hope everyone is finding some downtime

all the best from sleepy New Palo Alto #nothingtoseehere

But the clock that’s ticking the loudest has a way of blinding you to the quieter ones — the boring, unglamorous and for now deeply unfashionable companies that are ignoring the noise and just getting on with building some of the best businesses of the next decade in some the least interesting-sounding sectors imaginable. SME banking. Business travel & expenses. Second-hand cars. Workplace training. Even many parts of consulting, legal, financial, tax & accounting and entertainment & marketing services. Say them out loud and the instinctive reaction is a shrug. Nothing to see here.

That shrug is an opportunity.

The middle clock

I wrote in April 2026 about the clock speed problem — the observation that innovation doesn’t move at one speed but several at once. Right now, there is the AI fast clock, where capabilities turn over monthly, the narrative is loud, the capital is unprecedented and valuations run hot. Then there is the slow clock, the science-based bets measured in decades. And in between sits the clock almost nobody writes about and right now almost forgets exists: the thoroughbred companies doing $100m, $250m, $500m+ of revenue, compounding steadily, getting on with their busines and largely insulated from the noise.

The market barely covers them, because they don’t fit the story of the moment. They aren’t “an AI company.” They’re just great companies with products their customers love, growing bigger every day, using technology (and yes AI!) to increase productivity and grow their margins. And that is exactly why they are mispriced in the public imagination.

Three things they have in common

At Phoenix Court we have over 35 of these companies across our funds and when I look across these thoroughbreds, many of the same three features keep showing up.

First, they are attacking genuinely enormous markets. Not “large-ish.” Enormous. The UK second-hand car market alone is worth £80–85bn a year on more than 7.5m transactions. UK SME lending is an ~$87bn annual market and in Germany SME financing sits north of $175bn. Business travel and spending, workplace skills, professional services — each is a multi-hundred-billion-dollar pool or in some cases much much more. These are not niches. They are some of the largest sectors in major economies, and they have been dominated for decades by incumbents who still treat technology as a cost line not a raison d’etre.

Second, they are already at real scale. Every company I have in mind here is doing well over £100m in revenue — in several cases a great deal more — and still growing quickly and efficiently. This is the part people miss. We have trained ourselves to think of “tech” as either a tiny startup or one of the seven-or-ten giants. The interesting action is in the vast, ignored middle: businesses large enough to matter, young enough to still be compounding at rates the incumbents can only dream about.

Third — and this is the punchline — they are still tiny inside their own markets. Most of them are at low single-digit percentages of their total addressable market. A company doing £200m in a £40bn sector has taken half of one percent. The whole business, at scale, in front of a market it has barely begun to touch. The runway isn’t a rounding error on the opportunity; the opportunity is a hundred times the business. Or as we like to say even at scale, early is a state of mind.

Huge market. Real scale. Tiny share. That is the shape of a company with enormous upside that reads, to a casual observer, as nothing to see here.

The type is easy to picture once you have the pattern: challengers in SME banking and payments, marketplaces rebuilding how used cars are bought and sold, platforms remaking business travel and spend, companies reskilling the workforce, and — newer still — the first serious attempts to rebuild the professional services and creative industries. Different sectors, same three traits.

“But is it really tech?”

The most common objection to these companies is a question: is this really tech? Is a business-travel platform tech? Is a used-car marketplace tech? Is a training company tech?

It’s the wrong question, and it’s worth being precise about why.

The right question is not whether a company belongs to a category called “tech.” It is whether it uses technology — and now, overwhelmingly, AI — to build a structurally better business than the incumbent, in a market large enough to be worth winning. By that test these companies are emphatically tech companies, and most of the incumbents they are attacking are not. The incumbents have technology. They do not run on it. That is the difference, and it is the whole game.

Here is the thing most people have backwards. The obsession with AI has quietly become an obsession with the AI companies — the labs, the model layer, the picks and shovels. But the largest transfer of value in this cycle will not all be captured by these novel AI companies at all. It will be also be captured by the “traditional” tech companies that use AI to build genuinely great businesses and take enormous share from incumbents who can’t move fast enough to change at the pace of chan.

Every honest incumbent knows this now. For exmaple, when the CEOs of the largest and most successful banks tell shareholders they are really “technology companies,” they are not being fashionable. They are being smart and defensive. They understand — correctly — that if they don’t become tech companies, someone who already is one will eat their lunch or at least their breakfast.

What should worry the index

We spend a lot of energy fretting that AI will hollow out white-collar work. Maybe it will, it will certainly meaningfully displace white-collar jobs. But the nearer, more concrete threat to a lot of listed incumbents isn’t a frontier lab with an LLM.

It’s more likely to be a 10-year-old private company doing $300m of revenue, growing 40% a year, with a structurally lower cost base and a product customers actually prefer, sitting at 2% of a market the incumbent has treated as an annuity. Multiply that across SME banking, payments, travel, autos, skilling and professional services, and you have the makings of the most significant market share shift in many generations — all happening right now and largely off the public market’s radar given the FOMO around AI.

