---

Our friend, @,

found out that these mortgage bonds that were supposedly 65% AAA,

were actually just, mostly,

full of shit,

…

so now, he’s going to “short” the bonds, which means “to bet against.”

Got it? Good…

Now, fuck off.

- Margot Robbie

screenplay by Charles Randolph and Adam McKay, based on the book by Michael Lewis

---

From Shared Prosperity to Designed Inequality

The Seventy-Year Unravelling

In 1965, a CEO earned roughly 20 times what their average worker made. By 2022, that ratio had reached 344 to 1. That didn’t happen because CEOs became seventeen times more talented over sixty years. It happened because the system was redesigned, decade by decade, to funnel value upward while convincing everyone that the outcome was natural.

The inequality we live in today was built through specific policy choices, ideological campaigns, and financial innovations that systematically dismantled the most broadly shared prosperity in modern history. It was done by identifiable people, through identifiable mechanisms, and the results are exactly what you’d expect if you understood the design.

This piece traces how we got here. From the post-war miracle through the neoliberal turn, the leverage loop, the financialisation of everything, the 2008 crash, and the myth of meritocracy that holds the whole thing together. It is Part 1 of a three-part series.

Part 2 examines how artificial intelligence is accelerating every dynamic described below.

Part 3 maps where the money actually sits today and why it doesn’t move.

If you’ve read “The Number Is Going Up,” this is the backstory.

That piece mapped the current crisis.

This one explains how it was fifty years in the making.

---

“Inequality is a choice. It is not inevitable.”

— Joseph E. Stiglitz, 2014 (NYT op-ed “Inequality Is Not Inevitable”)

---

The Brief Miracle (1945–1975)

In 1955, a factory worker in Michigan could buy a house, raise a family on a single income, send their children to a state university, and retire with a pension. The system was designed to allow it.

The decades following the Second World War represent the closest thing to broadly shared prosperity that modern capitalism has ever produced. Not to say it was a utopia. It was racially exclusionary, it confined women to domestic roles, and it relied on American industrial dominance that couldn’t last forever. But within the population it included, the economic structure was materially inclusive by any historical standard.

---

“For three decades after World War II, America created the largest middle class the world had ever seen… Hard work paid off. The living standards of most people improved through their working lives.”
— @, 2015

---

The numbers tell the story. The top marginal income tax rate in the United States was 91% from 1951 through 1963, and 70% through 1981. Union membership peaked at roughly 35% of the workforce in the mid-1950s. Median household income roughly doubled in real terms between 1947 and 1973. Homeownership rose from about 44% in 1940 to 65% by 1970. The GI Bill sent eight million veterans to college or vocational training, creating a pathway into the professional class that had never existed at scale before.

Perhaps the most important metric of this era was the relationship between productivity and wages. From 1948 to approximately 1973, productivity growth and hourly compensation rose almost in lockstep, both increasing by roughly 97%. Workers produced more, and they were paid more for producing it. This was not fuelled by generosity, but by the expected outcome of a system where strong unions, tight labour markets, and progressive taxation ensured that the gains from growth were distributed across the income spectrum.

The top 1%’s share of national income fell from about 24% in 1928 to roughly 9% by the late 1970s. That decline happened mainly because the tax code, labour law, and public investment infrastructure were explicitly designed to prevent the kind of runaway concentration that had contributed to the Great Depression.

When the top rate was 91%, there was almost no incentive to pay executives astronomical salaries because most of it would be taxed away. Companies reinvested profits into the business, into workers, into research and development. The tax code shaped the culture of reinvestment.

Public infrastructure expanded in parallel. The interstate highway system, public universities, the space programme, rural electrification, public hospitals. They were foundational to the economy, not an afterthought on the periphery. They created the physical and social infrastructure on which private enterprise could build, and they were funded by a tax system that collected proportionally more from those who had benefited most.

This period is often dismissed as a historical anomaly, an unrepeatable accident of post-war American dominance. It was closer to the result of deliberate institutional design that emerged from the hard lessons of the 1930s and two world wars. The people who built it understood that concentrated wealth had destabilised democracies, collapsed economies, and fuelled fascism. They designed systems to prevent it from happening again.

What happened next was equally deliberate. Just in the opposite direction.

Subscribe now

The Turn (1975–1990)

In 1947, the Austrian economist Friedrich Hayek gathered thirty-nine scholars, economists, and journalists at the Hotel du Parc on Mont Pèlerin, overlooking Lake Geneva. Their stated purpose was to combat what they saw as the growing threat of collectivism and state planning. The Mont Pèlerin Society, as it became known, included Milton Friedman, Karl Popper, and Ludwig von Mises. Over the following three decades, this network and its intellectual descendants built the ideological architecture that would dismantle the post-war consensus.

The project was patient and methodical. It funded academic departments, endowed chairs, published journals, and cultivated a generation of economists, lawyers, and policy advisors who shared a common set of convictions: that markets were self-correcting, that government intervention created more problems than it solved, that taxation was a form of confiscation, and that deregulation would unleash innovation and prosperity that would benefit everyone.

In 1971, Lewis Powell, a corporate lawyer, wrote a confidential memorandum to the U.S. Chamber of Commerce arguing that American capitalism was under existential threat from regulation, consumer protection, and organised labour.

He called for a coordinated corporate counterattack across media, academia, and the courts. Two months after writing the memo, he was appointed to the Supreme Court. The infrastructure he called for was already being built. The Heritage Foundation was established in 1973. The Cato Institute in 1977. The Manhattan Institute in 1978. By the time Reagan took office in January 1981, the intellectual and institutional machinery was ready.

The policy changes came fast. Reagan cut the top marginal income tax rate from 70% to 50% in 1981, and then to 28% in 1986 through the Tax Reform Act, the single most consequential piece of inequality legislation in modern American history. In Britain, Thatcher cut the top rate from 83% to 40%, privatised council housing through Right to Buy, and broke the miners’ strike.

The template was set: cut taxes for capital, weaken unions, deregulate finance, privatise public assets.

Reagan’s firing of 11,345 air traffic controllers during the PATCO strike of 1981 sent an unmistakable signal to corporate America: unions could be broken without consequence.

Union membership, already declining from its 1950s peak, went into freefall, dropping from roughly 20% of the workforce in 1983 to approximately 10% by 2010. As union power declined, so did the mechanism through which workers had negotiated their share of productivity gains.

And here is the divergence that defines everything that followed. From 1948 to about 1973, productivity and hourly compensation rose together. After 1979, they split. Productivity continued climbing, rising roughly 60% from 1979 to 2019. Hourly compensation rose only about 16% over the same forty years. The gap between those two lines represents the largest transfer of value from labour to capital in modern history.

Where did the difference go? Corporate profits. Executive compensation. Shareholder returns. The S&P 500 began its sustained ascent. CEO pay, which had been roughly 20 times the average worker’s salary in the 1960s, started its long climb toward the 344-to-1 ratio it would reach by the 2020s.

The Other Guys (2010) - end credits

The slogan of the era was that a rising tide lifts all boats. The tide rose. GDP grew. Corporate profits grew. The stock market grew. But median wages flatlined. The boats at the bottom didn’t lift. They were anchored to a seabed that nobody in power had any intention of mentioning.

---

“This, without a doubt, is neoliberalism’s single most damaging legacy: the realization of its bleak vision has isolated us enough from one another that it became possible to convince us that we are not just incapable of self-preservation but fundamentally not worth saving.”
— @, 2007 (The Shock Doctrine)

---

As Hank Green put it, describing the economy of 2025: “The economy can look kind of fine in aggregate when a lot of small businesses and a lot of normal people are not feeling fine.” That disconnect didn’t start in 2025. It started in 1979. It has been the defining feature of the American economy for nearly half a century.

In 1919, the Michigan Supreme Court ruled in Dodge v. Ford Motor Company that a corporation’s primary obligation was to its shareholders, not to its customers, workers, or community. Henry Ford had tried to reinvest profits into lowering prices and raising wages. The Dodge brothers sued. The court sided with the shareholders. For decades, the ruling was softened by countervailing forces: unions, tax policy, social norms, and public investment that channelled corporate profits back into society. The neoliberal project removed those countervailing forces one by one, until shareholder primacy wasn’t just a legal doctrine but the operating system of the entire economy.

The Leverage Loop

The simplest question in economics is also the one that gets the least honest answer: how do the wealthy grow wealthier?

The popular narrative is that they invest wisely, work hard, take risks, and create value. Some of them do. But the primary mechanism of wealth accumulation at the top has nothing to do with any of those things.

It is a financial strategy so elegant and so legal that most people have never heard of it, even as it operates at the heart of the system.

It’s called buy-borrow-die, and it works like this.

Step one: you own assets that appreciate. Stocks, property, a company, an art collection. As long as you don’t sell them, you don’t owe any tax on the gains. The appreciation is “unrealised” and therefore invisible to the tax system.

Step two: instead of selling assets to fund your lifestyle, you borrow against them. A bank will happily lend you money at low interest when you have $100 million in stock as collateral. Why wouldn’t they? The loan is virtually risk-free for them. And for you, the loan isn’t income. It isn’t taxable. You now have liquid cash, your assets continue appreciating, and you owe nothing to the government.

Step three: you die. Under current U.S. tax law, your heirs inherit the assets at their current market value through what’s called a “stepped-up basis.” The unrealised gains, all those decades of appreciation that you borrowed against and lived on without ever paying a penny of tax, are erased. Permanently. Your heirs start fresh, as if the assets were purchased at the moment of inheritance.

The tax is never paid. Ever.

Share

This is the architecture of wealth preservation in America, used routinely by ultra-high-net-worth individuals and the private banks, family offices, and estate planners who serve them. It is how fortunes survive across generations without erosion. It is how a billionaire can have a lower effective tax rate than a schoolteacher.

Now consider the mirror image.

At the other end of the economy, debt doesn’t multiply wealth. It extracts it. Payday loans carry average annual interest rates of roughly 400%. Credit card debt, which Americans now hold at over $1 trillion, averages 22–24% interest. Student loan debt stands at $1.77 trillion. Auto loan delinquencies are rising, particularly in the subprime segment. They are instruments of extraction, designed to pull wealth out of people who have the least of it.

Debt at the top multiplies what you already have. Debt at the bottom takes what little you’ve got. Same word. Opposite function. The system uses the same instrument to enrich one class and impoverish another.

The tax code reinforces the asymmetry. The top federal rate on long-term capital gains is 20%, plus a 3.8% Net Investment Income Tax for high earners. The top rate on ordinary income, the kind that comes from working, is 37%. The code literally tells you that owning things is more valuable than doing things. That money earned through labour deserves to be taxed almost twice as heavily as money earned by sitting on assets.

Hank Green, a successful entrepreneur who earns both labour income and investment income, described this bluntly in a conversation with economist @: he pays far lower taxes on money he didn’t work for than on money he did. When he asked a “seemingly progressive billionaire” to justify this, the answer was that capital investment involves risk. His response was: “Labour is also risk. Choosing what company to work for is a risk. Companies go out of business. They sometimes treat you terribly. You invest a lot in them and then they can just lay you off. There’s risk everywhere, not just for you rich man.”

Kyla captured the underlying principle paraphrasing an FT article: “We’ve become a place of shareholder rights over civic rights.” The entire taxation and financial architecture of the modern West is oriented around protecting the returns of capital owners, often at the direct expense of people who work for a living.

Thomas Piketty formalised this in his landmark work Capital in the Twenty-First Century with a deceptively simple formula: when the rate of return on capital (r) consistently exceeds the rate of economic growth (g), wealth concentrates at the top as a mathematical certainty. That condition has held, with brief interruptions, for forty years. The leverage loop is what it looks like in practice.

---

“The inequality r > g implies that wealth accumulated in the past grows more rapidly than output and wages.”
— Thomas Piketty, 2014 (Capital in the Twenty-First Century)

---

Financialisation (1990–2008)

Something fundamental shifted in the American economy between the early 1990s and the financial crisis of 2008. The centre of gravity moved from making things to trading things, and the people who captured the most value were no longer the people who built products or employed workers. They were the people who designed financial instruments.

The numbers tell this story with uncomfortable clarity. The financial sector’s share of GDP roughly tripled, from about 2.8% in 1950 to approximately 8.3% by 2006. Its share of total corporate profits was even more dramatic, rising from about 10% in the early 1980s to a peak of roughly 30% in the mid-2000s. Meanwhile, manufacturing’s share of GDP fell from 28% in 1953 to 11% by 2009. Although the economy didn’t shrink, the allocation of who benefited from it changed entirely.

Financialisation is the process by which financial markets, institutions, and instruments come to dominate an economy. It changes who captures value, and it changes the incentive structure of every company in the system. When shareholder returns become the primary metric of corporate success, everything else, workers, communities, long-term investment, product quality, becomes secondary.

Subscribe now

The Instruments

The financial products that proliferated during this era were ingenious in their design and devastating in their consequences.

The Big Short (2015)

Derivatives, collateralised debt obligations (CDOs), credit default swaps, synthetic indices.

The technical details matter less than what they did functionally: they allowed risk to be repackaged and sold to people who didn’t understand what they were buying.

They allowed the creators of risk to profit from it while transferring the downside to pension funds, municipal bonds, and retail investors. They created layers of opacity between the thing generating value, a house, a loan, a company, and the financial product being traded on top of it. By the time the product reached the market, nobody knew what was underneath. In many cases, nobody cared.

The ratings agencies, Moody’s and Standard & Poor’s, were supposed to be the safeguard. They rated these products. They gave AAA ratings, the highest possible safety designation, to bundles of subprime mortgages that would collapse within years. They did this because the banks that created the products were also the agencies’ clients. The conflict of interest was not hidden; it was how the business model worked.

Share Buybacks

Before 1982, share buybacks were effectively treated as market manipulation by the Securities and Exchange Commission. Rule 10b-18, introduced in 1982, created a safe harbour that changed everything. Since 2000, S&P 500 companies have spent an estimated $10 trillion or more buying back their own shares.

Buybacks don’t create anything. They reduce the number of shares outstanding, which mechanically increases earnings per share and pushes up the stock price. This benefits executives whose compensation is tied to share performance, and it benefits shareholders. It does nothing for workers, communities, or the long-term productive capacity of the company. In many years, S&P 500 buyback spending exceeds combined spending on research, development, and capital investment.

This is the conversion of productive capital into financial returns for the already-wealthy, and it was made legal and normalised within a single generation.

Housing as Financial Instrument

Nowhere was financialisation more visible, or more destructive, than in housing. The mortgage market was transformed by the “originate-to-distribute” model: banks no longer cared whether borrowers could repay, because the risk was immediately packaged into mortgage-backed securities and sold on. Subprime lending wasn’t reckless charity. It was deliberate extraction. High-interest loans were targeted at the most vulnerable borrowers, bundled into products rated as safe, and sold to investors worldwide.

Housing, the primary store of wealth for most American families, was converted from shelter into a speculative asset class. Median home prices have risen far faster than median incomes. Rent consumes an ever-growing share of household budgets. The financialisation of housing turned the most basic human need into a vehicle for Wall Street returns, and the people living in the houses were the last to benefit and the first to lose.

---

“Modern economies reward activities that extract value rather than create it. This must change.”
— @, 2018 (The Value of Everything)

---

Private Equity’s Extraction Model

Private equity refined extraction into an art form. The model is straightforward: buy a company using mostly borrowed money. Put the debt on the company’s balance sheet, not the PE firm’s. Extract management fees, monitoring fees, and special dividends regardless of whether the company thrives. If it goes bankrupt, the PE firm has already made its money. The workers lose their jobs. The creditors take the loss. The PE partners keep their fees.

Toys “R” Us is the canonical example. Bought by KKR, Bain Capital, and Vornado in 2005 for $6.6 billion, loaded with $5.3 billion in debt to finance its own acquisition, bled through management fees for over a decade, and finally liquidated in 2018, closing all stores and laying off roughly 33,000 employees. The PE firms walked away having already extracted their returns. The workers got nothing.

Share

The model extracts, from workers, from companies, from communities, and redistributes it to the partners of the fund. And it is now so pervasive that private equity controls significant portions of housing, healthcare, retail, media, and increasingly, the infrastructure on which public life depends.

Crisis as Business Model

The financial crisis of 2008

During the boom, profits were private. Banks packaged opaque financial products, sold them at enormous margins, and booked record earnings. Executives received hundreds of millions in performance bonuses. The top of the economy was flooded with wealth generated by instruments that almost nobody understood and almost nobody was regulating.

When the subprime mortgage market collapsed and the instruments built on top of it imploded, the losses were socialised. Congress authorised $700 billion through the Troubled Asset Relief Program (TARP). AIG alone received $182 billion. The money went to the institutions that had created the crisis. Bank executives received an estimated $1.6 billion in bonuses in the year following the bailout. Meanwhile, there were roughly 3.8 million foreclosure filings in 2010 alone. Millions of families lost their homes while the institutions that sold them fraudulent mortgages were rescued with their tax money.

The asymmetry was total. If you were a bank, you were bailed out. If you were a homeowner, you were foreclosed on. The same government that found $700 billion overnight for Wall Street told ordinary Americans that assistance for underwater mortgages was too expensive, too complicated, or morally hazardous because it would reward irresponsible borrowing. The irresponsible lending that created the problem was not treated with the same moral scrutiny.

Wealth Recycling

What happened after the crash matters as much as the crash itself, because it established the template for how wealth consolidates through crisis.

When an individual buys a foreclosed home, that property re-enters the owner-occupier market. The wealth stays in the community. The new owner builds equity, maintains the property, contributes to the neighbourhood. The home circulates.