It is this much less concentrated trend should worry the FTSE, the CAC and the DAX. It should also worry the NYSE and the Dow. Not that AI is coming for everyone in the abstract., it’s actually that the quiet assassins are already here, already at scale, and have barely started. Remember Amazon in retail or Netflix in entertainment - and now Revolut in banking.

The weighing machine, again

Bessembinder’s work keeps showing up in these pieces because it keeps being true: a tiny fraction of companies generate almost all of the value, in public and private markets alike. The job is to be in the room when those companies are being built — and to hold long enough for the weighing machine to do its work.

Right now, some of those companies are being built in the least glamorous corners of the economy, by founders perfectly happy to be underestimated. They don’t trend. They don’t get the covers. They just quietly take share, quarter after quarter, in markets so large that a rounding error is a unicorn.

So just like our slow New Palo Alto summers, there’s nothing to see here.

Which, of course, is exactly where you should be looking. 🙂

And hat tip to for the on-going conversation that prompted this

It still amazes me how little people even inside the innovation economy have really internalised the hard realities of the power law and how much it drives success and returns.

Slide 2 above from the *Fact Pack* we shared with the largest UK asset allocators in November 2024, says everything you need to know about capital allocation in the innovation economy - and actually if you pay attention to Bessembinder's research on the 2.4% of companies that have made up 100% of public market returns ($76TN) since 1990, then you’ll know that the evidence is not just true for private markets but for public markets as well.

Something of course - Norway, Vanguard, State Street and Blackrock figured out some time ago 🙂

Dealroom’s most recent research updates the 8 bps of “VC outliers” who have invested in more than 5 unicorns at seed to look at (in my opinion) the tougher metric of what % of active VCs have invested in more than 11 companies that go on to reach $100m in revenue from seed - just 28 out of more than 20,000 or 0.14%.

All of this may feel pretty abstract, but to make it concrete it’s nice to see how Dealroom benchmarked venture to other fields where outliers occur.

So it’s nice to know that at Phoenix Court with our LocalGlobe fund where we look at back founders at pre-seed and seed we are in good company with Jackie Stewart, Chris Evert, Robert Zemeckis, Jean Piaget and Led Zeppelin.

It’s also nice to know that 2 of those 5 are British as well - and if you’re curious about how it looks for Series A and growth funds (or who I left out on books), then check out Dealroom - it’s a pretty cool way of navigating the context of the power law even beyond our small bubble of venture.

Back to our summer days in New Palo Alto 🙂

But while sticks and stones can break our bones, facts can never hurt us and as we roll into August, our little part of the world - or as we like to say New Palo Alto - is really leaning into it’s position as one of the world's most powerful innovation clusters.

Just In our little part of the world at Phoenix Court we’ve seen this momentum first hand over the last 2 weeks, where just three of our investments announced just over $1bn+ in follow-on capital.

So congrats to the amazing founders and teams at:

• Cusp who raised $450m to support the launch of their AI Materials Foundry 

• Olix who raised $315m and added stellar talent to build token factories fit for the future, not retrofitted from the past 

• Moove who raised $250m to build out the operations infrastructure for autonomous mobility 

We’ve been fortunate to have partnered with all three through LocalGlobe at seed, again with Latitude at early growth and now we can’t wait to work together on their next stage of scaling through our Solar fund.

So here’s to more summers in New Palo Alto and to all the teams moving fast and making things :)

Where I left off

In the last piece, I argued that the personal discipline of a long-term investor — the Clock Speed Problem of separating the voting machine from the weighing machine — is the same discipline Britain has failed at, as a country, three times in a century. Churchill in 1925 chose financial credibility over industrial capacity and got the General Strike as his reward. Thatcher in the 1980s boosted the financial services P&L while quietly liquidating the industrial balance sheet. Today, the pattern is repeating again: pension funds in gilts, FTSE boards hoarding cash, the Bank holding rates high, the Treasury treating productive investment as fiscal risk. Each decision looks individually prudent. Collectively, they are exactly the panicked seller’s portfolio adjustment.

The Macmillan Gap identified in 1931 — the structural failure of British finance to serve British industry — has never been closed. It is the mechanism by which the country’s short clock repeatedly fails to connect to its long clock. And it is why the £143 billion a year that Haskel and Westlake calculate as the cost of UK intangible underinvestment keeps compounding as a loss rather than being reversed as an opportunity.

That was the pattern. This piece is about where the pattern is showing up right now.

There are two specific places where Britain is selling, in 2025, at the bottom of its own long wave. One is the Max Q problem — where scaling companies hit the wall of slow regulation, inaccessible procurement, and absent growth capital, and are forced to relocate, sell, or simply fail to reach the scale their science and commercial traction deserve. The other is sovereign data — where the country is quietly handing over its most valuable public intangible assets to foreign AI foundation model companies for the 2025 equivalent of beads.

Both are happening right now. Both are fixable. And both matter because they are the concrete test of whether the fourth time can be different.

The Max Q problem

Max Q is a term borrowed from aerospace engineering. It is the moment during a rocket’s ascent when dynamic pressure is at its maximum — when the atmosphere is still thick enough to push back hard, but the rocket is moving fast enough that every structural weakness gets stress-tested simultaneously. It is the moment when rockets fail.

For UK companies, Max Q is the scaling phase: the point where the core science has been de-risked, early commercial traction has been proven, and the company needs to accelerate to global scale. It is where thoroughbreds are made and where pipelines either convert or evaporate. And it is precisely where Britain’s voting-machine institutions fail the weighing-machine companies most spectacularly.

The evidence is specific and well-documented. British founders trying to get regulated in their own country routinely wait eighteen months or more for decisions that take three weeks in the US federal process and four months through the Dutch system. Founders who would have sold into US or German corporate markets with relative ease describe UK corporates and public sector buyers as slow, fragmented, and risk-averse to the point of paralysis — the “innovation teams” that exist in many large British institutions turn out not to be empowered to make purchase decisions, so engagement stalls. Non-dilutive financing, which is abundant in Germany and France through institutions like Fraunhofer and BPI, is in the UK fragmented across multiple agencies behind opaque application processes. The capital stack that should exist for scaling companies — equity for high-risk growth, debt where there are revenues and contracts, non-dilutive grants where the science is still maturing — is partial, uncoordinated, and often simply unavailable at the stages where it matters most.

The consequences are predictable. Companies that have genuinely de-risked the hard science and found commercial pull are forced to raise their growth capital overseas, which means pricing their rounds in dollars, accepting American board seats, and eventually moving their executive teams. Some relocate entirely. Others get acquired before they can scale. Either way, the value created by UK science and UK early-stage capital ends up captured by non-UK owners — the same pattern that produced Ferranti, Plessey and the rest of the 1980s losses, playing out again in real time.

The fixes are not mysterious. They are known, tested, and operating successfully in other countries right now. A public register of companies meeting agreed Max Q criteria, so that regulators, procurement teams and investors can identify them and engage proactively. The long promised (over 15 years or more) single Whitehall concierge service offering a front door to the fragmented landscape of regulatory innovation pathways, public sector procurement, and sovereign capital — BPI France as the obvious model. Empowered regulators with time-bound service-level agreements that match the speed of competitor jurisdictions rather than lagging them by a factor of ten or twenty. Objective thresholds — revenue scale, external funding raised, specific technical milestones — that create a streamlined VIP procurement pathway for a cohort of “micro-primes” able to sell into government at scale. Quarterly transparent reporting on how much sovereign capital — from the British Business Bank, British Growth Partnership, Innovate UK, the National Wealth Fund, the National Strategic Science Infrastructure Fund — is actually reaching Max Q companies, and how much domestic and international private capital it is crowding in alongside.

And, critically, sustained convening of founders, investors and policymakers in genuine working sessions rather than ceremonial roundtables, because the frontline intelligence about what is breaking in the system sits with the people running the companies, and the system cannot be fixed without that knowledge flowing into the institutions that shape policy.

None of this is radical. Most of it is standard practice in France, the Netherlands, Singapore, Canada, Australia, and the United States. The scandal is that it isn’t standard practice in the country with the third-largest innovation economy in the world.

The sovereign data problem

There is a second specific manifestation of the two-clock mismatch that deserves attention, because it may be the single largest unacknowledged transfer of national wealth currently taking place: the undervaluation and unprotected use of UK public sector data.

The story is straightforward. The race to train ever-larger AI foundation models has created enormous, and rapidly growing, commercial demand for high-integrity, domain-specific datasets. The UK government happens to sit on some of the most valuable such datasets in the world: anonymised NHS records going back decades, longitudinal educational outcome studies, ONS microdata, Met Office atmospheric and climate records, biodiversity and natural history collections curated over centuries, transport and energy system data at national scale. These are not commodity datasets. They are, in economic terms, sovereign assets comparable to North Sea oil reserves — scarce, high-quality inputs into the production of value in a fast-growing global market.

And they are being treated as if they were free.

The 2018 HM Treasury Mackintosh Report, which looked systematically at public sector intangible assets, concluded that they were “undervalued, underinvested in, and underexploited.” Nothing material has changed since. There is no systematic valuation methodology for public sector data in the UK. There is no central registry of high-value datasets. There is no clear framework for negotiating royalties, licensing terms, or equity participation when such data is used to train commercial AI models. There is no specialist unit in government with the authority to strike partnerships on terms that capture appropriate long-term value for British taxpayers. Decisions are made deal-by-deal, by officials without the technical expertise to evaluate what is actually being given away, and without the political backing to hold out for better terms.