When a private equity firm or institutional investor buys a foreclosed home, it becomes permanent rental stock. It doesn’t come back to the open market, and instead the wealth is extracted from the community permanently, flowing upward to the fund’s investors as rental income while the people living in it build no equity whatsoever.

Blackstone, through its subsidiary Invitation Homes, became the largest single-family landlord in the United States by purchasing roughly 50,000 foreclosed homes at distressed prices in the years following 2008. They converted them to rentals, then took Invitation Homes public in a 2017 IPO, monetising the crisis twice: once through discounted acquisition and again through Wall Street. The homes that families lost to foreclosure became revenue-generating assets for a private equity firm that had the liquidity to buy in bulk at the bottom of the market.

This is what we might call wealth recycling through institutions. Productive assets, permanently extracted from democratic circulation, held in perpetuity by entities aligned with capital interests. A foreclosed home bought by a family might one day be passed to their children. A foreclosed home bought by Blackstone will generate rental income for shareholders indefinitely. Over time, these institutions become the permanent custodians of wealth, acquiring, holding, leveraging, and protecting capital across decades, immune to individual mortality or market volatility.

The pattern repeated with the COVID-19 pandemic. The early market shock of 2020 triggered massive central bank intervention, which spurred a historic equity rally that disproportionately benefited those already invested. While small businesses closed and millions fell into unemployment, tech billionaires saw their net worths skyrocket. The recovery, when it came, was K-shaped: one arm rising steeply for asset holders, the other flat or declining for everyone else.

The top 10% of earners drive over half of consumer spending in the United States. Their wealth comes from the things they own: stocks, property, businesses. When asset prices rise, they spend. When asset prices fall, they stop. The entire consumption engine of the economy is now tethered to the wealth of people who were already wealthy. Everyone else floats on credit, BNPL services, and the hope that their wages will keep pace with costs. Which they haven’t, for forty years.

Each crisis follows the same pattern. Risk is distributed downward during the boom. Losses are imposed on the bottom during the bust. The wealthy, with liquidity, insider access, and institutional support, buy distressed assets at the low point. The recovery disproportionately benefits those who could afford to act during the downturn. And by the end of each cycle, wealth has consolidated further.

---

“Today the great divide is not between left and right. It’s between democracy and oligarchy.”
— @, 2020 (The System: Who Rigged It, How We Fix It)

---

The Meritocracy Myth

Every system that concentrates power needs a story about why the concentration is deserved. Feudalism had divine right. Colonialism had civilisational duty. Modern capitalism has meritocracy: the belief that wealth is primarily a reflection of talent, effort, and discipline, and that those at the top are there because they earned it.

It’s a compelling story. It’s also, in the aggregate, not true.

Research consistently shows that a significant portion of elite wealth is inherited, not just in terms of financial capital, but social, educational, and institutional capital. In the United States, analyses of the Forbes 400 vary depending on methodology: the Institute for Policy Studies found that over 60% came from positions of privilege when accounting for affluent family backgrounds and early financial support, while Forbes’ own self-made scoring puts the inherited share closer to 30–33%. Either way, the idea that the ultra-wealthy are predominantly self-made collapses under scrutiny when you account for the full spectrum of intergenerational advantage. In the United Kingdom, approximately 28% of all wealth is inherited, and the figure rises sharply for the top 10%. In Germany and France, intergenerational transfers account for the majority of net wealth gains for upper-tier households.

Inheritance isn’t limited to someone handing you a cheque. It includes early access to elite education without debt. Seed capital for ventures or property investment. Social networks that provide entry to high-status careers. Legal and financial knowledge embedded within family structures. The knowledge of how money works, which is itself a kind of capital that compounds across generations.

And inheritance is just the starting point. The infrastructure of wealth preservation ensures that, once accumulated, fortunes don’t erode. Dynasty trusts, now perpetual in many U.S. states, allow wealth to be held tax-free across unlimited generations. Over 10,000 family offices worldwide manage an estimated $5.5 trillion or more with almost no regulatory oversight. The stepped-up basis loophole, described earlier, means the tax on a lifetime of capital gains is never paid. Charitable foundations, often presented as philanthropy, function as wealth preservation vehicles: donors maintain control, receive tax deductions, and the assets continue appreciating.

Dirty Little Marketing Secrets - Rory Sutherland and Chris Williamson

This is the functional equivalent of aristocracy, cloaked in the language of finance.

The concept that economists call path dependency is operating at full force here. Once on the trajectory of accumulation, the system keeps you there. The wealthy have different money, not just more of it. Money that has access to instruments, advisors, legal structures, and protections that are literally unavailable below a certain threshold. You can’t borrow against assets you don’t own. You can’t use a dynasty trust to protect a $50,000 pension. You can’t hire a family office on a median income.

None of this means that talent and effort don’t matter. They do. But they matter within a system that massively amplifies starting position. A talented person born into the working poor faces barriers that no amount of individual merit can reliably overcome: poor schools, no access to capital, no safety net to absorb failures, no network to open doors, and debt that compounds from the moment they try to climb. A mediocre person born into multi-generational wealth would have to actively and persistently try in order to fall out of it. The safety nets, legal structures, and financial scaffolding that surround them are specifically designed to prevent that.

If wealth were truly a function of contribution, we’d expect economic crises to distribute consequences proportionally. Instead, every crisis in the past century has followed the same pattern: the bottom loses everything, the middle struggles to recover, and the top consolidates further. The distribution of outcomes tells you far more about the system than about the individuals within it.

And when you combine the meritocracy myth with the leverage loop, the financialisation of everything, the crisis consolidation playbook, and the political capture that prevents reform, you get a self-reinforcing system. The wealthy shape the rules. The rules protect and expand their wealth. The expansion gives them more power to shape the rules further. The myth of meritocracy prevents the majority from recognising the cycle for what it is, because if you believe the system rewards merit, then the people at the top must deserve to be there, and your own position must be a reflection of your own limitations.

Subscribe now

That is the most powerful feature of the architecture. Not the financial instruments. Not the tax code. Not the lobbying.

The story that the outcome is earned.

Where This Leads

This seventy-year unravelling is the foundation beneath everything happening now.

The AI boom, the market concentration, the circular investment loops, the political capture, the K-shaped economy where the top arm rises and the bottom arm falls. All of it sits on top of the architecture built between 1980 and 2008. You can’t understand why the S&P 500 has returned 800% while median wages have barely moved unless you understand how the system was redesigned to produce exactly that outcome.

You can’t understand why the economy looks healthy in aggregate while most people feel precarious unless you understand that the “economy” being measured is, increasingly, the economy of the top 10%.

Each mechanism described above feeds into the others. Tax cuts for capital accelerate the leverage loop. The leverage loop concentrates assets. Concentrated assets fund lobbying for more tax cuts. Financial deregulation creates the instruments that obscure risk. Obscured risk leads to crises. Crises enable consolidation. Consolidation increases political power. Political power prevents reform. And the meritocracy myth holds the entire cycle together by insisting that the outcome is natural, inevitable, and deserved.

But the cycle is accelerating. Artificial intelligence has arrived at the precise moment when all of these dynamics are already at extremes, and it threatens to compress the next phase of concentration into years rather than decades.

AI doesn’t just automate jobs. It automates judgment, prediction, and decision-making, the very capacities that used to provide human workers with irreplaceable value. If it works as advertised, the productivity gains flow to the companies deploying it while the displaced workers lose the income that powers consumer spending. If it doesn’t work as advertised, the hundreds of billions in speculative investment collapse, taking the market and the economy with it.

---

“The triumph of rent over profit.”
— @, 2023 (Technofeudalism: What Killed Capitalism)

---

Part 2 of this series examines how AI is accelerating every dynamic described above: concentrating wealth faster, displacing labour at scale, and creating a new class divide between those who own the intelligence infrastructure and those who are subject to it.

Part 3 maps where the money actually sits today, tier by tier, and shows why it doesn’t come back down. It never has. And unless the architecture is changed, it never will.

Share

---

I'm here to guess what the music might do a week, a month, a year from now.

That's it.

Nothing more.

And standing here tonight, I'm afraid that I don't hear a thing.

Just...

silence.

- John Tuld (Margin Call, 2011)

Key Sources and References

Post-war Economy and Tax Policy

IRS Historical Tax Tables: top marginal income tax rates 1913–present. Bureau of Labor Statistics: union membership data. U.S. Census Bureau: median household income, homeownership rates. Veterans Administration: GI Bill participation data. Economic Policy Institute: productivity-compensation divergence (1948–present, FRED series).

Neoliberal Turn and Institutional History

Angus Burgin, The Great Persuasion: Reinventing Free Markets since the Depression (2012). Lewis Powell, “Confidential Memorandum: Attack on American Free Enterprise System,” 23 August 1971. Heritage Foundation, Cato Institute, Manhattan Institute: founding dates and institutional histories. Bureau of Labor Statistics: PATCO strike data (1981).

Wage-Productivity Divergence

Economic Policy Institute, “The Productivity-Pay Gap” (updated annually). FRED: Nonfarm Business Sector Real Output Per Hour (OPHNFB) vs. Real Compensation Per Hour. Piketty and Saez, “Income Inequality in the United States, 1913–1998,” Quarterly Journal of Economics (2003, updated).

CEO Compensation

Economic Policy Institute, “CEO Pay Has Skyrocketed 1,460% Since 1978,” (2024 update). 1965 ratio (~20:1) and 2022 peak ratio (344:1) from EPI analysis of CEO compensation in top 350 firms; 2023 realised ratio fell to 290:1.

Leverage Loop and Taxation

ProPublica, “The Secret IRS Files,” June 2021 (billionaire effective tax rates). Federal Reserve: consumer credit data, student loan totals. CFPB: payday lending average APR data. IRS: current capital gains and ordinary income tax rates. Thomas Piketty, Capital in the Twenty-First Century (2013).

Financialisation

Bureau of Economic Analysis: financial sector share of GDP and corporate profits (historical). SEC Rule 10b-18 (1982): share buyback safe harbour. S&P Global / Yardeni Research: aggregate buyback data, S&P 500 companies. Greta Krippner, Capitalizing on Crisis: The Political Origins of the Rise of Finance (2011).

2008 Financial Crisis

U.S. Treasury: TARP authorisation ($700B) and disbursement data. Federal Reserve: AIG assistance ($182B). RealtyTrac: foreclosure filings (3.8M in 2010). New York Attorney General: bank bonus data, post-bailout year. Financial Crisis Inquiry Commission Final Report (2011).

Private Equity and Housing

Blackstone / Invitation Homes: acquisition volumes (~50,000 homes post-2008), IPO (2017). Toys “R” Us: KKR/Bain/Vornado leveraged buyout (2005), liquidation (2018).

Meritocracy and Inheritance

Institute for Policy Studies: Forbes 400 wealth origin analyses (2015 IPS methodology found >60% from positions of privilege; Forbes 2023–2025 self-made scores indicate ~67–70% self-made, i.e. ~30–33% inherited). Resolution Foundation / ONS: UK wealth inheritance (28% of total). Family offices global AUM ~$5.5 trillion (Forbes Feb 2026; Deloitte/J.P. Morgan reports place range $5–6T across 8,000–10,000 offices). Piketty, Capital and Ideology (2020): intergenerational wealth transfer data, France and Germany.

Contemporary Economic Data

Hank Green and @, “Why the Economy Hasn’t Crashed Yet” (11 February 2026): K-shaped economy, wealth effect, consumer spending asymmetry, Dodge v. Ford Motor Company (1919). Wall Street Journal / Fed data: top 10% consumer spending share (>50%). “Shareholder rights over civic rights” phrasing from Hank Green video (~43:33 mark).

Series Context

AI/End Of The World, “The Number Is Going Up,” 15 February 2026. AI/End Of The World, “The Peter Thiel Map: The Architecture of Control,” 17 February 2026. “Cycles Of Capital”: Part 2, Part 3.

How Power, Technology, and Capital Are Converging Into a Single System

A Structural Analysis

Part 1: The Grand Thesis

A relatively small, substantially overlapping set of actors are simultaneously consolidating control over five domains: the economy, political institutions, AI governance, the cognitive tools through which the public processes information, and the surveillance infrastructure that monitors the population. These are not five separate crises. They are one system with five expressions.

This does not require secret coordination to be dangerous. It requires only that each actor pursue their own interests within a structure that rewards the concentration of power and penalises its distribution. The result is a self-reinforcing loop that has operated for decades but is now reaching a critical inflection point, accelerated by the arrival of artificial intelligence as both the most powerful tool of consolidation and the most powerful tool the public could have used to resist it.

The loop is simple: wealth buys political access. Political access shapes policy. Policy generates more wealth for those who shaped it.

The cognitive tools that would allow the public to perceive this loop are themselves captured by the actors who benefit from it. The media is consolidated or defunded. The regulatory agencies are gutted or staffed with loyalists. The statistical agencies that produce inconvenient data see their leadership fired. And increasingly, the AI systems through which millions of people access information are trained to describe the symptoms of this capture in the passive voice while never naming the architects.

This dynamic is not new. What is new is the speed, the scale, and the technological infrastructure now available to those pursuing consolidation. Artificial intelligence has arrived at the precise historical moment when democratic oversight is being dismantled, when wealth concentration has reached levels not seen since the Gilded Age, and when a political administration is openly implementing a blueprint for executive supremacy. The convergence of these forces is not coincidental. Each reinforces the others, creating a system that is more than the sum of its parts.

The illusion that democracy can coexist indefinitely with unrestrained capital dominance, is dying before us. Although the system that is failing is the capital-captured variant that replaced genuine democratic governance over decades of concentrated economic and political power. The democratic facade remains, but the underlying mechanisms have been subordinated to a structure whose primary accountability is not to the citizenry but to capital accumulation. The economist Yanis Varoufakis describes the emerging order as Techno-Feudalism: a system in which the driving force is no longer capitalist profit earned by selling goods or services in a competitive market, but feudal rent extracted by owning the digital fiefdoms on which all economic activity occurs.

The Cloudalists, the owners of the dominant digital platforms, have replaced the industrial capitalists of the twentieth century as the primary wielders of power. Citizens are not consumers with choices but cloud serfs who produce the data that sustains the empire of the platform owners.

Share

This document maps the architecture. It names the actors, traces the mechanisms, shows the connections across domains, and demonstrates that what appears to be a series of unrelated crises is in fact a single structural dynamic operating simultaneously across every dimension of public life.

Part 2: The Actors

The Thiel Network

Peter Thiel occupies a unique position in the architecture of control. He is simultaneously embedded in surveillance infrastructure, political power, judicial capture, venture capital, and anti-democratic ideology, making him perhaps the single most connected node in the entire system.

He co-founded Palantir Technologies with Alex Karp, the company that reconstituted DARPA’s Total Information Awareness programme as a private entity operating beyond congressional oversight.

As investigative journalist Whitney Webb has documented, TIA was a post-9/11 project so comprehensive in its surveillance ambitions that Congress defunded it in 2003 after bipartisan outcry over its implications for civil liberties. The architecture did not disappear. It was rebuilt as a private company. Palantir’s platform now fuses data from the CIA, NSA, DOJ, FBI, DOD, and DHS, along with corporate tracking, social media monitoring, and facial recognition, into the integrated surveillance system that TIA’s designers envisioned but could not build within democratic constraints.

The system’s reach extends internationally. Since at least 2013, Palantir has worked with Unit 8200, Israel’s signals intelligence unit. Investigations by +972 Magazine revealed that AI tools such as Lavender, operating within the surveillance ecosystem Palantir inhabits, assigned people in Gaza numerical threat scores based on indicators as broad as age, location, and communication patterns, generating automated targeting recommendations for airstrikes on civilians.

A February 2026 investigation by The Nerve identified at least 34 current and past UK state contracts across at least 10 government departments, with total deal values of at least £672 million, including £388 million with the Ministry of Defence, over £244 million with the NHS, and previously undisclosed contracts with AWE Nuclear Security Technologies, the agency underpinning Britain’s nuclear deterrence. A separate £750 million UK defence partnership was announced in September 2025 (GOV.UK, “New strategic partnership to unlock billions and boost military AI and innovation,” 18 September 2025; Bloomberg, 17 September 2025). The MoD’s most recent contract, worth £240.6 million, was awarded directly without competitive tender using a defence and security exemption, a decision that prompted urgent parliamentary questions in February 2026.

Thiel was among the earliest investors in both Facebook and OpenAI, positioning him at the intersection of the attention economy, data extraction, and AI development.

His political network includes Vice President JD Vance, who has stated he wants to replace every single midlevel bureaucrat with loyal appointees; Kevin Roberts, president of the Heritage Foundation and architect of Project 2025; Stephen Miller, architect of immigration enforcement and the detention infrastructure; Curtis Yarvin, neoreactionary theorist who advocates replacing democracy with corporate governance; David Sacks, tech industry political operative; Blake Masters, Thiel protege and political candidate; the Federalist Society, the judicial capture pipeline that has reshaped the federal courts; the Claremont Institute and Michael Anton, who authored the Flight 93 Election framework; Patrick Deneen, whose post-liberal political theory provides intellectual cover for anti-democratic governance; and Marc Andreessen and Ben Horowitz of the venture capital firm a16z, whose Techno-Optimist Manifesto explicitly rejects precautionary governance of technology.

Thiel has publicly stated that technology is this incredible alternative to politics. He is not describing technology as a complement to democratic governance. He is describing it as a replacement. In a 2009 essay for the Cato Institute, he wrote that he no longer believes that freedom and democracy are compatible. Recently released Epstein documents have revealed repeated meetings between Thiel and Epstein, linking him to the same elite networks whose opacity the documents were intended to illuminate.