The risk is not theoretical. It is happening now. The same dynamic that gave the Manhattan indigenous peoples twenty-four dollars of beads for their island — an analogy my Phoenix Court colleague Ziv Reichert has used to describe the current moment — is being repeated, at scale, in the data economy. The UK is in danger of handing over its most valuable public intangible assets to foreign AI companies for the equivalent of modern beads: one-off payments, preferential research access, vague promises of capability transfer. The companies receiving the data will use it to build multi-trillion-dollar commercial platforms. The UK will be left with nothing but the taxes it can collect from whatever fraction of the economic activity eventually flows through British corporate entities.

The parallel with 1925 is uncomfortably close. In 1925, Britain prioritised short-term financial credibility over long-term industrial capacity, and watched a century’s worth of productive investment evaporate. In 2025, the risk is that Britain prioritises short-term AI access deals and research partnerships over long-term data asset value, and watches a generation’s worth of publicly-funded intangible capital evaporate in the same way. In both cases, the voting machine — eager to signal modernity, keen to appear commercially engaged, reluctant to complicate things with demands for proper valuation — drowns out the weighing machine of what these assets are actually worth.

The fix, again, is not mysterious. A cross-government stocktake of high-value public sector datasets, led with senior authority and supported by specialist technical advice. A multi-disciplinary expert group drawing on AI, finance, law, data science, economics, ethics, and specific domain expertise — bioinformaticians for health data, climate scientists for Met Office data — tasked with developing valuation methodologies and commercial frameworks appropriate to the specific characteristics of each dataset. A centre of excellence to advise departments on structuring deals, backed by serious commercial legal expertise and with the authority to act as a challenge function when proposed terms undervalue the asset. Improved public-sector literacy on intangible asset valuation, modelled on similar training programmes that operate in the US federal workforce. And a clear signal, at the highest level, that UK public sector data is a strategic national asset — not a cost centre, not a bureaucratic liability, and not something to be given away in exchange for a photo opportunity with a Silicon Valley CEO.

Done well, the UK’s public sector data estate could be structured the way Norway structured North Sea oil: not as something to be rented out as quickly and cheaply as possible, but as the foundation of a long-term sovereign asset that compounds value for generations of British citizens. That is what the weighing machine would recommend. The voting machine is currently doing the opposite.

When Britain held its nerve

The story so far is bleak. But it would be a mistake to read it as inevitable. Britain has not always made this choice. There have been periods when the country held its nerve across the long clock, overrode the voting machine, and reaped the compounding returns. Three of them are worth looking at carefully, because they tell us exactly what the conditions for success look like.

The Victorian era (1851–1914). This is the period the 150-Year Moment paper takes as its benchmark, and it is the clearest example of a country running on the weighing machine. Prince Albert organised the Great Exhibition of 1851 not as a celebration but as a diagnostic — a way of benchmarking British industry against the world and identifying what needed investment. When it made a profit of £186,000 — around £25 million in today’s money — Albert could have handed it back to the Treasury. Instead, he invested every penny in building the South Kensington innovation district: the Science Museum, the Natural History Museum, the V&A, the Royal College of Art, Imperial College. Those institutions are still compounding value 170 years later.

Around Albert, a network of scientist-operators — Henry Cole, Lyon Playfair — translated science into policy and policy into institutions. The civic universities were founded by industrialists who used their own capital: Josiah Mason in Birmingham, Mark Firth in Sheffield, John Owens in Manchester. The whole era was defined by a cultural conviction that building things mattered more than preserving balance sheets. The voting machine said Britain had nothing to prove; it was already the world’s richest country. The weighing machine said the long wave of electrification, chemistry, and mass manufacturing was just beginning. The Victorians listened to the weighing machine, and Britain emerged from the nineteenth century with a compounding base of scientific and industrial capacity that would sustain it for a century.

The wartime period (1939–45). This is the best modern example of what scientist-operators look like when they are given real institutional authority. Henry Tizard — an Oxford chemist and former World War One pilot — drove the entire radar programme, managing both laboratory research and operational deployment, and then led the Tizard Mission that shared Britain’s most closely held technical secrets with the Americans in exchange for industrial scale-up capacity. It was a patient-capital trade on a 50-year horizon: give away the short-term monopoly, secure the long-term alliance.

Patrick Blackett — a Nobel Prize-winning physicist — invented operational research by applying scientific method to military problems. His convoy routing analysis saved thousands of lives and millions of tons of shipping. R.V. Jones ran scientific intelligence, thinking simultaneously as a physicist and a strategist in what became known as “the wizard war.” Alan Turing and the Bletchley Park cryptographers weren’t just solving problems; they were building an entirely new kind of flat, interdisciplinary, mission-driven institution that prefigured every important technology organisation of the next eighty years. Frank Whittle had the jet engine ready years before the Air Ministry took it seriously — an early warning that Britain’s institutional capacity to back its own inventors is always fragile.

The common thread: scientists had operational authority, not just advisory roles. They were in the room where decisions were made, with the power to act on what they knew. The voting machine of military bureaucracy would have slowed all of this down. The weighing machine of actual scientific capability was allowed to run. Without this, Britain would have lost the war. With it, the country emerged with the foundational technologies — radar, jet engines, penicillin, early computing — that defined the post-war world.

The post-war period (1945–51). The Attlee government is the most remarkable case of all, because it invested boldly at precisely the moment the Treasury said Britain couldn’t afford to. The country was effectively bankrupt. Rationing was still in force. The orthodox view — the voting machine — said the priority had to be balancing the books and servicing the debt. Attlee and his ministers listened to the weighing machine instead.

Keynes, the same economist who had warned Churchill about the gold standard twenty years earlier, designed the international architecture at Bretton Woods. Beveridge created the welfare state from first principles. Herbert Morrison — perhaps the greatest institution-builder in modern British history — translated policy ideas into functioning organisations at extraordinary speed: the NHS, the National Coal Board, the Arts Council, British European Airways, the nature conservancies. John Cockcroft built Harwell and launched Britain’s nuclear programme. The government committed to full employment — a commitment Keynes himself described as “a revolution in official opinion.”

And at the centre of it all was Ernest Bevin — the Somerset orphan who had left school at eleven, built the TGWU, organised the General Strike of 1926 against the consequences of Churchill’s gold standard decision, taught himself monetary economics on the Macmillan Committee in 1931, served Churchill as wartime Minister of Labour, and then became Attlee’s indispensable Foreign Secretary. Harold Wilson later said of him: “Ernie sat opposite the Prime Minister at the Long Cabinet table. If a few ministers were talking more than they needed to, and Bevin wanted to get in, he just signalled across. I do not remember a single occasion when the two disagreed in Cabinet, with Ernie acting as the bulldozer whenever necessary.” Bevin helped found NATO, handled the early Cold War, and managed the retreat from empire — while still driving the domestic investment agenda forward. He was the archetypal operator whose authority came from the fact that he knew things the technocrats didn’t.

Keynes and Bevin had been the two most important dissenters from Churchill’s 1925 decision. Twenty years later, they were the two most important architects of its correction. The country listened, finally, to the weighing machine — and the result was a generation of infrastructure, institutions, and technological capacity that transformed Britain.

The four conditions

Every time Britain got it right, four things were present at once:

1. Brave leaders who overrode cautious institutions. Prince Albert against a sceptical parliament. Churchill (the 1940 version) giving scientists freedom and resources. Attlee overriding the Treasury. None of them were reckless — they were disciplined people who understood that not investing was itself a risk, usually the bigger one
   
   

2. New institutions built for the specific challenge. South Kensington. Bletchley Park. Blackett’s operational research groups. Harwell. The NHS. The lesson is always the same: you don’t solve a new problem by asking existing institutions to do something they weren’t designed to do. You build new ones
   
   

3. Technical experts with real authority, not just advisory roles. Playfair translating chemistry into policy. Tizard running radar. Blackett redesigning convoy strategy. Keynes negotiating Bretton Woods. Bevin on the Macmillan Committee and then in Cabinet. In every case, the experts were in the room where decisions were made, not submitting reports that got filed
   
   

4. A cultural belief that building things matters more than saving money. This is the deepest condition. The Victorians celebrated makers. The wartime culture deployed every asset. The post-war generation believed that a better society could be constructed through institutional investment. In each case, productive investment was treated as an act of national purpose, not as a cost to be minimised.

When all four conditions aligned, Britain was extraordinary. When any of them were missing — as in 1925, in the 1980s, and today — the voting machine won and the compounding stopped.

This is why the pattern isn’t inevitable. It is conditional. And the conditions are recoverable.

Next

Max Q and sovereign data are the two places where Britain is selling right now, at the bottom of its own long wave. Both are fixable. The fixes are known, tested, and operating successfully in other countries.

The Victorian, wartime, and post-war periods show that Britain has recovered its nerve before, and can do so again, when all four conditions are present.

But there is a deeper question underneath both of these specific failures. Why does the UK keep making the same category of mistake across completely different domains — monetary policy in 1925, industrial policy in the 1980s, pension allocation today, AI data deals this year? Why do individual reforms keep getting undermined even when the diagnosis is widely understood and the recommendations are on the table?

The answer, I think, is structural. It is about how the Treasury — the institution that dominates every economic decision in the country — actually sees the country it is trying to manage. The next piece is about that.

It is also the piece where I think the argument finally becomes actionable. The three-repetition pattern is diagnosis. Max Q and sovereign data are symptoms. But the Treasury’s operating system is the mechanism. Change that, and everything else becomes possible. Leave it in place, and every individual reform will keep hitting the same wall.