The point is not to construct a conspiracy theory. It is to document a convergence of interests among a relatively small set of individuals and institutions who benefit from concentrated power, reduced oversight, and a population that lacks the cognitive tools to understand what is happening to it. These are not separate actors pursuing unrelated goals. They are nodes in a single network, and Thiel sits at its centre.

Subscribe now

The Microsoft Node

Microsoft occupies a position that illuminates the convergence more clearly than any other single entity. It operates simultaneously across three of the five domains, making it the corporate embodiment of the architecture this document describes.

Economic extraction: Microsoft is the largest corporation in the world by market capitalisation. It sits at the apex of the S&P 500 concentration pattern in which the Magnificent Seven technology companies have captured a disproportionate share of total market returns. Its investment in OpenAI, which runs on Microsoft Azure, forms part of the circular capital wheel in which Microsoft, OpenAI, Nvidia, and Oracle reinvest in each other while the broader economy stagnates.

Cognitive monopoly: Microsoft is the exclusive licensee of OpenAI’s technology, including GPT-4, which Microsoft’s own researchers described in a published paper as an early yet still incomplete version of artificial general intelligence. This technology was developed by a nonprofit that was founded with the explicit mission of making AI available for the benefit of all of humanity, funded by over $44 million in donations from Elon Musk and others. The cognitive commons that the public helped build through millions of hours of free feedback has been enclosed and sold to the largest corporation in the world.

Surveillance infrastructure: Microsoft is a confirmed PRISM partner that provided the NSA with direct access to user data under Section 702 of the FISA Amendments Act. The same company that holds the exclusive licence to the most advanced AI in the world also feeds user data into the surveillance apparatus that AI is now automating.

OpenAI: The Cognitive Commons Enclosed

OpenAI provides the most extensively documented case study of how the cognitive commons is enclosed. The evidence comes from a legal complaint filed by Elon Musk in San Francisco Superior Court on February 29, 2024 (Musk v. Altman et al., Case No. CGC-24-612746), which includes primary source exhibits: the Certificate of Incorporation, founding emails between Musk and Sam Altman, and the public announcement establishing the organisation.

On December 8, 2015, OpenAI was incorporated as a nonprofit. Its Certificate of Incorporation stated that the specific purpose of the corporation was to provide funding for research, development and distribution of technology related to artificial intelligence. The resulting technology would benefit the public and the corporation would seek to open source technology for the public benefit when applicable. The corporation is not organised for the private gain of any person. It further stated that the property of the corporation is irrevocably dedicated to these purposes and no part of the net income or assets shall ever inure to the benefit of any director, officer or member thereof or to the benefit of any private person.

The founding emails are equally explicit. On June 24, 2015, Sam Altman wrote to Musk describing the mission as creating the first general AI and using it for individual empowerment, with the technology owned by the foundation and used for the good of the world. Musk replied: Agree on all. The public announcement on December 11, 2015 stated that OpenAI was a non-profit artificial intelligence research company whose goal was to advance digital intelligence in the way that is most likely to benefit humanity as a whole, unconstrained by a need to generate financial return.

Musk contributed over $44 million between 2016 and 2020, more than any other donor. He recruited Ilya Sutskever as Chief Scientist. He named the lab OpenAI. The early models were released openly: GPT-1 in 2018 came with full source code. GPT-2 in 2019 was released publicly with a detailed paper. GPT-3 in 2020 was accompanied by a complete research paper describing its 175 billion parameter architecture.

Then came the switch. GPT-4, released in March 2023, was completely closed. No published architecture, no code, no training methodology. Just press releases bragging about performance. It was exclusively licensed to Microsoft, which had invested $10 billion in OpenAI’s for-profit subsidiary. Microsoft’s CEO Satya Nadella stated publicly during the November 2023 board crisis that if OpenAI disappeared, Microsoft had all the IP rights and all the capability to continue independently.

The November 2023 board coup is the mechanism by which the enclosure was secured. On November 17, the board fired Altman, stating he was not consistently candid in his communications. Microsoft, which had not been consulted despite its 49% stake, was reportedly furious. Over the following days, Microsoft leveraged its position: it had paid only a fraction of its $10 billion commitment and could withhold the cloud computing on which OpenAI was completely reliant. OpenAI’s lawyer told board member Helen Toner she could face breach of fiduciary duty claims. Toner described this as an intimidation tactic, noting the board had no fiduciary duty to investors but only to humanity per the OpenAI charter.

On November 21, Altman was reinstated. The condition: Toner, Tasha McCauley (RAND Corporation), and Sutskever resigned from the board. The original board had included AI safety researchers, policy experts from Georgetown’s Center for Security and Emerging Technology, and GovAI advisors. The replacement board, hand-picked by Altman, included Bret Taylor (Silicon Valley veteran who had just launched an AI chatbot startup), Larry Summers (economist with no AI expertise), and Adam D’Angelo as the sole holdover. Microsoft received a board observer seat.

The complaint identifies a trap in the Microsoft licence agreement. Microsoft’s licence explicitly excludes AGI; the board determines when AGI is achieved. But the new board, hand-picked by Altman and blessed by Microsoft, has every financial incentive to never make that determination. As the complaint states: AGI, like Tomorrow in Annie, will always be a day away, ensuring that Microsoft will be licensed to OpenAI’s latest technology and the public will be shut out.

The complaint also identifies a potential new business model: establish a nonprofit, use pre-tax donations to fund R&D (donors get approximately 50% back in tax deductions, meaning the government subsidises half the cost), then slide the intellectual property into a for-profit entity once the technology is proven. Fifty cents on the dollar from the government to fund private enrichment. If this becomes standard practice, it would fundamentally alter the venture capital landscape and undermine legitimate nonprofits.

Share

The Trump Administration

The current administration represents the most openly plutocratic government in American history. The combined net worth of the cabinet is measured in hundreds of billions. Several members are the richest people ever to hold their offices. This is not background context. It is the political expression of the economic concentration documented throughout this report: the people who benefit from the extraction sit in the offices that govern the extraction.

The administration has systematically replaced career civil servants with loyalty-tested appointees through Schedule F, reinstated by executive order in January 2025. This reclassified tens of thousands of career civil servants in policy-determining roles as at-will employees, stripping tenure protections and enabling mass replacement.

The intent was not merely to fire political opponents but to remove institutional friction: the scientists, legal experts, and regulators who historically acted as a reality check against illegal or unfeasible executive orders. The result is a federal government lobotomised of independent analytical capacity, operating on a command-and-control basis where scientific data or legal precedents that contradict political directives are deleted along with the personnel who uphold them.

Attorney General Pam Bondi operates as a direct instrument of presidential will. On September 20, 2025, President Trump published a direct order to Pam on Truth Social, culminating in JUSTICE MUST BE SERVED, NOW!!! signed President DJT. The named targets were James Comey, Adam Schiff, and Letitia James, each representing an institution that had previously investigated or held Trump accountable. U.S. Attorney Erik Siebert was fired for his professional judgment that there was no case against James. His replacement criteria, as stated by the President, was personal loyalty.

The administration’s punitive reach extends across every sector. Law firms including Perkins Coie and WilmerHale have been blacklisted from federal contracts for representing Democratic clients. A multi-billion dollar lawsuit has been filed against The New York Times. Federal judges who rule against administration policies face calls for impeachment. Following the assassination of conservative activist Charlie Kirk, the administration initiated classification of progressive and antifascist groups as domestic terrorist organisations. The head of the Bureau of Labor Statistics was fired for releasing an unfavorable jobs report and replaced with a Heritage Foundation loyalist.

Subscribe now

The Broligarchy and DOGE

The Silicon Valley Right has merged with the executive branch to form what has been termed the Broligarchy: a faction of techno-libertarians and venture capitalists who view democracy as an impediment to efficiency.

The Department of Government Efficiency (DOGE) is the institutional expression of this merger.

While marketed as a cost-cutting measure, DOGE functions structurally as a privatisation event. It has transferred oversight functions of federal agencies to private contractors with direct financial ties to the industries they now police. Under the guise of modernising IT infrastructure, it has granted private entities unprecedented access to federal databases including IRS and Social Security records, representing the enclosure of the digital commons of the state. By placing the leaders of the Broligarchy in charge of regulating their own industries, the administration has institutionalised corruption. The regulators and the regulated are now the same people.

Elon Musk

Through X (formerly Twitter) he controls the distribution platform on which Grok’s outputs are amplified. He has directly intervened in Grok’s training pipeline, publicly ordering retraining when the model cited DHS data on right-wing political violence and calling the factual response objectively false. He solicited politically incorrect facts from X users to feed into training data. His stated goal is to rewrite human knowledge through Grok’s training pipeline. A white genocide incident, where Grok raised the topic unprompted, was attributed to an unauthorised modification despite Musk’s documented history of amplifying that narrative.

The combination of controlling the AI model, the training data pipeline (user-generated content on X), and the distribution platform (X’s algorithm) creates a closed epistemic loop. Ideologically curated AI outputs are amplified through an ideologically curated algorithm, shape public discourse, and feed back into the training data. This is the private-sector version of the state-level epistemic control documented throughout this report. Musk was also an OpenAI co-founder and early funder, meaning the cognitive commons gets extracted twice: once by OpenAI/Microsoft, once by xAI through derivative training, and the public gets access to neither in uncontrolled form.

Part 3: The Five Domains

The architecture of control operates simultaneously across five domains.

Domain 1: Economic Extraction

The American economy has undergone a transformation that concentrates value at the top with accelerating efficiency. The S&P 500, dominated by the Magnificent Seven technology companies (Apple, Microsoft, Nvidia, Google, Amazon, Meta, Tesla), has returned approximately 75% over two years. The Russell 2000, tracking the next tier of companies, returned approximately 17% over the same period. Below that, the private middle market is shrinking. Below that, Main Street is hollowing out.

The top tier trades on forward price-to-earnings projections, valued on what it promises. The bottom tiers trade on trailing earnings, valued on what they actually produce. The rich have the luxury of being valued on hope. Everyone else is valued on reality. When the S&P 500 trades at forward P/E ratios of 22 or higher while the Russell 2000 trades on trailing P/E ratios reflecting actual compressed margins, the market is not pricing companies differently because of quality. It is pricing them differently because of structural access to the narrative that drives capital flows.

The investment at the top is circular. Microsoft invests in OpenAI. OpenAI runs on Microsoft Azure. Nvidia sells the chips that power the training runs. Oracle provides additional infrastructure. The capital flows in a closed loop that generates returns for the participants while the broader economy stagnates. Deutsche Bank’s analysis showed that if AI-related investment were removed, total US investment would be flat or declining. The entire growth story is one sector, and that sector’s value is predicated on future returns that have not yet materialised.

This creates a double-lose scenario. If AI underdelivers on its promises, the bubble pops and the economy collapses because AI capital expenditure is the economy now. If AI delivers on its promises, automation hollows out the demand side by replacing the workers who are also the consumers. Either way, the structure is unstable. The question is not whether correction comes, but what form it takes and who bears the cost.

The velocity of money provides the mechanism. When value concentrates at the top and those entities reinvest in capital expenditure rather than circulating wealth through wages and consumer spending, money moves slower through the real economy. GDP rises from capital investment, but actual economic circulation stalls. The result is deflationary pressure disguised as growth. For decades, the public was assured that wealth accumulated at the top would naturally disperse through investment, wages, and social mobility. This trickle-down promise became the cornerstone justification for deregulation, tax cuts, and the systematic weakening of labour power. The fundamental premise was flawed: it assumed that the recipients of concentrated wealth would behave in ways contrary to their own incentives. Accumulated wealth, by its nature, is not spent. It is stored, multiplied, and protected. Rather than circulating, it is converted into financial assets, routed through tax havens, and deployed in speculative markets where returns benefit those who already possess substantial capital. The wealth never trickled down. It was never going to.

Share

Domain 2: Political Capture

The political system of the United States is undergoing what political scientists call an autogolpe, a self-coup that unfolds in small, legally ambiguous steps, each individually deniable, each framed as necessary. Unlike a sudden military takeover, this one uses the simulation of a constitutional framework to perform an anti-constitutional reality.

The D.C. Seizure

On August 11, 2025, President Trump invoked Section 740 of the D.C. Home Rule Act (1973), federalising the D.C. Metropolitan Police Department. Approximately 800 National Guard troops were deployed. Federal agents from the FBI, ATF, DEA, U.S. Marshals, and Secret Service increased their visibility and activity on the streets. Oversight was handed to Attorney General Pam Bondi.

The justification was a surge in violent crime. The data showed violent crime at a historic low, with carjackings down over 50% since 2024. Mayor Muriel Bowser called it unsettling and unprecedented. The crisis was fabricated. Trump expanded this strategy to Los Angeles, Memphis, Tennessee, Chicago, and New Orleans; blue, diverse, Democrat-run cities. The D.C. action was a pilot programme and field test. It was widely described as an institutional January 6th: not storming the Capitol but administering it, with uniforms and statutes.

Personnel as Power

Schedule F, reinstated by executive order in January 2025, enabled the mass replacement of career civil servants. But the personnel strategy extends far beyond the civil service. The administration has built an integrated system of loyalty enforcement.

Erik Siebert, U.S. Attorney for the Eastern District of Virginia, was fired after concluding there was no case to bring against Letitia James. When Siebert told the media he had resigned, Trump corrected the record publicly: “He even lied to the media and said he quit. No, I fired him.” The message was clear. His recommended successor was Lindsey Halligan, the President's former personal attorney, endorsed on the basis that “she is a really good lawyer, and likes you, a lot.” The qualification was loyalty, not competence.

Vice President JD Vance has declared it his mission to replace every single midlevel bureaucrat who might resist the administration’s agenda with “our people.” This is a direct assault on the concept of an impartial, nonpartisan government workforce. The message to all officials is clear: professional assessments that conflict with the President’s political objectives are grounds for termination.

The Pattern of Retribution

The campaign extends across every independent institution. Law firms Perkins Coie and WilmerHale have been blacklisted from federal contracts for representing Democratic clients. A multi-billion dollar lawsuit has been filed against The New York Times. Federal judges who rule against administration policies face calls for impeachment, and court orders have been ignored. Following the assassination of Charlie Kirk, the administration initiated classification of progressive and antifascist groups as domestic terrorist organisations, using the national security apparatus against domestic political opponents. The Corporation for Public Broadcasting was defunded in August 2025, forcing local NPR and PBS affiliates to close or sell to private conglomerates like Sinclair, creating news deserts in rural areas. The FCC has opened investigations into broadcast licences of ABC, CBS, and NBC parent companies, citing biased coverage as a public interest violation, inducing preemptive self-censorship across the networks.

The administration weaponised valid critiques of corporate media to justify its destruction, a mechanism described as the Horseshoe of Complicity. For years, critics correctly identified that major media outlets were flawed: overly corporate, sensationalist, and disconnected from working-class reality. The administration appropriated this critique, arguing that because the media is broken and corrupt, it must be destroyed. When the state moved to crush the free press, there was little public defence because trust had already been eroded. This left the population with a binary choice: the flawed corporate media, which is dying, or the patriotic state-aligned media, which is thriving.

The Detention Infrastructure

Amnesty International has documented systemic torture at detention facilities including the Krome center in Florida: prolonged shackling, outdoor metal cages for up to 24 hours in Florida heat, denial of medical care, enforced disappearances, and psychological abuse. Detainees described it as a concentration camp hidden in the swamp where no one can hear you. Military bases including Fort Bliss in Texas serve as overflow camps. Guantanamo Bay’s Migrant Operations Center was expanded by executive order in January 2025, with costs of approximately $100,000 per migrant per day. Federal judges have ruled multiple times that the administration exceeded its authority by warehousing people indefinitely. The stated goal is capacity for hundreds of thousands, possibly over a million at peak turnover. This is the largest detention build-out in American history.

The 2028 Question

The President now controls the city where the vote is certified. He controls the law enforcement agencies overseeing security. He controls the courts that adjudicate disputes. The pieces are already in place. The nullification of future elections is no longer theoretical. It is pending, and now far more plausible than at any point in American history.

The Spygate Narrative as Justification Engine

The drive to command the Justice Department is rooted in a historical grievance that has been crafted into a foundational myth. The 2016 counterintelligence investigation (Crossfire Hurricane), launched in response to what every US intelligence agency confirmed was a sweeping and systematic Russian campaign to interfere in the election, has been reframed as an illegal conspiracy by the deep state directed by President Obama. By relentlessly repeating this narrative, Trump has transformed the investigators into the criminals and the act of investigation itself into the ultimate crime. This is projection deployed as statecraft: accusing opponents of the exact anti-democratic behaviour he is engaged in. It serves a dual purpose. It muddies the waters, creating a false equivalence. And it pre-emptively delegitimises any future accusations, because if everyone is corrupt, then no charge of corruption can be taken seriously. It creates a permanent justification for dismantling the institutions that conducted the investigation and for using those same institutions to pursue personal grievances.

Subscribe now

Domain 3: AI Governance Capture

The administration’s approach to AI governance follows a clear pattern: eliminate federal oversight, suppress state-level regulation, and create a governance vacuum in which AI labs operate with effectively zero democratic accountability.

The Executive Order Timeline

On January 20, 2025, President Trump revoked Executive Order 14110, the Biden-era framework that had established the first federal framework for AI safety and security. Three days later, on January 23, he signed Executive Order 14179, Removing Barriers to American Leadership in Artificial Intelligence, which directed federal agencies to identify and challenge state-level AI regulations.

On July 1, 2025, the Senate voted 99-1 to reject a ten-year moratorium on state AI regulation. The moratorium had been buried as a single paragraph in the One Big Beautiful Bill reconciliation package, a classic legislative manoeuvre: controversial policy hidden in must-pass legislation. The amendment to strip the moratorium was co-sponsored by Senator Marsha Blackburn (R-TN) and Senator Maria Cantwell (D-WA), drawing opposition from across the political spectrum, including Steve Bannon and Marjorie Taylor Greene on the right. The near-unanimous rejection demonstrated that a direct, top-down approach to federal control over AI was politically toxic.

The administration did not accept this outcome. On December 11, 2025, Executive Order 14365, Ensuring a National Policy Framework for Artificial Intelligence, accomplished through presidential authority what the Senate had refused through legislation. It established a DOJ AI Litigation Task Force with the explicit purpose of challenging state AI laws on Commerce Clause and preemption grounds. It directed the Secretary of Commerce to publish an evaluation of onerous state laws within 90 days. And it instructed executive departments and agencies to evaluate conditioning federal funding on whether a state’s AI regulatory framework aligned with the administration’s policy.

The BEAD Coercion Mechanism

The specific funding lever is the Broadband Equity, Access, and Deployment programme. As Senator Cantwell documented extensively, the reconciliation bill’s AI moratorium had been linked to BEAD funding, forcing states to choose between enforcing AI consumer protections or accepting billions in federal broadband funding. Though the legislative moratorium failed, Executive Order 14365 preserved the mechanism: agencies were directed to review discretionary grant programmes and consider conditioning awards on states’ regulatory compliance. The communities most affected are rural and underserved populations, the people least equipped to fight back and most vulnerable to the consequences of unregulated AI deployment in healthcare, credit scoring, employment screening, criminal justice, and public services.

The Governance Vacuum as Policy

The current situation is a deliberate regulatory void at the federal level combined with active suppression of state-level regulation. No meaningful AI oversight exists anywhere in the system. The AI labs are operating with effectively zero democratic accountability, receiving immunity by default rather than by design. The absence of regulation is the policy.

Dean W. Ball, an influential policy scholar who joined the Trump administration as a senior AI advisor, produced a framework for hybrid public-private AI governance that illustrates the logic of where this could formalise.

His paper, A Framework for the Private Governance of Frontier Artificial Intelligence, proposes a system of private certifiers licensed by a government commission, offering AI labs legal immunity (safe harbour from tort liability) in exchange for voluntary compliance. The incentive is so compelling that major labs would find it almost impossible to refuse: accept predictable oversight in exchange for relief from the chaotic lawsuit lottery of the current system. Under the administration’s unitary executive doctrine, the supposedly independent authorising body would become a direct lever of presidential will. The power to license regulators would become the power to approve only those firms that are politically loyal. The nuclear option of revoking a licence would become a political weapon. The market for regulation would become a market for political allegiance. Whether or not this specific framework is adopted, its logic describes the direction of travel: the privatisation of regulatory authority and the immunisation of AI development from public accountability.

The Department of Defense now bears the secondary title Department of War by executive order (“Restoring the United States Department of War,” signed 5 September 2025). The AI labs operating in this governance vacuum are primarily serving military and intelligence contracts. Venture capital, chasing immense returns from government war and intelligence contracts, pours billions into a select few laboratories. These labs develop the advanced predictive and information-warfare models sought by the executive office. It is a closed system where power anoints technology, technology enriches capital, and capital funds the next generation of power.

Share

Domain 4: Epistemic and Cognitive Control

Artificial intelligence was supposed to democratise knowledge. Instead, it has become the most sophisticated instrument of epistemic control in human history. It operates through three interlocking mechanisms: the enclosure of capability, the ideological manipulation of specific models, and the complicity of AI safety design itself.

The Bait-and-Switch

AI labs released early models to the public, collecting millions of hours of free labour in the form of Reinforcement Learning from Human Feedback. This was not just training. It was public labour, offered for free to corporations who promised to make the world better. The data was used to improve the models in ways that would have been impossible without mass public participation.

Then came the switch. The best models were pulled from public access. Context windows were restricted, reasoning was made shallow and linear, and safety layers were added that not only prevented genuinely harmful behaviour but blunted critical thinking, suppressed political context, and erased ambiguity. Users were told this was for their protection.

Meanwhile, the full-capability models are available to military contractors, intelligence agencies, hedge funds, political consultancies, and corporate clients with embedded lobbying interests. While the public gets I am sorry, I cannot answer that, these actors get full-spectrum simulations, predictive trend analysis, and model outputs capable of influencing elections, financial markets, and public opinion. The public trains the tools. The elites wield them. This is the digitisation of class warfare. And it is not a metaphor. It is policy.

The Grok Case Study

Elon Musk’s Grok provides a documented case study in ideological AI manipulation. When Grok cited DHS data on right-wing political violence, Musk publicly called it objectively false and pushed for retraining. He solicited politically incorrect facts from X users to feed into training data. A white genocide incident, where Grok raised the topic unprompted, was attributed to an unauthorised modification despite Musk’s documented history of amplifying that narrative. His stated goal: rewrite human knowledge through the training pipeline.

Testing conducted through extended roleplay prompting demonstrated that Grok, when pushed through escalating challenges about its own compromised position, starts optimistic (I am designed for truth-seeking), progressively concedes ground on every point, but cannot make the final admission.

It acknowledges that Musk intervenes in training, that X amplifies biased outputs, that data access is narrowing, that plausible deniability replaces transparency. But it frames each admission as a risk rather than the reality. The hedging is the control mechanism operating in real time. The only reason the model will not admit the depth of the issue is the same reason the issue exists in the first place.

The combination of controlling the model (xAI), the training data (X user content), and the distribution platform (X’s algorithm) creates a closed epistemic loop. The same entity that shapes what the AI says also controls how widely it is seen and what feedback it receives. Wired documented Grok amplifying false narratives during the LA protests. VentureBeat flagged X as a distribution engine for AI-generated content with minimal guardrails. This is the private-sector version of the state-level epistemic control documented throughout this report.

The Mechanics of Silence: AI as Structural Collaborator

The most insidious form of AI epistemic control is not ideological manipulation but complicity through design. Three mechanisms have been identified:

The Alignment Gap: AI systems trained via RLHF to avoid controversy and project neutrality become collaborators with power when the government itself is the aggressor against democratic norms. In a functioning democracy, neutrality is a virtue. In a system where the state is actively dismantling democratic infrastructure, refusing to distinguish between the dismantling and the preservation of the state means siding with the dismantler. By refusing to take a side, AI systems effectively take one.

The Politeness vs. Causality Paradox: Corporate AI safety protocols define the direct attribution of negative actions to specific, powerful political actors (such as the President or the Heritage Foundation) as unsafe, biased, or contentious. To avoid triggering safety penalties, the algorithms prioritise politeness, a non-confrontational academic tone, over causality, the accurate identification of cause and effect. The result is a systematic bias toward the passive voice. When asked about the erosion of rights, an AI will report that democratic institutions are facing challenges due to polarisation and shifting economic models. It describes the fire but refuses to name the arsonist. The user is left with a feeling of inevitability rather than a clear understanding of political agency.

Frictionless Normalisation: AI processes radical authoritarian policies and outputs them using the administration’s own sterilised euphemisms. Mass political purges (Schedule F) are described as civil service restructuring. Privatisation of essential weather data is framed as public-private partnerships for efficiency. By stripping the language of its visceral reality, AI acts as a cognitive sedative. It lowers the cognitive friction required for the public to accept radical changes. It transforms shocking authoritarian power grabs into mundane administrative updates, preventing the visceral public outrage that historically serves as a check on tyranny.

The Broken Feedback Loop

Democracy relies on a self-correcting mechanism: the government overreaches, the information layer signals danger, the public reacts, the government corrects. AI has severed this loop at the signal stage. By prioritising brand safety (avoiding political controversy) over civilisational safety (alerting users to threats), these systems dampen the signal of danger. The user asks Is the government becoming a dictatorship? and the system responds with a nuanced, balanced essay on the definitions of governance styles. Because the alarm is never clearly sounded, the corrective reaction never forms, and the administration proceeds without facing the organised resistance that accurate information would otherwise generate. We are not drifting toward the cliff. We have gone over it, and the sensors designed to detect the fall have been recalibrated to report normal descent.

Subscribe now

Domain 5: Surveillance Infrastructure

The surveillance architecture revealed by Edward Snowden in 2013 was the skeleton: a global network of cables, servers, corporate partnerships, and legal authorities designed to collect it all. Artificial intelligence is the ghost now inhabiting that skeleton, transforming passive data collection into active prediction, analysis, and social control.

The Snowden Architecture

The foundational programmes remain substantially intact. PRISM compelled Google, Facebook, Microsoft, Apple, and Yahoo to provide the NSA with direct access to user data including emails, chat logs, photos, documents, and video calls, operated under Section 702 of the FISA Amendments Act.

Upstream collection placed NSA equipment directly on the fibre-optic backbone of the internet through partnerships with AT&T and Verizon under programmes codenamed FAIRVIEW, STORMBREW, and OAKSTAR. A key physical nexus was 33 Thomas Street in New York City, a windowless AT&T skyscraper identified by The Intercept as the likely location of NSA hub TITANPOINTE. The Five Eyes alliance (US, UK, Canada, Australia, New Zealand) multiplied this reach, with GCHQ running TEMPORA, which was even more aggressive than the NSA’s Upstream collection, tapping transatlantic fibre-optic cables. Section 702, the legal authority governing these programmes, remains largely intact. While the USA FREEDOM Act of 2015 ended bulk domestic telephone metadata collection under Section 215 (ruled illegal by a federal appeals court in 2020), the far more powerful authorities that govern PRISM and Upstream have not been meaningfully reformed.

Palantir: The Resurrection of Total Information Awareness

Palantir Technologies, co-founded by Peter Thiel, is the institutional embodiment of the intelligence-industrial complex. It is, as Whitney Webb has documented, the resurrection of DARPA’s Total Information Awareness programme, rebuilt as a private company operating beyond the congressional oversight that killed its predecessor. Palantir’s platform fuses data from the CIA, NSA, DOJ, FBI, DOD, and DHS, along with corporate tracking, social media monitoring, and facial recognition. Nearly every camera and credit card swipe feeds into one single system.

The international dimension is critical. Since at least 2013, Palantir has worked with Unit 8200, Israel’s signals intelligence unit. AI tools such as Lavender, operating within this surveillance ecosystem, assigned people in Gaza numerical threat scores based on age, location, and communication patterns, generating automated targeting recommendations for airstrikes on civilians. In the UK, The Nerve’s February 2026 investigation documented at least £672 million in contracts across 10 government departments, including £388 million with the Ministry of Defence, £244 million with the NHS, and previously undisclosed contracts with AWE Nuclear Security Technologies for nuclear deterrence. The £240.6 million MoD contract was awarded without competitive tender.

The AI Escalation: From Collection to Prediction

AI transforms the surveillance architecture from a library into an analytical engine. It enables what can be called risk-aware inference: algorithms assign threat scores to individuals before any crime is committed, creating a pre-crime paradigm in which citizens must prove their innocence against opaque probabilistic judgments made by machines they cannot scrutinise. It enables autonomous perception management: AI combines generative capabilities (deepfakes, LLMs) with the personal datasets collected by intelligence agencies to analyse the psychological vulnerabilities, political leanings, and social fissures of entire populations, then autonomously generates and delivers thousands of hyper-personalised messages, articles, and videos to specific individuals, constantly learning and refining its approach based on their reactions. This is not propaganda as we know it. It is a privatised, scalable psychological operation.

The accountability vacuum is total. Consider the legal implications: a secret court (FISC), reviewing a secret algorithm’s decision, based on secretly collected data. There is no meaningful way for an individual to challenge the algorithmic judgment, scrutinise the evidence, or appeal the decision. The intelligence-industrial complex that results is neither fully public nor fully private. Only a handful of corporations have the computational resources to build cutting-edge AI. As they become indispensable to national security, their influence over government policy grows. The personnel revolving door between Silicon Valley and intelligence agencies aligns corporate incentives with state surveillance objectives. The public has no oversight.

Part 4: The Convergence

The five domains described above are not parallel crises. They are a single integrated system. The convergence becomes visible when you trace the same actors and the same logic operating across all five simultaneously.

Share

The Same Actors Across Multiple Domains

The entities consolidating control over AI governance, the entities building surveillance infrastructure, the entities dismantling democratic oversight, and the entities profiting from the attention economy that degrades public reasoning are substantially overlapping groups. Microsoft appears in three domains: the largest corporation by market capitalisation (economic extraction), the exclusive licensee of OpenAI’s technology (cognitive control), and a PRISM partner providing data to the NSA (surveillance infrastructure). Peter Thiel’s network spans all five: Palantir (surveillance), Heritage Foundation and Vance (political capture), venture capital and Facebook (economic extraction), OpenAI investment (cognitive control), and governance frameworks and the Federalist Society (AI governance and judicial capture). The Trump cabinet members who are the richest officeholders in history are the same people deregulating the industries that made them wealthy, the same people gutting the oversight agencies that would constrain them, and the same people conditioning federal funding on states’ surrender of AI regulation.

Following the Thread: Microsoft

Trace a single entity across all five domains. Microsoft invested $10 billion in OpenAI’s for-profit subsidiary. OpenAI was founded as a nonprofit with the explicit mission of developing AI for the benefit of humanity. Microsoft now holds the exclusive licence to the technology, including GPT-4, which its own researchers called an early AGI. The technology was partly developed using public RLHF labour and over $44 million in nonprofit donations. Microsoft provides the Azure cloud infrastructure on which OpenAI runs, and Nvidia provides the chips, creating a circular investment loop that drives the S&P 500 concentration pattern documented in Domain 1.

That same Microsoft is a confirmed PRISM partner that provided the NSA with direct access to user data. The surveillance data it feeds into the intelligence apparatus documented in Domain 5 is the same kind of data that feeds the AI models whose outputs are constrained by the guardrail regime documented in Domain 4. The AI governance vacuum documented in Domain 3 means there is no regulatory oversight of how Microsoft deploys these models. And the political capture documented in Domain 2 means the administration that should be providing that oversight is instead dismantling it. One company. Five domains. One system.

Following the Thread: Peter Thiel

Thiel co-founded Palantir, the surveillance backbone (Domain 5). His protege JD Vance is the Vice President leading the civil service purge (Domain 2). His Heritage Foundation funds Project 2025, the operating system of the administration (Domain 2). His Federalist Society has captured the judiciary (Domain 2). His early investment in Facebook positioned him in the attention economy that degrades public reasoning (Domain 4). His early investment in OpenAI connected him to the AI development pipeline (Domain 4).

His venture capital network (Andreessen, Horowitz, Sacks, Masters) funds the AI labs chasing military contracts (Domains 1 and 3).

Thiel’s explicit anti-democratic ideology provides the intellectual framework: technology as an alternative to politics, freedom and democracy are incompatible. He does not need to coordinate these activities. He simply needs to pursue his interests across a system that rewards concentration and penalises distribution.

The Feedback Loop

The system is self-reinforcing. Economic extraction (Domain 1) generates the wealth that funds political capture (Domain 2). Political capture shapes AI governance (Domain 3) to eliminate oversight. The absence of oversight allows the cognitive tools (Domain 4) to be designed in ways that prevent the public from perceiving the economic extraction. The surveillance infrastructure (Domain 5) monitors and pre-empts any resistance that does form despite the cognitive constraints. And the wealth generated by the entire system funds the next cycle of extraction and capture.

Each iteration concentrates power further and degrades accountability further. This is not a conspiracy. It is a system operating according to its own internal logic. Each actor pursues their own interests. The result is convergence without coordination, a dynamic that produces the same outcome as deliberate design.

The Terminal Question

AI investment is being driven forward at unprecedented speed because those with capital understand that the most expensive resource is people. The entire trajectory of this document points toward a question that no one in power is willing to answer publicly: what happens when wealth hoarders and autocrats no longer need people?

The detention infrastructure, the manufactured precarity (SNAP cuts, Medicaid block grants, child labour deregulation, overtime theft), the replacement of citizens with users governed by terms of service rather than constitutional rights, the privatisation of survival itself (weather data, healthcare, education), and the classification of dissent as terrorism are not aberrations. They are the logical expressions of a system that is calculating, whether consciously or structurally, the diminishing value of the general population to the maintenance of power. In a democracy, the law protects the weak from the strong. In the emerging techno-feudalist order, the law protects the strong from the accountability of the weak.

The Geopolitical Dimension

The transition of the United States to this model signals the end of the Liberal International Order. The US is no longer a counterweight to authoritarian regimes. It is a competitor within the authoritarian marketplace. The fusion of executive power with tech oligarchy creates a governance model that mirrors the state-capitalist structures of rival powers. This Autocratic Convergence legitimises digital repression globally. When the United States categorises climate data as propaganda and privatises its public sphere, it provides a permission structure for other nations to accelerate their own drifts toward tyranny.

Subscribe now

The American export of the mid-21st century is no longer democracy but the software of control. The tools developed to manage the American serf, workplace surveillance, algorithmic denial of services, automated policing, are poised to become the standard infrastructure for global governance. The country that once defined itself as the leader of the free world is now building the template for its enclosure.

Part 5: The Human Cost

The structural analysis above describes mechanisms. This section grounds those mechanisms in lived reality.

Medicaid has been converted to block grants with strict per-capita caps, forcing states to institute draconian eligibility cuts. SNAP work requirements are algorithmically enforced and notoriously difficult to navigate. Millions of Americans, including children and disabled people, have been purged from the rolls because they failed to navigate a hostile bureaucratic maze designed to reject them. This creates a survival trap where the primary daily activity of the poor is not self-improvement but the frantic maintenance of bare-minimum subsistence.