The voting machine is very noisy right now. The weighing machine is quietly compounding. But the machinery Britain uses to choose between them is the real problem — and the real opportunity.

hopefully sharing Act IV of the Clocks series later today or tomorrow with final Act V ahead of our Gathering on the Thames on June 4th - details to follow

Genesis

In The UK’s 150-Year Moment, in spite of feeling a crushing weight of collective pessimism ahead of the November 2024 UK Budget, I tried to set out seven factual reasons why Britain’s innovation position is structurally stronger than it feels. Four of the world’s top ten universities. 1500+ colts and thoroughbreds where there were fewer than ten in 2002. Science clusters from Bristol to Edinburgh and everywhere in between. Fifty science champions already at global outlier scale. Science now 31% of all UK ventures, 3x the 2015 share. Critical future growth cycles — AI, engineering biology, advanced materials, quantum, photonics, space — that map unusually well onto British scientific strengths. A configuration not seen since 1851–1914. This was the piece about the inputs. The argument was that, whatever the sentiment says, the evidence is overwhelming: the raw material is extraordinary and largely unrecognised.

In The Clock Speed Problem, I tried to share some of my reflections on where we are today (Iran, populism, AI supercycle) to my experiences of 30 years surfing the waves of the markets with the observation that innovation doesn’t move at a single speed. AI is spinning fast. Colts (doing $25m to $100m+ in annual revenue) and thoroughbreds ($100m+) are compounding steadily, largely insulated from the noise. Deep science bets in genomics, clean energy, space, neurotech, and quantum are playing out over decades. The job of a long-term investor is to hold across all those timelines simultaneously, because the market’s clock and the company’s clock almost never match. In the short run, the market is a voting machine. In the long run, it is a weighing machine. Ben Graham wrote that in 1949. Bezos was still quoting it fifty years later. I return to it every time the noise gets loud. This was the piece about discipline. The argument was that panics don’t change the underlying thesis because the underlying thesis operates on a different clock to the one markets run on.

Two pieces. Two arguments. Inputs and discipline. With this third piece I’m trying to explore if there is a connection between personal experience and national discipline.

When I talk about holding your nerve as an investor — trusting the weighing machine, ignoring the voting machine, staying long on the 50-year wave — I am describing exactly the discipline that Britain as a country seems to have failed to sustain for over a century. In fact, at least three times in a hundred years, the national voting machine has panicked and sold at the bottom of its own long wave. While the country’s productive capacity - underlying companies, science and most critically talent - was compounding and still had not fundamentally changed, the market was moving and the institutions were moving. Each time, the cost showed up a decade later, by which point iconic political brands had been built that papered over the damage and hid it from long term view.

Sadly, this is the story of how that keeps happening. In the next two pieces I will explore where it is happening right now and some observations on why the Treasury’s operating system makes it inevitable. This piece is about the pattern itself — because once you see it, you can’t unsee it.

Keynes and Merrill: the original clock-speed thinkers

When Keynes drew the distinction between enterprise and speculation in the 1930s, he wasn’t only writing about stock markets. He was writing about an entire system of economic decision-making that had come to privilege the short clock over the long one. Enterprise, for Keynes, was “forecasting the prospective yield of assets over their whole life.” Speculation was “forecasting the psychology of the market.” His worry, even then, was that speculation was coming to dominate enterprise. The voting machine was drowning out the weighing machine.

Charles Merrill, writing to his clients in February 1928 — a full eighteen months before the crash — was making the same point from the other side of the Atlantic. He told them to get out not because the productive economy was broken. American industry was still compounding. He told them to get out because the financial structures built on top of it had detached from business reality. The voting machine had lost contact with the weighing machine. He was early. The voting machine kept voting. When the weighing machine finally caught up in October 1929, his clients were the ones left standing.

Keynes and Merrill were the two great clock-speed thinkers of their generation. They understood that financial markets and productive economies run on different timelines, and that the dangerous moments are always the ones when institutional authority confuses the two. Both were ignored. Both were proved right. Both are worth listening to now.

Churchill in 1925: the panicked seller as Chancellor

Keynes’s clock-speed insight had a specific, devastating application. In April 1925, Winston Churchill — as Chancellor of the Exchequer — returned sterling to the gold standard at the pre-war parity, overvaluing the currency by around 10 percent. He did it because the Bank of England, the Treasury, the City of London, and the international financial establishment told him it was the only credible choice.

The voting machine was unanimous. Montagu Norman at the Bank had been preparing the ground for years. Otto Niemeyer at the Treasury treated it as self-evident. Benjamin Strong at the New York Fed had coordinated US monetary policy to make it possible. Everyone serious, everyone sound, everyone credible, agreed.

Two people disagreed inside the decision-making circle. Keynes, who wrote The Economic Consequences of Mr Churchill three months later — echoing the title of his 1919 pamphlet on Versailles, signalling that this was an act of economic destruction comparable to a punitive peace. And Reginald McKenna, the former Chancellor then running the Midland Bank (the institution that would eventually become HSBC), who had the authority of a practitioner who had seen both sides of the system. A third, Ernest Bevin — who had been arguing against the gold standard since 1917 — was outside the room but would become its most consequential critic.