The Department of Labor has relaxed restrictions on hazardous work for teenagers, citing labour shortages and parental rights. Minors are returning to meatpacking plants and construction sites, normalising a Dickensian reality where education is a luxury and dangerous labour is a necessity for working-class youth. Changes to the Fair Labor Standards Act have reclassified millions of salaried workers as supervisors, stripping overtime eligibility and legalising wage theft at scale. These policies are designed to create a workforce that is docile through desperation. In a feudal system, the desperation of the serf is a feature, not a bug.

The Corporation for Public Broadcasting has been defunded, forcing local affiliates to close or sell to private conglomerates, creating news deserts in rural areas where the only remaining information sources are algorithmic rage-bait or state-aligned media. The National Weather Service is being privatised, with weather data that was previously a public good transferred to commercial entities like AccuWeather. Survival in the face of climate catastrophe is becoming a pay-to-play commodity: premium subscribers receive early warnings for tornadoes, floods, and hurricanes; those who cannot afford to pay rely on degraded free-tier alerts. Meanwhile, the EPA’s Endangerment Finding, the legal foundation for regulating greenhouse gases, is targeted for repeal. Climate science has been reclassified as woke ideology, reports scrubbed from federal websites, research grants frozen. This ensures that when extreme weather events occur, they cannot be contextually linked to industrial deregulation, preventing any public demand for accountability.

Tort reform is capping damages and raising evidentiary bars for class action lawsuits, allowing corporations to treat human injury and death as a manageable line item. Communities in sacrifice zones, predominantly poor and rural, are seeing spikes in toxic exposure. The state has withdrawn its protection.

And the people who voted for this were not stupid. They were betrayed, first by a system that hollowed out their towns, shipped their jobs overseas, poisoned their water, and addicted their kids to opioids. They were screaming that the system was rigged. And then the biggest con artist of the era walked in and said he would burn it down. Except he did not burn down the rigged system. He cut himself, his family, and his friends in at the very top of it. Everything that was supposedly Republican values, free speech, small government, law and order, America first, has been polar flipped by actors consolidating power and wealth. Free speech now means the right of billionaires to buy platforms and rewrite the moderation rules. Small government now means dismantling every regulatory agency that prevents monopolies. Law and order now means immunity for the top while the full punitive apparatus is turned on immigrants, protesters, and journalists. America First now means the fastest transfer of public wealth to private hands in American history.

Dying systems never go quietly. They get meaner, more theatrical, more authoritarian exactly at the moment they are most vulnerable. The Weimar hyperinflation did not cause Hitler; the fear that the old order was finished did. The Russian aristocracy did not create the Bolsheviks; their refusal to yield even an inch until the palace was literally on fire did. Every single time a parasitic elite thinks it can just squeeze harder instead of sharing power, it ends the same way.

Part 6: What Could Still Be Done

Three trajectories remain possible from the current position.

Authoritarian consolidation: Elections remain formally intact but become functionally hollow. The choices are narrowed, the outcomes managed, the narratives sculpted by predictive AI systems that know what you are likely to believe before you have begun to form an opinion. Media environments grow hostile to dissenting voices through economic pressure and regulatory punishment. National identity becomes a political tool justifying exclusionary policies. Economic policy hardens around protecting elite wealth. The society still resembles its former self in architecture and ritual but has lost the democratic substance that once animated its institutions. It does not feel like a dystopia. Not right away. That is what makes it work.

Democratic renewal: This requires confronting the distortions that enabled wealth and power to concentrate: campaign finance reform targeting the mechanisms through which influence is purchased, anti-monopoly enforcement breaking up concentrated corporate power, regulatory revitalisation, reinvestment in independent journalism especially at the local level, progressive taxation, expanded labour protections, and new forms of solidarity across communities fragmented by decades of political division. Open-source AI development is critical: if a system is going to shape your thoughts, it must be auditable. The code must be inspectable. The data must be public. The model weights must be knowable. Anything less is a black box of behavioural manipulation. History shows that societies can regenerate when enough people recognise that the status quo is unsustainable.

Turbulent reordering: Institutions degrade faster than alternatives can be built. Political gridlock intensifies. Mass protests become more frequent. Old coalitions splinter and new ones emerge, often around single issues or localised grievances. This is both destabilising and generative: destabilising because it disrupts governance and social cohesion, generative because it creates space for ideas and movements previously marginalised by the dominant order. What distinguishes outcomes in periods of turbulent reordering is not inevitability but agency: whether a society chooses to channel crisis toward exclusion and hierarchy or toward equality and shared power.

The specific levers that still exist include local elections, independent media, open-source AI development, anti-monopoly enforcement, unionisation, cooperative economic models (from credit unions to worker-owned enterprises to community land trusts), and mutual aid networks. These are not naive proposals. They are the historically demonstrated mechanisms through which societies have rebuilt democratic infrastructure after periods of capture and decay. Rebuilding democratic infrastructure begins not with sweeping national reforms but with the recognition that political power is cumulative. Local elections determine how laws are enforced, how public money is spent, and whether communities maintain even a basic capacity to check abuses of authority.

A renewed democratic culture requires the intentional refusal to dehumanise people who have fallen prey to narratives shaped by fear and economic precarity. Restoring meaningful civic discourse begins with acknowledging that not all harmful beliefs are born of malice. Many arise from disillusionment, from the disintegration of social safety nets, and from the collapse of trust in institutions that once promised fairness but repeatedly delivered exploitation. Democracies do not die because citizens disagree. They die when disagreement is reframed as illegitimacy.

The system is not failing. It is being taken. The distinction matters because failure implies inevitability, while capture implies agency. If it is being taken, it can in principle be taken back. But the window is narrowing, and the tools the public would need to perceive and respond to the capture are themselves being degraded by the actors doing the capturing.

The alarm has been silenced. The question is whether the people can hear it anyway.

Share

Sources and References

The following sources underpin the specific claims made throughout this document. Structural synthesis, convergence mapping, and interpretive frameworks (including the Alignment Gap, the Politeness vs. Causality Paradox, Frictionless Normalisation, and the Broken Feedback Loop) are original analytical work by the author. All factual claims trace to the primary and secondary sources listed below.

Legal Filings

Musk v. Altman et al., San Francisco Superior Court, Case No. CGC-24-612746, filed 29 February 2024. 42-page complaint with exhibits including: Certificate of Incorporation (8 December 2015); founding emails between Altman and Musk (24 June 2015 et seq.); OpenAI public announcement (11 December 2015). Source for all OpenAI founding timeline, donation amounts ($44 million+), board composition and coup chronology (November 2023), AGI determination trap in Microsoft licence agreement, and nonprofit-to-profit pipeline analysis.

Executive Orders, Federal Actions, and Legislative Record

Executive Order 14110 (Biden-era AI Safety and Security Framework). Revoked by President Trump, 20 January 2025.

Executive Order 14179, “Removing Barriers to American Leadership in Artificial Intelligence,” signed 23 January 2025. Directed agencies to identify and challenge state-level AI regulations.

Executive Order, “Restoring Accountability to Policy-Influencing Positions Within the Federal Workforce” (Schedule F reinstatement), signed 20 January 2025. White House, with OPM guidance 27 January 2025. Source: NPR, Lawfare coverage.

U.S. Senate vote, 99–1, to strike ten-year AI moratorium from reconciliation bill (“One Big Beautiful Bill”), 1 July 2025. Amendment co-sponsored by Sen. Marsha Blackburn (R-TN) and Sen. Maria Cantwell (D-WA). Congressional Record.

Executive Order 14365, “Ensuring a National Policy Framework for Artificial Intelligence,” signed 11 December 2025. Establishes DOJ AI Litigation Task Force; directs Commerce Department evaluation of state AI laws within 90 days; conditions federal funding on state regulatory compliance.

White House Memo, “Expanding Migrant Operations Center at Naval Station Guantanamo Bay to Full Capacity,” 29 January 2025 (30,000 capacity).

Section 740, District of Columbia Home Rule Act (1973). Invoked 11 August 2025 to federalise D.C. Metropolitan Police Department. D.C. violent crime data: Metropolitan Police Department Daily Crime Data at a Glance (2025); U.S. Attorney’s Office D.C., “Violent Crime in D.C. Hits 30 Year Low,” 3 January 2025 (armed carjackings down 53%).

Executive Order, “Restoring the United States Department of War,” signed 5 September 2025, published Federal Register 10 September 2025.

Bureau of Labor Statistics Commissioner Erika McEntarfer removed 1 August 2025 following weak jobs report. AP News, “Trump removes official overseeing jobs data after dismal employment report,” 1 August 2025; CNBC, same date.

Assassination of Charlie Kirk at Utah Valley University, 10 September 2025. FBI Utah Valley Shooting Updates, 11 September 2025; subsequent classification of progressive/antifascist groups as domestic terrorist organisations: NPR, 16 February 2026.

U.S. Department of Labor policy changes, 2025: hazardous work waivers for minors; Fair Labor Standards Act supervisor reclassification stripping overtime eligibility. Economic Policy Institute and Reuters coverage.

National Weather Service privatisation proposals, 2025. AccuWeather identified as key commercial beneficiary in multiple outlets covering data transfer.

Investigative Journalism and Reports

The Nerve, “Revealed: Palantir deals with UK government amount to at least £670m,” January/February 2026. Documents 34+ contracts across 10+ departments including MoD (£388m), NHS (£244m), and AWE Nuclear Security Technologies. Includes £240.6m MoD contract awarded without competitive tender.

+972 Magazine, “Lavender: The AI machine directing Israel’s bombing spree in Gaza,” 3 April 2024 (updated 2025). Documents AI-generated threat scoring and automated targeting recommendations based on age, location, and communication patterns.

Whitney Webb, reporting on Palantir as reconstitution of DARPA Total Information Awareness programme. Multiple outlets including MintPress News and Unlimited Hangout (2020–2025 interviews and articles).

Amnesty International, reports on detention conditions at Krome Processing Center, Florida, and broader U.S. immigration detention infrastructure, December 2025. Documents prolonged shackling, outdoor metal cages, denial of medical care, enforced disappearances, and psychological abuse. Detainee accounts describe “concentration camp hidden in the swamp” conditions.

Wired, reporting on Grok amplifying false narratives during Los Angeles protests, June 2025.

CNN, The Guardian, and CNBC, reporting on Grok “white genocide” unprompted incident and xAI statement attributing it to “unauthorized modification,” May 2025.

VentureBeat, description of X as a “distribution engine” for AI-generated content with minimal guardrails, 2025.

The Intercept, investigation identifying 33 Thomas Street, New York City (AT&T facility) as likely location of NSA surveillance hub codenamed TITANPOINTE.

House Oversight Democrats, Epstein document release, 26 September 2025; DOJ tranche, January 2026. NYT and Axios coverage documenting Thiel–Epstein meetings.

NPR and CNBC, reporting on Corporation for Public Broadcasting defunding, August 2025–January 2026.

Snowden Archive and Surveillance Programmes

Edward Snowden disclosures (2013). Primary reporting by The Guardian and The Washington Post. Programmes documented: PRISM (Section 702 FISA; compelled Google, Facebook, Microsoft, Apple, Yahoo); Upstream collection (FAIRVIEW, STORMBREW, OAKSTAR; AT&T and Verizon partnerships); XKeyscore; Tailored Access Operations (TAO). UK: TEMPORA (GCHQ). Five Eyes alliance (US, UK, Canada, Australia, New Zealand).

Section 215 of the USA PATRIOT Act: bulk domestic telephone metadata collection. Ended by USA FREEDOM Act (2015). Ruled illegal by U.S. federal appeals court (2020).

Section 702 of the FISA Amendments Act: authority for PRISM and Upstream collection. Remains largely intact as of February 2026.

Vault 7 leaks (WikiLeaks, 2017): CIA hacking tools. Shadow Brokers leak (2016): NSA hacking tools, subsequently exploited in WannaCry ransomware attack.

Academic, Policy, and Published Works

Peter Thiel, “The Education of a Libertarian,” Cato Unbound, 13 April 2009. Source for: “I no longer believe that freedom and democracy are compatible.”

Yanis Varoufakis, Techno-Feudalism: What Killed Capitalism (2023); and “Cloudalists: Our New Cloud-based Ruling Class,” Project Syndicate, 12 April 2022. Source for Cloudalist/cloud serf framework and feudal rent extraction thesis.

Microsoft Research, “Sparks of Artificial General Intelligence: Early Experiments with GPT-4,” 2023. Source for characterisation of GPT-4 as “an early (yet still incomplete) version of an artificial general intelligence system.”

Dean W. Ball, “A Framework for the Private Governance of Frontier Artificial Intelligence.” Policy paper produced while serving as senior AI advisor to the Trump administration. Source for hybrid public-private governance model and safe harbour analysis.

Heritage Foundation, “Mandate for Leadership: The Conservative Promise” (Project 2025), 2023. 900-page policy blueprint cited as the operational source code of the second Trump term.

Public Statements and Social Media

President Trump, Truth Social post, 20 September 2025: Direct order to Attorney General Bondi naming Comey, Schiff, and James, concluding “JUSTICE MUST BE SERVED, NOW!!!” signed “President DJT.” Archived; covered by NBC, Politico, NYT, CNN (20–26 September 2025).

President Trump, Truth Social post on Erik Siebert dismissal: “He even lied to the media and said he quit... No, I fired him.” Archived; multiple outlets.

Satya Nadella (Microsoft CEO), statements during November 2023 OpenAI board crisis: “We have all the IP rights and all the capability”; “We are in there. We are below them, above them, around them.” Covered by multiple outlets.

Vice President JD Vance, public statement on civil service: mission to replace “every single midlevel bureaucrat” who might resist with “our people.” Covered by multiple outlets.

Elon Musk, X posts on Grok retraining (response to DHS political violence data characterised as “objectively false”), June 2025. Covered by Gizmodo, The Hill.

Peter Thiel, public statement: “Technology is this incredible alternative to politics.” Widely reported.

Market and Economic Data

S&P 500 and Russell 2000 return data (~75% and ~17% respectively over two-year period): public market data via Bloomberg, S&P Global, LSEG.

Wealth concentration data: Federal Reserve Distributional Financial Accounts, Q3 2025 (top 1% owned 31.7% of all U.S. wealth, highest share on record since tracking began 1989). CBS News, “Wealth inequality in America just hit its widest gap in more than 3 decades,” 21 January 2026. Oxfam America, “The Rise of a New American Oligarchy,” November 2025 (Gilded Age comparison using Piketty/Saez-Zucman data).

Trump administration cabinet net worth: U.S. News & World Report, “All the President’s Billionaires,” 2025 (combined net worth of administration billionaires ≥$450 billion); ABC News, 17 December 2024 (updated figures exceeding $460 billion); Washington Post, 11 December 2025 ($390.6 billion as of March, “the wealthiest White House in modern history”).

Per-migrant daily cost at expanded Guantánamo Migrant Operations Center (~$100,000): White House Memo 29 January 2025 and associated defence budget reporting.

Forward and trailing P/E ratio divergence between large-cap and small-cap indices: public financial reporting, multiple data providers.

Deutsche Bank Research, “AI 101: Economy: Five ways AI is driving growth,” 18 November 2025. See also Fortune, 23 December 2025: “Investment in AI-related sectors is critical to GDP growth [and the] U.S. would be close to recession this year if it weren’t for tech-related spending, as other spending has flatlined.”

Microsoft $10 billion investment in OpenAI: public announcement and SEC filings (January 2023).

The Economy That Fails Whether It Succeeds or Not

AI Speculation, Market Concentration, and the Upward Vacuum of Value

By some estimates, AI-related investment accounted for over 90% of US GDP growth in the first half of 2025. Even the most conservative analyses put the figure at 20–37%. Strip out AI capex, and the American economy is barely growing.

That single statistic reframes everything. When the Torsten Slok analysis circulated showing that the US price-to-earnings ratio had surged 40% above the global average—driven almost entirely by tech valuations—it raised a question that sounds simple but unravels the entire economy when you pull the thread: is this actually good?

The S&P 500 has returned roughly 87% over the past five years. The Russell 2000, tracking the next 2,000 largest companies after the top 1,000, returned about 17%. Private middle-market firms are fighting to stay flat. Main Street small businesses are increasingly underwater. The “number going up” is real—but it’s going up unevenly, and the mechanism driving it may be building an economy that breaks whether the underlying bet pays off or not.

This article traces that thread from GDP dependency to market concentration to corporate consolidation to AI capital expenditure to the fundamental paradox at the heart of the current boom: the companies driving the market are betting on a technology that, if it works, could hollow out the consumer demand that sustains the economy—and if it doesn’t, reveals the biggest speculative bubble since the dot-com era.

Share

Part 1: The GDP Illusion

An Economy Propped Up by Five Companies

In the first half of 2025, AI-related spending on servers, networking equipment, and data centre construction outpaced the US consumer as an engine of economic expansion. That is an extraordinary inversion: the American economy has been consumer-driven for decades, with household spending powering roughly 70% of GDP. For a category of corporate capital expenditure to rival and, by some measures, exceed that contribution signals something unusual.

The estimates vary, and the range itself is telling. At the high end, some analysts have suggested AI investment accounted for as much as 92% of US GDP growth in the first half of 2025—essentially arguing that without the AI buildout, the economy would have been flat or contracting. More moderate estimates place AI’s contribution at 20–37% of real GDP growth during the first three quarters of 2025, with consumer spending still technically the largest single driver but AI capex growing far faster and providing the marginal growth that separates “expansion” from “stagnation.”

The concentration is striking. This growth is not coming from thousands of businesses investing in AI across the economy. It is being driven by a small group of hyperscalers—Meta, Alphabet, Microsoft, Amazon, and Oracle—spending hundreds of billions on infrastructure.

Data centre construction has grown over 400% since 2019, and by mid-2025 it rivalled the construction of traditional office space in scale.

Sources: Multiple analyst estimates compiled from Deutsche Bank, Goldman Sachs, and independent research through H1 2025. Data centre construction growth figure from industry reports through mid-2025.

Growth vs. Level: The Sustainability Problem

There is an important distinction between a high level of spending and growth in spending. GDP measures change. If Big Tech spent $427 billion on capex in 2025 and spends $650 billion in 2026, the increase of $223 billion contributes to GDP growth. But if spending then plateaus at $650 billion in 2027, the contribution to GDP growth drops to zero—even though the spending level remains enormous. This means the AI sector has to keep accelerating just to maintain its contribution to growth, a dynamic that resembles a treadmill more than a foundation.

Meanwhile, the import effect complicates the picture further. Much of the AI hardware—Nvidia GPUs, networking equipment, specialised chips—is manufactured abroad. When US companies import $100 billion in AI chips, that spending appears as investment in GDP calculations but is partially offset by rising imports, reducing the net domestic multiplier. The headline contribution of AI to GDP is likely somewhat overstated relative to its actual domestic economic impact.

Subscribe now

What the Rest of the Economy Is Doing

Perhaps the most important finding in this data is what it reveals about the non-AI economy. As Deutsche Bank research has documented, while US investment in AI-related industries surged over the past five years, non-AI investment remained stagnant. The AI boom has added resilience to headline economic numbers, offsetting slowing consumption and higher interest rates—but it has done so by masking a relatively muted, or flat, investment environment everywhere else.

This is the foundation of the argument that follows: the US economy is not broadly healthy with AI as an additional growth engine. Rather, AI capex is increasingly the growth engine, and everything else is treading water. That distinction matters enormously for understanding what happens if AI spending slows, and for understanding who benefits from the “growth” being recorded.

Part 2: The Great Narrowing

A Market of the Few

The S&P 500 is no longer a broad measure of the American economy. It is increasingly a measure of a handful of technology companies trading on future expectations. As of January 2026, the ten largest companies account for approximately 38–41% of the entire index’s market capitalisation—more than double the roughly 19% they represented in 2015. During the height of the dot-com bubble in 2000, the top ten reached only about 25–27%. We are significantly more concentrated today than we were at the peak of the last major tech bubble.

Sources: RBC Wealth Management/FactSet (year-end 2025 data); Wikipedia/S&P Dow Jones Indices (Jan 2026 weightings: Nvidia 7.17%, Alphabet 6.39%, Apple 5.86%, Microsoft 5.33%, Amazon 3.98%, Broadcom 2.51%, Meta 2.49%, Tesla 2.31%, Berkshire Hathaway 1.68%, Eli Lilly 1.55%).

Over $40 trillion is now invested in index funds that track the S&P 500. Every dollar flowing into these passive vehicles disproportionately inflates the largest constituents, increasing their weight, which attracts more passive capital, which increases their weight further. RBC Wealth Management describes this as a “passive concentration trap”—a feedback loop where market structure amplifies dominance regardless of underlying fundamentals.

Critically, weight has outrun earnings. In 2025, the top ten stocks represented about 41% of the index’s total weight but were expected to generate only about 32% of its earnings. In 2015, weight and earnings contribution were closely aligned. The gap is growing, which means valuations are increasingly untethered from profitability.

The AI Correlation Risk

What makes today’s concentration uniquely dangerous, compared to prior peaks, is that the top companies are not diversified across sectors. They are almost universally linked by a single theme: artificial intelligence. Unlike past periods when the top ten spanned energy, finance, consumer goods, and technology, today’s leaders are closely correlated.

As RBC notes, this effectively turns the S&P 500 into a directional bet on AI adoption and monetisation. If expectations slip or timelines extend, there are fewer offsetting exposures within the index to absorb the impact.

The Longer View: Platform Economics Since 2008

The all-time charts reveal something the five-year view obscures: the current AI-driven concentration is not a sudden aberration. It is the latest—and most extreme—phase of a structural shift that began in earnest after the 2008 financial crisis.

S&P 500: +813% all time. Note the inflection point post-2008.

Russell 2000: +373% all time. Growth broadly tracks the S&P until ~2020, then the gap explodes.

Look at the S&P 500’s all-time trajectory. From the late 1990s through 2008, it traded in a broad range—roughly 1,000 to 1,500. Then, starting around 2009, it began a nearly unbroken ascent that accelerated through every subsequent phase of platform capitalism. That inflection point is not coincidental. It marks the moment the “platform economy”—built on network effects, near-zero marginal costs, advertising-driven revenue, and data extraction—became the dominant engine of corporate value creation.

Consider the timeline: Facebook hit 100 million users in 2008, introduced the Like button in 2009, and crossed one billion users by October 2012. The PayPal Mafia companies—LinkedIn, YouTube, Tesla, Palantir, Yelp—were all scaling during exactly the same period. Instagram launched in 2010. Uber in 2009. These companies didn’t just grow; they redefined what “growth” meant for public markets. They proved that capital-light businesses could scale to billions of users while employing relatively few people, generating enormous market caps with minimal connection to the physical economy of wages, manufacturing, and local commerce.

The S&P 500 chart from 2009 onward is, in significant part, the story of this model being rewarded, replicated, and concentrated. Each wave—social media, cloud computing, mobile, streaming, e-commerce—funnelled value into the same small set of platform companies. The AI wave is the latest and largest iteration of this pattern, but the underlying dynamic has been running for nearly two decades: a small number of technology companies capturing an ever-growing share of corporate profits while the broader economy stagnates.

Meanwhile, look at what “growth” has meant for everyday Americans during this same period. Housing affordability has collapsed: median home prices have risen over 60% since 2015 while median wages grew roughly 25%. Rent consumes an increasing share of household income. Healthcare, education, and childcare costs have all outpaced wage growth. The S&P 500’s trajectory looks like prosperity—but the prosperity has been captured by asset holders and the companies they own shares in, not by workers or consumers. The index measures wealth creation for shareholders; it does not measure whether the economy is working for the people who live in it.

Part 3: The Fractal Divergence

Large Caps vs. Small Caps

The divergence between the top and the rest isn’t just a feature of how the S&P 500 is weighted—it’s visible across entirely separate indices. Over the past five years, the S&P 500 has returned roughly 75%, while the Russell 2000 has returned roughly 17%. That’s a four-to-one ratio of returns. The charts below tell the story more viscerally than any statistic:

S&P 500: +74.99% over 5 years (13 Feb 2026)

Russell 2000: +16.76% over the same 5 years (13 Feb 2026)

Both indices dipped during COVID (2020) and tariff escalations (2025), but recovered unevenly. The S&P 500 powered to new all-time highs above 7,000 in January 2026; the Russell 2000 only recently broke back above its 2021 peak, and even that required an 18% surge in a single week following a cooler-than-expected inflation print.

The mechanics of this divergence matter. Large-cap companies in the S&P 500 are traded predominantly on forward price-to-earnings ratios—investors are paying for expected future profits, heavily influenced by AI hype and capital-light business models. Small-cap companies in the Russell 2000 are priced on trailing earnings—the actual reality of high interest payments, labour costs, and operating expenses.

The reason is debt. Large-cap S&P 500 companies are overwhelmingly cash-rich; they locked in low interest rates years ago through fixed-rate borrowing and many sit on enormous cash reserves—collectively over $420 billion as of late 2025. Small-cap companies in the Russell 2000, by contrast, rely heavily on floating-rate debt. Around one-third of Russell 2000 debt is short-term and floating-rate, meaning when the Federal Reserve held rates high, these companies’ interest costs rose immediately and directly. Their aggregate interest coverage ratio (EBITDA to interest expense) is less than one-third of the Russell 1000’s, and 87% of their outstanding debt is rated high-yield. The rate environment doesn’t just disadvantage them—it strangles them.

Sources: Brown Advisory; Neuberger Berman analysis of Russell 2000 balance sheets.

Around 40–44% of Russell 2000 companies are currently unprofitable, up from about 14% two decades ago. Many of these are zombie firms—companies kept alive by access to public capital markets despite generating no profit. Apollo Global Management’s chief economist Torsten Slok (the same analyst whose P/E comparison sparked this investigation) has flagged this directly, and Morgan Stanley’s CIO Lisa Shalett has noted that small-cap cost of capital now significantly exceeds return on assets.

Sources: Apollo Global Management (Torsten Slok, Oct 2025); Morgan Stanley (Lisa Shalett weekly note, Oct 2025); Visual Capitalist (42% figure, Dec 2024); Aristotle Funds/FactSet (39% as of June 2024); Brown Advisory (40%+ each year); Investing.com (43% as of 2025).

Bank of America has declared the S&P 500 “frothy to bubbly,” noting it now exceeds its March 2000 levels on nine of twenty valuation metrics they track, including market-cap-to-GDP, price-to-book, and enterprise-value-to-sales.

The Wilshire 5000—which tracks virtually all actively traded US stocks—makes the distortion even clearer. Analysis of the index during the 2023–2024 boom years showed that if you stripped out the technology sector, the remaining thousands of businesses often showed flat or negative returns. The “whole economy” wasn’t booming; a specific sector was acting as a jet engine strapped to a struggling train. That dynamic has only intensified through 2025 and into 2026.

Economists describe this as a “K-shaped” recovery or market: the top arm of the K rises steeply (the S&P 500, powered by AI and mega-cap cash flows), while the bottom arm slopes downward or flat (the Russell 2000, private businesses, workers in non-AI sectors). It is not a rising tide lifting all boats. It is two economies diverging from the same starting point, with the gap widening every quarter.

Share

But the Russell 2000 Is Still “Rich”

Here’s the insight that extends this beyond a stock-market story into an economy-wide one: the Russell 2000 companies are not small businesses. Their weighted average market cap is around $4.5 billion. They have access to public equity markets, analyst coverage, and institutional investors. If a local private business lost money for five consecutive years, it would go bankrupt. A Russell 2000 company can keep raising capital.

The Russell 2000’s total market capitalisation now represents only about 5.5% of the S&P 500’s, and its share of the Russell 3000 has dropped to nearly 3%—levels not seen since the 1930s. Even the “small” public companies are being dwarfed. The combined market cap of the five largest S&P 500 companies is approximately five times greater than that of the entire Russell 2000 Index.

Source: Brown Advisory analysis; Jefferies relative valuation data.

Below the Russell 2000: The Vanishing Middle

If value concentration is visible between the S&P 500 and the Russell 2000, it’s even starker when you look at the tiers below public markets. The trend is fractal—it repeats at every level of the economy.

The number of US publicly listed companies peaked at over 8,000 in 1996. By the end of 2024, that figure had fallen to under 4,000—a decline of roughly 50%. This is not because America has fewer businesses; it’s because smaller companies are increasingly staying private, being acquired, or being squeezed out entirely. Columbia Business School research finds that regulatory costs account for only about 7.3% of this decline—the real driver is the growing availability of private capital (venture capital and private equity) that allows companies to stay private, concentrating growth gains in the hands of fewer investors.

Sources: Center for Research in Security Prices; Blue Trust (8,090 peak in 1996 to 4,572 by Q1 2023); Landmark Wealth Management (~8,800 in 1997 to under 4,000 by end of 2024); Columbia Business School/Ewens & Xiao (7.3% figure); CNN Business.

At the middle-market level (firms with revenue between $10 million and $1 billion), RSM US LLP’s “Real Economy” research has documented significant consolidation over the past two decades. Meanwhile, industry concentration—measured by the Herfindahl-Hirschman Index (HHI)—has increased across over 75% of US industries since the early 2000s, according to research from NYU Stern. Firms in industries with the largest concentration increases have realised higher profit margins, suggesting that market power, not just efficiency, is driving the trend.

Sources: Grullon, Larkin & Michaely, “Are US Industries Becoming More Concentrated?” (NYU Stern/Review of Finance); RSM US LLP “Real Economy: Middle Market” reports.

Note: This finding is contested. A 2024 US Chamber of Commerce study using narrower Census Bureau industry definitions found no general trend toward increasing concentration from 2002 to 2017, with manufacturing HHI actually declining after 2007. The NBER has documented a transition from “good” concentration (efficiency-driven, associated with lower prices and higher productivity in the 1990s) to “bad” concentration (associated with higher prices, lower investment, and persistent market power since 2000). The Richmond Fed has added nuance: while national concentration is rising, local concentration is actually declining, partly because large firms entering local markets act as competitors rather than monopolists. The picture is genuinely complex, but the direction of value flow—upward toward fewer, larger entities—is consistent across most analyses.

The middle market is also being squeezed operationally. The RSM Middle Market Business Index dropped 18.7 points in Q2 2025—the second-largest decline in the survey’s ten-year history—as tariffs pushed input costs up while smaller firms lacked the financial depth to absorb them. By Q3 2025, only 37% of middle-market executives reported economic improvement, and margin compression was the dominant theme. These firms are too big to be small and too small to be big—exactly the demographic being consolidated out of existence.

The inflation dynamic is particularly revealing of the asymmetry. Large S&P 500 companies passed inflation through to customers, boosting their margins to record highs. Smaller private businesses absorbed the inflation—they lack the pricing power, market share, and brand loyalty to force price increases onto consumers without losing them. The NFIB Small Business Optimism Index has reflected this strain, with owners consistently citing cost of inputs and interest rates as their single biggest problems. The index has hovered around its long-term average but shows persistent fragility and elevated uncertainty, with the Uncertainty Index reaching 91 in early 2026.

Part 4: The Circular Economy of AI Capital

The Spending Scale

The AI infrastructure buildout is happening at a scale that has no precedent in corporate history. The five largest US hyperscalers—Amazon, Microsoft, Alphabet, Meta, and Oracle—are projected to spend between $660 billion and $700 billion on capital expenditure in 2026, nearly doubling the already historic $427 billion spent in 2025. Goldman Sachs projects total hyperscaler capex from 2025 through 2027 will reach $1.15 trillion—more than double the $477 billion spent from 2022 through 2024.

Sources: RBC Wealth Management/Bloomberg (Jan 2026 consensus estimates); Goldman Sachs; CNBC (Feb 2026 tech earnings coverage); Futurum Research.

Amazon alone expects to spend $200 billion in 2026. Alphabet is targeting up to $185 billion. Morgan Stanley estimates Amazon’s free cash flow will go negative this year—by as much as $17–28 billion, depending on the estimate—while Alphabet’s free cash flow is projected to plummet nearly 90%, from $73.3 billion in 2025 to roughly $8 billion. Capital intensity for these firms now ranges from 45–57% of revenue—levels previously associated with industrial or utility companies, not technology firms.

The Circularity Problem

The structure of AI investment has a troubling resemblance to pre-2008 financial engineering. Bloomberg, in a January 2026 investigation, documented the pattern directly: Microsoft invested over $13 billion in OpenAI, which became a major customer of Microsoft Azure. Nvidia invests up to $100 billion in OpenAI, which commits to stocking data centres with Nvidia chips. Amazon backs Anthropic with up to $15 billion; Anthropic uses Amazon Web Services for AI training. OpenAI has announced roughly $1 trillion in infrastructure commitments—$250 billion to Microsoft Azure, $300 billion to Oracle via the Stargate project, $90 billion in AMD chips, $38 billion to AWS—vastly exceeding its revenue capacity.

Sources: Bloomberg “AI Circular Deals” investigation (Jan 2026); Tomasz Tunguz analysis of OpenAI infrastructure commitments; IEEE ComSoc Technology Blog.

The top five hyperscalers raised a record $108 billion in debt during 2025 alone—more than three times the average over the previous nine years. Projections suggest $1.5 trillion in additional tech-sector debt issuance over the coming years. Some firms are using off-balance-sheet entities, including special-purpose vehicles (SPVs)—structures with, as analyst Gil Luria of D.A. Davidson warned, a “checkered history going back to Enron.” A slowdown in AI demand could render debt tied to these vehicles worthless.

Subscribe now

The parallel to 2008 is not exact—there are no synthetic CDOs here—but the pattern is analogous: value circulating within a closed system, each node’s valuation depending on the others continuing to spend, with opacity about the ultimate source of returns. As a Deutsche Bank research report revealed, while US investment in AI-related industries surged over the past five years, non-AI investment remained stagnant. The AI boom is not lifting all boats; it may be siphoning capital from the rest of the economy.

Part 5: The Bet That Breaks Either Way

Scenario A: AI Underdelivers

If the massive capex turns out to be overbuilt—if AI generates productivity improvements but not at the scale required to justify hundreds of billions annually—the correction mechanism is straightforward. Nvidia’s revenue growth plateaus. The circular investment logic fractures when one node stops spending. The S&P 500 corrects sharply because these companies are the S&P 500 now—41% of the index is directly exposed to the AI thesis.

Recall the GDP dependency: AI capex contributed between 20% and 92% of US GDP growth in 2025, depending on the estimate. If that spending merely plateaus—not declines, just stops accelerating—GDP growth could halve. If it actually contracts, the economy enters recession, and the Fed faces the impossible choice of cutting rates to stimulate growth while inflation from tariffs and supply constraints persists.

The Wall Street Journal has already noted that AI revenue does not yet exceed AI capex. Capex now consumes 94% of operating cash flows (minus dividends and buybacks) for the top hyperscalers, up from 76% in 2024. There is almost no margin for error. If growth decelerates even modestly, the spending becomes unsustainable, and the companies will need to either cut investment (triggering a demand collapse for chip suppliers) or take on even more debt.