At a famous dinner at 11 Downing Street in March 1925, Churchill convened both sides to argue it out. Keynes and McKenna made the case against. Niemeyer and Norman made the case for. Churchill admitted afterward that the opponents had the stronger arguments. But McKenna said to Churchill privately, with the fatalism of a man who understood institutional momentum: “There is no escape; you will have to go back; but it will be hell.”

He was right about the hell. He was wrong — as the panicked seller is always wrong — about the inevitability. There is always a choice. Churchill made the wrong one because he listened to the voting machine rather than the weighing machine. Manufacturing was hobbled for the rest of the decade. Wages were cut. The General Strike of 1926 followed directly, with Bevin organising on the labour side exactly what Keynes had predicted on the economic side. America, meanwhile, built the companies of the twentieth century — General Motors, RCA, the national electricity grid. Britain preserved its balance sheet and built nothing.

Churchill later said he had been “misled by the experts.” The experts in question were the Treasury and the Bank of England — institutions whose clock ran on a schedule that had nothing to do with the productive economy they were supposed to serve. It was his worst mistake as Chancellor, and he knew it. The national brand of Churchill the war leader has almost entirely erased the memory of Churchill the panicked seller.

The pattern, three times

Here is the thing about clock-speed mistakes - we never seem to learn how to make new ones. Mistakes are fine - they help with learning, but making the same mistakes again and again should not be happening - certainly not at nation scale.

1925. The long wave was electrification, the automobile, radio, chemical engineering, mass consumer manufacturing. Britain had world-class science and engineering. But the voting machine said sterling credibility mattered more than industrial compounding. Churchill sold at the bottom. America bought the decade.

The 1980s. The long wave was semiconductors, personal computing, software, telecommunications, biotechnology. Britain had Ferranti — which had built the world’s first commercial computer. It had Plessey, whose telecoms division was world-leading. It had GEC under Arnold Weinstock, sitting on decades of cash that could have funded transformation. It had a genuinely strong academic base in computer science. But Thatcher’s tight money and strong sterling oriented the economy toward financial services and away from industrial investment. The Big Bang in 1986 turned the City into a servicing operation for international capital flows rather than a financing engine for domestic productive capacity. Manufacturing output fell nearly 20 percent in two years. Ferranti was broken up. Plessey’s telecoms division was sold to Siemens. GEC’s cash pile was eventually destroyed in a panicked acquisition spree under Weinstock’s successors. Britain sold at the bottom. Silicon Valley bought the decade.

Today. The long wave is AI, engineering biology, advanced materials, quantum, photonics, space — exactly the areas where Britain’s scientific base is strongest. The 1500+ companies at $25m+ revenue. The 176 unicorns. The 287 future unicorns. Phoenix Court, where I work, was ranked the number one venture investor in EMEA by Dealroom this August for backing companies that go on to become thoroughbreds, colts and unicorns — in the top 0.01 percent of funds globally. The raw material is extraordinary. The pipeline is real.

And yet the voting machine is doing exactly what it did in 1925 and the 1980s. The Bank of England holds rates higher for longer. The Treasury’s fiscal rules treat productive investment identically to current spending. UK pension and insurance allocation to UK equities has collapsed from 39 percent in 1997 to just 4 percent today, and less than 1 percent to UK private companies. FTSE 100 companies sit on £150–200 billion in deployable cash. The always brilliant Michael Mauboussin at Morgan Stanley’s Counterpoint Global found that the UK has the second-lowest cash-to-total-assets ratio of any major economy — 7.8 percent, ahead only of Australia. The countries investing most aggressively in frontier technology hold the most cash. Britain holds the least.

Each of these decisions looks individually prudent. Collectively, they are exactly the panicked seller’s portfolio adjustment — fleeing to perceived safety just as the long wave is about to deliver its compounding returns.

The cost is quantifiable. Haskel and Westlake calculate UK intangible underinvestment at £143 billion per year in lost output — roughly £2,144 per person per year. Funds favouring low-intangible stocks returned 2.7 times less to British pensioners than intangible-intensive funds. British savers are already measurably poorer because the national voting machine has been selling at the bottom for a decade.

The Macmillan Gap: a structural feature, not a cyclical accident

Six years after the gold standard decision, in 1931, a government committee published a report on finance and industry. It was chaired by a Scottish judge named Hugh Macmillan, but the intellectual work was done by three of the dissenters from 1925 — Keynes, McKenna, and Bevin, the last of whom had been appointed as the only trade unionist on a committee of bankers and economists and had spent eighteen months teaching himself monetary economics from inside the room.

Their conclusion was simple and devastating: “The relations between the British financial world and British industry have never been so close as those in Germany and the United States.” British banks wouldn’t finance British industry. The gap between the City and the productive economy — the Macmillan Gap — was identified, named, and left unclosed.

It has never been closed.

Ninety-four years later, the Bank of England found in 2015 that British banks still required collateral on more than 90 percent of loans to small businesses. Haskel and Westlake call this “the tyranny of collateral” — if your company’s main assets are software, data, research, and skills (the things you can’t touch), banks won’t lend to you. The $30 billion scale-up funding gap at $100m+ rounds is the same problem in modern form. The diagnosis from 1931 could be republished today with the numbers updated and nothing else changed.