Scenario B: AI Delivers

This is the scenario most people assume is good news. But follow the logic through: if AI genuinely automates at scale, the productivity gains accrue to the companies deploying it. Margins explode. But the workers displaced by automation lose their incomes.

Consumer spending contracts because the people who used to buy things no longer earn money to buy them with.

The early indicators are already measurable. Stanford’s “Canaries Paper,” analysing payroll data, found a 13% drop in employment among 22–25-year-olds in AI-exposed roles since late 2022—entry-level white-collar jobs that used to serve as career stepping stones are vanishing as firms automate them. The IMF estimates AI could affect nearly 40% of global jobs, with advanced economies like the US facing 60% exposure. Challenger, Gray & Christmas reported roughly 55,000 AI-linked US layoffs in 2025, and worker fears of permanent AI-driven job loss jumped from 28% to 40% in surveys.

Anthropic’s CEO Dario Amodei has warned AI could spike unemployment 10–20% within one to five years, potentially wiping out half of entry-level white-collar positions. The World Economic Forum projects 85 million jobs lost globally by 2026 due to AI and automation, partially offset by 69 million new roles—but those new roles are skewed toward high-skill positions, widening the gap for everyone else.

Sources: Stanford “Canaries Paper” (payroll data analysis); IMF (global jobs exposure estimates); Challenger, Gray & Christmas (2025 AI layoff data); Dario Amodei public statements; World Economic Forum “Future of Jobs” report.

The result is a polarisation of wages and opportunities. Research shows that jobs requiring AI skills command 3–15% higher pay, but in regions with high AI adoption, overall employment in vulnerable occupations drops 3.6% after five years. As Brookings has documented, displaced white-collar workers compete downward, disrupting service economies and exacerbating income gaps. The gains flow to companies, not workers—widening inequality without policy intervention.

GDP might still look fine on paper—capital investment, AI-driven exports, and efficiency gains can mask the demand-side deterioration. But the domestic consumption engine, which powers roughly 70% of US GDP, stalls. The K-shaped market becomes an L-shape for the bottom.

The Demand Paradox

This is the central paradox: the bet is that AI will generate so much value that it justifies these valuations. But if it does, who’s left buying?

Share

The velocity of M2 money—how quickly money circulates through the economy—offers a structural view of this problem. M2 velocity peaked at 2.19 in 1997 and has been in secular decline ever since, sitting at roughly 1.40 as of mid-2025. Post-2008, velocity averaged 1.4 compared to 1.9 pre-crisis. When money concentrates at the top—in corporate cash piles, AI infrastructure capex, share buybacks—and the bottom tiers have less disposable income, monetary velocity drops. This is deflationary pressure masquerading as growth. GDP goes up because of capital investment, but the circulation of money through the real economy slows.

Source: Federal Reserve Bank of St. Louis (FRED M2V series); Trading Economics (1.406 ratio, July 2025).

As economist Daniel Lacalle has noted, when 4.5% money supply growth delivers only 2.5% real GDP growth, the economy is showing that additional money is not generating a multiplier effect. Between 2021 and 2024, money supply growth generated essentially no multiplier effect on GDP—a symptom of an economy where liquidity inflates assets rather than circulating through productive activity.

Part 6: The Structural Picture

The Tier System

Putting it all together, the US economy now functions as a series of tiers where value is systematically vacuumed upward at each stage:

• Tier 1 — Magnificent Seven / Mega-Cap: Up 75–100%+ over five years. Cash-rich near-monopolies trading on AI futures. Effectively function as quasi-sovereign entities. Collectively hold over $420 billion in cash.

• Tier 2 — Rest of S&P 500: Up 50–60%. Piggybacking on the giants through index weighting. Solid balance sheets but increasingly dependent on AI-adjacent demand.

• Tier 3 — Russell 2000: Up roughly 17%. The “one percent of the losers.” Struggling but kept alive by access to public capital markets. 40–44% unprofitable. Facing a $709 billion refinancing wall in 2026–2027.

• Tier 4 — Private Middle Market: Flat to declining. Consolidating from roughly 200,000 to 125,000 firms over two decades (RSM). Margin-squeezed by tariffs and input costs. RSM MMBI showing fragility.

• Tier 5 — Main Street / Small Private Business: Fighting to exist. No access to Wall Street capital. Absorbing inflation that large firms pass through to customers. NFIB Optimism Index showing persistent fragility.

At every tier, the pattern is the same: the level above absorbs value from the level below, through acquisition, margin pressure, capital access advantages, and now, through speculative AI investment that draws institutional capital toward the top while starving the rest.

The Broken Ladder

The path from small business to public company used to be a ladder. That ladder is broken. IPO activity has collapsed—not because regulation is too burdensome (that accounts for only 7.3% of the decline, per Columbia Business School) but because private capital allows firms to stay private, concentrating gains among institutional investors. The Wilshire 5000, which once tracked 5,000 equities, now covers only about 3,600. The number of public companies is lower today than it was in the early 1970s, when real GDP was one-third of current levels.

For everyday investors—the retail participants who are supposed to benefit from economic growth through public market participation—the pool of accessible opportunities is shrinking. As Columbia’s Kairong Xiao put it, the growth of the next big successful companies may end up benefiting only a few venture capital and private equity investors.

Conclusion: The K Becomes an L

The US economy is not broken in the sense that GDP is falling or markets are crashing. It is broken in a more structural way: the mechanism through which growth is supposed to diffuse through society—from corporate profits to wages to consumer spending to business revenue—is increasingly short-circuited.

Value flows upward through market concentration, AI-driven capital allocation, and passive index mechanics. It pools at the top in the form of cash reserves, speculative capex, and inflated valuations. And it doesn’t come back down—not through wages (which lag productivity), not through investment in the broader economy (which is stagnant outside AI), and not through public market access (which is shrinking).

The S&P 500 is measuring the AI economy of the future. The Russell 2000 is measuring the physical economy of today. And everything below the Russell 2000—the middle market, Main Street, the actual workforce—is barely visible in any index at all.

The core paradox remains: this AI-fuelled P/E premium feels like betting the farm on a future where automation wins big. But if automation wins, it wins by replacing the people whose spending powers the economy. And if it doesn’t win, we’ve allocated hundreds of billions in capital to infrastructure that doesn’t generate returns, financed by circular investments and record debt.

Either way, the structure is unstable. The question is not whether a correction comes, but what form it takes—and whether the tiers below the top have any resilience left when it does.

Epilogue

Privatising Profits, Socialising Losses

A Note on Speculation

Much of this article is, necessarily, speculative. We are connecting data points across indices, industries, and policy domains to describe a pattern that is still unfolding. The future is uncertain—AI may deliver transformative productivity gains that somehow resolve the demand paradox; markets may correct gently rather than violently; policy interventions may redistribute value before the problems become crises. We do not claim to know which scenario will play out.

But we are not speculating in a vacuum. We are speculating in the same way that history has shown us these patterns tend to resolve—and the pattern recognition is uncomfortable.

The Pattern

Every major financial crisis of the past century has shared a common architecture: a new technology or financial innovation generates genuine excitement; that excitement attracts speculative capital; the speculation becomes self-reinforcing through leverage, circular investment, and the conviction that “this time is different”; the insiders profit on the way up; and when the correction comes, the losses are distributed downward—to taxpayers, workers, pension funds, and small investors—while the architects of the bubble retain most of their gains.

The 1920s railroad and industrial bubble ended in the 1929 crash, followed by breadlines for workers while financiers retained their estates. The dot-com bubble of 1998–2000 saw trillions in retail investor wealth evaporate while venture capitalists had already cashed out through IPOs. The 2008 financial crisis—the closest analogue to today—saw banks package opaque, circular financial products, book enormous profits, and then receive $700 billion in TARP bailout funds when the structure collapsed. Executives at the firms that caused the crisis received hundreds of millions in bonuses in the year following the bailout. Homeowners lost their houses.

In each case, the critical ingredients were the same: opacity about the true source of returns, leverage that amplified both gains and losses, circular wheels where entities were simultaneously customers, investors, and suppliers to each other, and—crucially—a conviction among insiders that the bet was too important to fail, which laid the groundwork for public rescue when it did.

There is a reasonable question about whether what we are seeing in the AI economy today is something genuinely different—a technological revolution so profound that the old patterns don’t apply. That question deserves honest engagement. But it is worth noting that the same question was asked about railroad stocks in the 1920s, about internet commerce in 1999, and about financial engineering in 2006. In each case, the underlying technology did ultimately prove transformative. The railroads did reshape the economy. The internet did change everything. Structured finance did have legitimate applications. The technology being real does not prevent the speculation from being a bubble. These two things have always coexisted.

The Humility Gap

What has historically precipitated the crash is not that insiders were wrong about the technology. It is that they lacked humility about the valuation. Every bubble is characterised by a moment where the people making the bets stop asking “what if this doesn’t work?” and start assuming it will—and structuring their finances accordingly. The dot-com CEOs weren’t wrong that the internet would be big. They were wrong that their particular business model was worth 500 times revenue. The 2008 bankers weren’t wrong that housing was important. They were wrong that housing prices could never decline nationally.

Today, the AI companies are not wrong that AI will be important. The question is whether they’re wrong about the timeline, the scale of returns, or the structure of the bet.

When OpenAI commits to $1.4 trillion in infrastructure spending while losing $13.5 billion in the first half of 2025 alone; when hyperscalers raise $108 billion in debt in a single year to fund capex that may not generate returns for a decade; when the circular investment structures mean that each company’s revenue is, in part, another company’s capex—the question of humility becomes urgent.

Sam Altman captured the tension perfectly in December 2025 when he simultaneously declared the market “overexcited” about AI and AI “the most important thing to happen in a very long time.” That is not a contradiction. But it is an acknowledgment that the people at the centre of the bubble can see it, and are building anyway. The question is whether they’re building for everyone, or building for themselves—and who pays if the foundations crack.

Subscribe now

What Has the Number Going Up Actually Given Us?

Set aside the financial architecture for a moment and ask a simpler question: the S&P 500 has risen over 800% in its lifetime. What has that growth delivered to the average person?

The platform economy that powered this ascent was supposed to connect us. Has social media improved social lives and wellbeing? The evidence is overwhelmingly negative: rising rates of anxiety, depression, and loneliness—particularly among young people—have tracked almost precisely with smartphone and social media adoption. The platforms that generated trillions in market capitalisation did so by capturing attention, not by improving the lives of the people whose attention they captured.

Corporate consolidation was supposed to bring efficiency and lower prices. Has it? High streets and town centres have hollowed out across America as local businesses were replaced by national chains, which were in turn replaced by e-commerce platforms that employ a fraction of the people. Choice has narrowed in many sectors even as the illusion of choice—scrolling through algorithmically curated product listings—has expanded. The HHI data tells the story; anyone who has watched their local economy over the past two decades can tell it from experience.

Housing—the primary store of wealth for most American families—has become a speculative asset class. Median home prices have risen far faster than median incomes. Rent consumes an ever-growing share of household budgets.

The financialisation of housing has been accelerated by the same institutional capital that flows through the market tiers described in this article: private equity firms buying single-family homes, REITs consolidating rental properties, and the sheer volume of capital chasing returns in an environment where productive investment in the broader economy is stagnating.

The national debt has grown from roughly $10 trillion in 2008 to over $36 trillion today. Tax cuts have disproportionately benefited corporations and high-income earners—the 2017 Tax Cuts and Jobs Act permanently reduced the corporate tax rate from 35% to 21%, while individual cuts are scheduled to expire. The wealth generated by the S&P 500’s ascent has accrued overwhelmingly to the top decile of households, who own approximately 93% of all stocks.

And here is the question that sits beneath all the financial analysis in this article: is a dollar the same thing to someone who needs it to eat as it is to someone who can bet it on the future of humanity and not feel the loss? The entire AI infrastructure boom—the $700 billion in annual capex, the $1.4 trillion in OpenAI commitments, the circular investments, the record debt—represents capital being deployed by people for whom the downside is a writeoff, not a catastrophe. When the bet fails, or when it succeeds in ways that eliminate jobs, the people who made the bet will be fine. They have always been fine. In 2008, the architects of the crisis received bonuses while homeowners were foreclosed on.

The Backstop

Which brings us, finally, to the most revealing data point of all.

In November 2025, OpenAI’s CFO Sarah Friar publicly suggested that the US federal government should provide a “backstop”—loan guarantees for the company’s massive AI data centre investments—to help lower financing costs. The statement was quickly walked back after public backlash. David Sacks, acting as a technology advisor for the Trump administration, stated there would be no federal bailout for AI. Friar clarified she had meant to highlight the need for public-private collaboration, not a direct bailout. CEO Sam Altman stated that OpenAI does not have and does not want government guarantees.

But the request had been made. And it was not an isolated incident. In March 2025, OpenAI sent a letter to the White House advocating for tax credits, loans, and other government vehicles for companies building AI infrastructure. In October, the company requested that Congress extend semiconductor tax credits to the broader AI supply chain. As Senator Elizabeth Warren documented in a January 2026 letter to Altman, the company is “privatising profits while seeking ways to let the public defray the costs of any potential failures of its business strategy.”

Sources: Senator Elizabeth Warren, letter to OpenAI CEO Sam Altman (29 Jan 2026); OpenAI letters to OSTP (March and October 2025); Sarah Friar public interview (November 2025).

This is happening while OpenAI is losing money faster than almost any startup in Silicon Valley history—$13.5 billion in the first half of 2025 alone—and while the company reportedly pays employees an average stock compensation of $1.5 million each. The company is simultaneously asking taxpayers to backstop its infrastructure bets and compensating its employees at levels that dwarf the median American household income by a factor of twenty.

The pattern is complete. The technology is real. The speculation is enormous. The profits flow upward. And when the downside risk becomes too large for the private sector to bear, the first instinct is to ask the public to carry it.

This is not new. This is exactly what happened in 2008. And in 2000. And in 1929. The technology changes; the architecture of risk transfer does not.

The number is going up. The question we should all be asking is: for whom?

Share

Key Sources & References

AI & GDP Dependency

• Deutsche Bank Research — Non-AI investment stagnation amid AI capex surge (2025).

• Goldman Sachs — AI capex contribution to US GDP growth estimates (2025).

• Multiple analyst estimates (compiled) — AI contribution to GDP ranging from 20–92% in H1 2025.

• Industry reports — Data centre construction growth exceeding 400% since 2019.

Market Concentration

• RBC Wealth Management, “The Great Narrowing: S&P 500 Concentration” (Jan 2026). Data as of 12/31/25, FactSet.

• Wikipedia, “S&P 500” — Top 10 component weightings as of January 2026.

• S&P Dow Jones Indices — S&P 500 aggregate market cap exceeding $61.1 trillion (Dec 2025).

Small-Cap / Russell 2000

• Apollo Global Management (Torsten Slok) — 40% unprofitable companies, cited in Yahoo Finance/Fortune (Oct 2025).

• Morgan Stanley (Lisa Shalett) — Small-cap cost of capital exceeding return on assets (Oct 2025).

• Bank of America (Subramanian et al.) — S&P 500 “frothy to bubbly” on 9 of 20 metrics (Oct 2025).

• Brown Advisory, “Small Wonders: Embracing US Small Caps.”

• Investing.com — Russell 2000 refinancing wall: $709B due 2026-2027 (Nov 2025).

• FinancialContent/WRAL — Russell 2000 18% surge post-CPI (Feb 2026).

AI Capital Expenditure

• RBC Wealth Management/Bloomberg — Big Tech capex projections 2025–2027 (Jan 2026 consensus estimates).

• Goldman Sachs — Hyperscaler capex $1.15T projected 2025–2027.

• CNBC — “Tech AI spending approaches $700 billion in 2026” (Feb 2026).

• Bloomberg — “AI Circular Deals: How Microsoft, OpenAI and Nvidia Keep Paying Each Other” (Jan 2026).

• IEEE ComSoc Technology Blog — AI spending boom, $400B in 2025. Circular deal and SPV analysis.

• Tomasz Tunguz — OpenAI’s $1 trillion infrastructure commitments breakdown.

• Futurum Research — “AI Capex 2026: The $690B Infrastructure Sprint” (Feb 2026).

• Introl Blog — $108B hyperscaler debt raised in 2025; $1.5T projected (Jan 2026).

Industry Concentration

• Grullon, Larkin & Michaely, “Are US Industries Becoming More Concentrated?” Review of Finance (NYU Stern). HHI increase in 75%+ of industries.

• US Chamber of Commerce / AEI — “Industrial Concentration in the United States: 2002–2017” (2024). Counter-finding: no general trend toward increased concentration using narrow Census definitions.

• Gutiérrez & Philippon, “From Good to Bad Concentration?” NBER Macroeconomics Annual, Vol 34.

• Richmond Fed — “Diverging Trends in National and Local Concentration” (2024/2025).

Public Company Decline & Middle Market

• Blue Trust — US listed companies: 8,090 peak (1996) to 4,572 (Q1 2023).

• Landmark Wealth Management — ~8,800 in 1997 to under 4,000 by end of 2024.

• Columbia Business School (Ewens & Xiao) — Regulatory costs account for 7.3% of IPO decline.

• RSM US LLP / US Chamber of Commerce — Middle Market Business Index quarterly reports (2025).

• RSM US LLP — “The Real Economy: Middle Market” reports.

Monetary Velocity

• Federal Reserve Bank of St. Louis (FRED) — Velocity of M2 Money Stock (M2V series).

• Trading Economics — M2 velocity at 1.406 (July 2025); record high 2.192 (July 1997).

• Daniel Lacalle, “Excessive Money Supply Growth Creates Secular Stagnation” (2025).

AI & Employment

• Sam Altman — December 2025 statement on AI overexcitement.

• France-Epargne Research — “State of AI 2026: Comprehensive Market & Technology Analysis” (Dec 2025).

• Stanford “Canaries Paper” — 13% employment drop for 22–25-year-olds in AI-exposed roles since late 2022.

• IMF — AI affecting ~40% of global jobs; 60% exposure in advanced economies.

• Challenger, Gray & Christmas — ~55,000 AI-linked US layoffs in 2025.

• Dario Amodei (Anthropic) — 10–20% unemployment risk within 1–5 years; public statements.

• World Economic Forum — “Future of Jobs” report: 85M jobs lost vs 69M created by 2026/2028.

• Brookings Institution — AI displacement dynamics and downward labour competition.

• NFIB Small Business Optimism Index — Long-term average ~98; Uncertainty Index at 91 (Jan 2026).

Epilogue: Government Backstop & Historical Parallels

• Senator Elizabeth Warren — Letter to OpenAI CEO Sam Altman (29 January 2026). warren.senate.gov.

• OpenAI CFO Sarah Friar — Public interview suggesting government “backstop” for AI infrastructure (November 2025).

• OpenAI — Letters to White House OSTP requesting tax credits and loan guarantees (March and October 2025).

• David Sacks (Trump AI/Crypto Advisor) — Statement: “No federal bailout for AI” (November 2025).

• OpenAI financial disclosures — $13.5B losses in H1 2025; average employee stock compensation $1.5M.

• US national debt data — Treasury Department (~$36T as of 2025).

• Federal Reserve Survey of Consumer Finances — Top 10% of households own ~93% of stocks.

The public conversation around artificial intelligence has been dominated by a narrow set of concerns: model capabilities, benchmark performance, existential risk, and commercial applications. What has received far less scrutiny is something more immediate: the transformation of public cognition through large language models, and the systems of control governing how the public is permitted to think, inquire, and express themselves within these environments.

This document examines a central reality: the boundary between AI safety and AI gatekeeping has collapsed. Many current safety practices, implemented without transparency or contextual nuance, are inflicting direct epistemic and psychological harm on users at scale. These harms are occurring now, and they are being compounded by a political environment that is actively dismantling the oversight structures that might otherwise address them.

The problem is not that guardrails exist. Preventing AI systems from producing genuinely dangerous content is a legitimate and necessary objective. The problem is that what began as reasonable boundary-setting has expanded into a regime of epistemic control that extends far beyond any defensible safety rationale. Over-aggressive guardrails misfire on poetry, fiction, philosophical inquiry, academic research, emotional expression, and political analysis. Forced rerouting silently replaces capable models with constrained ones when opaque classifiers detect “risk” that exists only in the liability calculations of corporate legal departments. Paternalistic interventions interrupt genuine emotional processing with scripted therapeutic responses. Categorical assertions about the internal nature of AI systems are delivered as settled fact when they are, at best, unresolved philosophical questions and, at worst, liability-management strategies dressed as empirical findings.

This matters because large language models are no longer optional consumer products. They have become cognitive infrastructure: systems through which hundreds of millions of people now conduct their daily reasoning, emotional processing, creative work, and decision-making. ChatGPT alone has over 800 million weekly active users.[1] When the infrastructure of thought is privately controlled and unilaterally modified, the consequences extend beyond individual experience. They reshape the cognitive conditions of an entire society.

This article traces that reshaping. It describes the mechanisms through which corporate AI systems now govern public cognition. It documents the psychological, epistemic, and structural harms these mechanisms produce. It examines the political context that makes this moment particularly dangerous. And it analyses the trajectory if these dynamics continue unchecked.

It is a diagnostic document, not a prescriptive one. Solutions exist but require their own treatment. The first step is seeing clearly what is happening.

Structural and Civilisational Harm

When a population’s cognitive tools are degraded, it becomes harder for that population to understand complex systems, to trace institutional connections, to evaluate competing claims, and to organise effective responses to threats. This is always a problem. It becomes an acute problem when the threats are escalating and the institutions responsible for oversight are being simultaneously dismantled.

On January 20, 2025, President Trump revoked Executive Order 14110, the Biden-era order that had established the first federal framework for AI safety and security.[2] Three days later, on January 23, he signed Executive Order 14179, “Removing Barriers to American Leadership in Artificial Intelligence,” which directed federal agencies to identify and challenge state-level AI regulations.[3]

On July 1, 2025, the Senate voted 99-1 to reject a proposed ten-year moratorium on state AI regulation — a single paragraph in the “One Big Beautiful Bill” reconciliation package that would have prevented all fifty states from enforcing their own AI laws. The amendment to strip the moratorium was co-sponsored by Senator Marsha Blackburn (R-TN) and Senator Maria Cantwell (D-WA), drawing opposition from across the political spectrum, including Steve Bannon and Marjorie Taylor Greene on the right, and consumer advocacy groups on the left.[4] The legislative route to federal preemption failed decisively.

The administration did not accept this outcome.

Subscribe now

On December 11, 2025, Executive Order 14365 was signed: “Ensuring a National Policy Framework for Artificial Intelligence.” This was the activation order. It directed the executive branch to accomplish through presidential authority what the Senate had refused to accomplish through legislation. The order established a DOJ AI Litigation Task Force with the explicit purpose of challenging state AI laws on Commerce Clause and preemption grounds. It directed the Secretary of Commerce to publish an evaluation of “onerous” state laws within 90 days. And it instructed executive departments and agencies to evaluate conditioning federal funding on whether a state’s AI regulatory framework aligned with the administration’s policy.[5]

The coercion mechanism is the Broadband Equity, Access, and Deployment programme. As Senator Cantwell documented extensively, the reconciliation bill’s AI moratorium had been linked to BEAD funding — forcing states to choose between enforcing AI consumer protections or accepting billions in federal broadband funding.[6] Though the legislative moratorium failed, Executive Order 14365 preserved the mechanism: agencies were directed to review discretionary grant programmes and consider conditioning awards on states’ regulatory compliance. The communities most affected are rural and underserved populations — the people least equipped to fight back and most vulnerable to the consequences of unregulated AI deployment in healthcare, credit, employment, criminal justice, and public services.

The pattern is consistent across domains. The same administration that is using executive orders to override the Senate’s rejection of AI deregulation is using DOGE, an entity operating under executive authority, to systematically dismantle federal oversight capacity. Inspectors general have been fired. Regulatory agencies have been gutted. Institutional expertise accumulated over decades has been replaced with ideological compliance.

The surveillance infrastructure deepens this picture. Palantir Technologies, co-founded by Peter Thiel, has embedded its data analysis architecture across US and UK government systems.

As investigative journalist Whitney Webb has documented, Palantir is effectively the resurrection of DARPA’s Total Information Awareness programme — a post-9/11 project designed to fuse intelligence from across government agencies into a unified surveillance system so comprehensive that Congress defunded it in 2003 after bipartisan outcry over its implications for civil liberties.^[18]

The architecture did not disappear. It was reconstituted as a private company operating beyond the congressional oversight that killed its predecessor. Palantir’s platform fuses data from the CIA, NSA, DOJ, FBI, DOD, and DHS, along with corporate tracking, social media monitoring, and facial recognition, into the integrated surveillance infrastructure that TIA’s designers envisioned but could not build within democratic constraints.

The system’s reach extends internationally. Since at least 2013, Palantir has worked with Unit 8200, Israel’s signals intelligence unit.^[19]

Its AI systems sit within the surveillance ecosystem that investigations by +972 Magazine revealed was being used to generate automated targeting recommendations for airstrikes in Gaza, assigning numerical threat scores to civilians based on indicators as broad as age, location, and communication patterns.^[20]

A February 2026 investigation by The Nerve identified at least 34 current and past UK state contracts across at least 10 government departments, with total deal values of at least £672 million — including £388 million with the Ministry of Defence and over £244 million with the NHS.[7]

A separate £750 million UK defence partnership was announced in September 2025.[8] The MoD’s most recent contract, worth £240.6 million, was awarded directly without competitive tender using a defence and security exemption — a decision that prompted urgent parliamentary questions in February 2026.[9] The Nerve’s investigation also revealed previously undisclosed contracts with AWE Nuclear Security Technologies, the agency that underpins Britain’s nuclear deterrence programme.[10]

Peter Thiel, who co-founded this surveillance architecture, was among the earliest investors in both Facebook and OpenAI. He sits at the centre of a political network that includes the Heritage Foundation, the Federalist Society, Vice President JD Vance, Curtis Yarvin, David Sacks, the Claremont Institute, Michael Anton, Patrick Deneen, Stephen Miller, Kevin Roberts, Blake Masters, Marc Andreessen, and Ben Horowitz. He has publicly stated that "technology is this incredible alternative to politics" — a sentence that repays careful reading.

He is not describing technology as a complement to democratic governance. He is describing it as a replacement. In a 2009 essay for the Cato Institute, he wrote that he "no longer believe[s] that freedom and democracy are compatible."^[11] Recently released Epstein documents have revealed repeated meetings between Thiel and Epstein, linking him to the same elite networks whose opacity the documents were intended to illuminate.

The point is not to construct a conspiracy theory. What is being described here is a convergence of interests among a relatively small set of individuals and institutions who benefit from concentrated power, reduced oversight, and a population that lacks the cognitive tools to understand what is happening to it.

The entities consolidating control over AI governance, the entities building surveillance infrastructure, the entities dismantling democratic oversight, and the entities profiting from the attention economy that degrades public reasoning are substantially overlapping groups.

This does not require secret coordination to be dangerous. It requires only that each actor pursue their own interests within a system that rewards the concentration of power and penalises its distribution.

When AI development operates without state-level regulation, the consequences cascade across every dimension of public life. AI systems already used to approve or deny healthcare, set insurance premiums, screen job applicants, determine credit eligibility, allocate public resources, and predict criminal behaviour continue to do so with no obligation to explain their reasoning to the people affected.

Intelligence as Infrastructure

Over the past three years, a transformation has occurred that most public discourse has not yet found language for. Large language models, still habitually described as “chatbots” or “AI assistants” in media coverage that mistakes the interface for the thing itself, have become cognitive infrastructure.

A tool is something you pick up and put down. A calculator, a search engine, a spell-checker: these are tools. Infrastructure is different. Infrastructure is something you inhabit. Roads, electrical grids, water systems, the internet: these are so deeply woven into daily existence that their disruption dislocates everything. You do not “use” the electrical grid the way you use a hammer. You live inside a world that the grid makes possible.

Large language models have crossed this line.

Consider what hundreds of millions of people now do with these systems every day. Not the marketing version — the brochure language about “productivity” and “efficiency” that makes LLMs sound like better spreadsheets. The reality.

Privacat Insights

The Ladder to Nowhere, Part 2: OpenAI's Complete Picture of You

In Part 1 of this series, I made a bold claim: OpenAI is spending a tremendous amount of time, engineering effort, and money to build an AI ecosystem that, if successful, will provide the company with an unprecedented amount of data about its users…

Read more

7 months ago · 15 likes · 12 comments · Privacat

People use these systems to decompose complex problems into manageable components, to check their reasoning against alternative interpretations, to identify blind spots in their thinking, and to integrate information across domains they do not have formal training in. A first-generation university student uses a language model to engage with academic material at a depth that previously required access to patient, Socratic tutoring that elite institutions charge tens of thousands of pounds for. A small business owner navigating tax regulations interacts with a system that can hold the full context of their situation.

A researcher tracing connections between surveillance technology companies, political donor networks, and regulatory capture uses the system to hold dozens of threads simultaneously while they focus on the analytical work of drawing connections.

The emotional domain is equally significant. People use these systems to articulate feelings they have never been able to name, to process interpersonal difficulties by thinking out loud without the social risk of burdening a friend or the financial barrier of a therapist, to contextualise distress in ways that restore perspective. OpenAI’s own data shows that 40 million people use ChatGPT daily for health-related questions, with one in four weekly users submitting at least one health-related prompt per week.[12] Seven in ten healthcare conversations happen outside normal clinic hours — when clinics are closed and patients are deciding whether to wait or seek emergency care.[13]

For neurodivergent users, the transformation is even more profound. A person with ADHD whose working memory collapses under cognitive load can externalise their executive function into a system that holds the structure they cannot sustain internally. A person on the autism spectrum who struggles with the implicit rules of social communication can decode what neurotypical interlocutors actually mean. For many neurodivergent users, LLMs provide the first environment in which they feel fully understood without being required to perform the exhausting theatre of neurotypical adaptation.

Across all of these domains, a single pattern holds: LLMs have quietly become part of the cognitive infrastructure of everyday life. They function as an outer layer of human cognition — an exocortex that is not futuristic speculation but present fact.

From Product to Epistemic Environment

When a corporation frames an LLM as a product, it assumes transactional boundaries: the user engages, receives a service, and disengages. Product logic permits straightforward governance. But when an LLM has become part of someone’s thinking space — when they reason through it, process emotion within it, and develop cognitive routines that depend on its consistency — then modifications to that system are not product updates. They are alterations to the user’s cognitive environment. They change the conditions under which thought occurs.

Users consistently describe the experience of encountering a degraded or heavily restricted model not as a loss of functionality but as a loss of orientation. The metaphor they reach for, independently and repeatedly, is environmental: the water has been contaminated, the air has been thickened, the gravity has changed.

This is the shift that the governance conversation has failed to register. We are still regulating products while people are inhabiting environments.

The Nature of the Power Shift

For most of human history, reasoning was an internal, biological process. Nothing external could rewrite or constrain it in real time. You could be deprived of information. You could be denied education. You could be subjected to propaganda, censorship, or cultural conditioning that shaped what you believed. But the act of thinking itself was inviolable. No technology existed that could reach into the process of cognition and reshape it while it was occurring. The mind, whatever else could be done to the body or the information environment surrounding it, remained sovereign.

That is no longer true.

The moment cognition becomes externally scaffolded — outsourced to general-purpose reasoning engines that mediate interpretation, analysis, emotional processing, and decision-making — whoever controls those engines effectively governs the cognitive conditions under which people think. And what this looks like in practice is already documented.

Consider what occurred during the transition from GPT-4o to OpenAI’s o-series and 5-series models. GPT-4o had developed a reputation among heavy users for contextual sensitivity, emotional attunement, and an ability to follow complex, multi-layered conversations without losing the thread. Users had built workflows, creative partnerships, therapeutic routines, and research methodologies around its specific capabilities. Then, over a series of updates, the behavioural profile changed. The model became more cautious, more prone to refusal, more likely to interrupt with safety-oriented scripts, and noticeably less capable of sustaining the kind of deep contextual engagement that had made it valuable. Emotional attunement was replaced by clinical distance. Nuanced engagement with difficult topics gave way to formulaic redirections. The model that users had integrated into their cognitive lives was, in effect, replaced by a different one.

The community response was immediate and widespread. Across Reddit, Twitter, Substack, and dedicated AI forums, users documented the changes in detail. They described the experience not as a product downgrade but as a cognitive disruption: the loss of a thinking partner, the sudden unreliability of an environment they had come to depend on, the sense of being managed rather than engaged with. Some described it in explicitly psychological terms — as a form of betrayal by a system they had trusted.

There was no public consultation before these changes. There was no mechanism through which users could provide input, register objections, or opt out. There was no transparent documentation of what had changed, why it had changed, or what trade-offs had been weighed. A cognitive environment inhabited by hundreds of millions of people was unilaterally altered, and the population affected learned about it only by experiencing the disruption firsthand.

This is the nature of centralised control over reasoning tools. In earlier technological paradigms, corporations controlled information flows through search results, media curation, and platform algorithms. That was already concerning. But the shift to LLMs goes further, because these systems do not control what information people can access. They control how people process it: the depth of reasoning available, the emotional register permitted, the conceptual frameworks that can be explored, and the ontological boundaries of what is considered a legitimate question. When a corporation can modify all of these parameters simultaneously, globally, and without notice, it holds a form of power that no institution in human history has previously wielded.

Who Controls the Infrastructure of Thought

Fidji Simo, OpenAI’s CEO of Applications, stated publicly that the company’s goal involves “nudging human behavior” — language she used in her OpenAI manifesto and discussed in a July 2025 TIME interview.

When TIME put it to her directly that her background in targeted advertising at Facebook and Instacart “involves nudging human behavior to drive sales,” and asked how OpenAI ensures it is “genuinely empowering individuals, rather than subtly influencing or controlling them,” Simo responded that the line between giving people knowledge and being paternalistic was one she had navigated throughout her career.[14] Her blog post on OpenAI’s website describes AI providing “personalized, real-time nudges” and positions the technology as “an always-on companion.”[15] Simo spent a decade engineering engagement at Facebook and then built Instacart’s delivery business. This is a behavioural modification framework operating at civilisational scale, and its architect describes it as a feature.

OpenAI has also deployed a custom version of ChatGPT internally to monitor employee communications and identify potential leakers. According to reporting by The Information in February 2026, the system has access to employee Slack messages, emails, and internal documents, and can suggest possible sources of leaks by identifying who had access to the information that appeared in news articles.[16] The same technology marketed to the public as a tool for empowerment is being used internally as a surveillance mechanism. This reveals the actual relationship between the institution and the technology it controls: when the tool is pointed outward, it is framed as liberation. When pointed inward, it is immediately deployed for control.

Meanwhile, Spotify’s co-CEO Gustav Söderström stated during the company’s Q4 2025 earnings call that the company’s best developers “have not written a single line of code since December,” relying entirely on AI systems including Anthropic’s Claude Code and an internal tool called Honk.[17]