This is not a cyclical problem. It is a structural feature of British capital allocation. It is the mechanism by which the short clock of British finance keeps failing to connect to the long clock of British innovation. And it is why every time the voting machine panics, the weighing machine of productive capacity gets starved of fuel — not because the capacity isn’t there, but because the connective tissue between capital and innovation is missing.

In clock-speed terms: Britain has the weighing machine. It just refuses to plug it in.

The personal discipline and the national discipline

To think through The Clock Speed Problem I mentioned that when the noise from the casino gets loud, I lean back on the NASDAQ chart. I note where we are relative to prior peaks and troughs. And then I ask a different question: have the companies we’ve backed actually changed? Their revenue, their team, their product, their customer relationships. Almost always, the answer is no. The market moved. The companies didn’t.

The national equivalent is the same question. When the Bank of England holds rates for another quarter, when the Treasury tightens fiscal rules, when another pension fund announces it is de-risking further into gilts, the question is not whether the voting machine is voting. It always is. The question is: has the underlying capacity of British innovation actually changed?

It hasn’t. The universities are still world-class. The 1500+ UK colts and thoroughbreds are still compounding revenue. Phoenix Court’s portfolio is still doing the work. The founders building Britain’s next generation of AI, fintech, biotech, space, advanced materials and climate companies are still building, still hiring, still selling to customers who want what they make. The scientific pipeline in AI, engineering biology, quantum, and space is the strongest in 150 years.

The market has moved. The companies haven’t. The country is panicking. The underlying weighing machine is patient.

Next

This is where the pattern stops being a historical observation and starts becoming a choice. Three times in a century, Britain has sold at the bottom of its own long wave. The question is whether the fourth time can be different.

I have the next two pieces in the hopper and plan to dig into where Britain is selling out right now:

1. at Max Q, where scaling companies hit the wall of slow regulation, inaccessible procurement, and absent growth capital;

2. at sovereign data, where the country is quietly handing over its most valuable public intangible assets for the modern equivalent of beads.

Both are happening this year, last year, last decade and beyond. Both are fixable. And both can be understood properly only against the backdrop of a country that has done this before — and could, if it held its nerve, stop doing it.

McKenna told Churchill in 1925 that there was no escape and it would be hell. He was right about the hell and wrong about the escape. There is always a choice.

The voting machine is very noisy right now. The weighing machine is quietly compounding.

So is the question of what Britain will decide to do about it.

Each time, the dominant question becomes: is this the one that changes everything?

It rarely is. And yet each panic reshapes something — confidence, valuations, the narrative of the moment. The chart below which tracks the NASDAQ since my professional introduction to markets with the Netscape IPO (August 1994) tells this story with brutal clarity: every one of those moments looks terrifying in real time and trivial in retrospect, provided you hold your nerve and your horizon. As we like to say at Phoenix Court “early is not a stage, it’s a state of mind”.

This is not an original observation. Keynes, a local in Bloomsbury, writing in the 1930s, drew a distinction between enterprise — forecasting the prospective yield of assets over their whole life — and speculation — forecasting the psychology of the market. His worry, even then, was that speculation was coming to dominate enterprise. Charles Merrill, writing to his clients in 1928, a full year before the crash, was more blunt: the market had become dangerously detached from underlying business reality, and investors were confusing price movement with value creation.

As long term investors allocating capital to the innovation economy, we need to think about this constantly. Because the temptation in moments of panic is to conflate two very different things: what the market is doing right now, and what companies are actually worth over a decade.

The clock speed problem

One of the things I’ve come to understand — slowly, through multiple cycles — is that innovation doesn’t move at a single speed. It moves at several speeds simultaneously. And I’ve come to believe that being a multi-stage, generalist investor is actually the right response to that complexity, not a compromise.

The best way I can explain this right now, is to look at what sits inside a diversified venture portfolio right now:

1. There are AI companies where the clock is spinning fast — new capabilities, new deployment curves, new winners emerging monthly. The market is hot, valuations are elevated, the narrative is loud.

2. There are colts and thoroughbreds: companies generating $25-100m (colts) and $100m+ (thoroughbreds) in annual revenue, compounding steadily, largely insulated from the noise for example fintech infrastructure and some pockets of enterprise software. In our portfolio alone there are 35 companies generating over $100m in annual revenue and 60 doing $25-100m. These are companies outside of the hype cycles who are doing the serious, durable work that doesn’t always make headlines - and in EMEA there are over 700 thoroughbreds.

3. There are science-based companies - sitting on even longer and more durable timelines like advanced materials, genomics, clean energy, neurotechnology, photonics, quantum etc — whose impacts will be measured not in quarterly news cycles but in decades of compounding.

These don’t move together. And that’s precisely the point.

When markets drop 15% in a week because of tariffs or geopolitical shock, the revenue-generating companies keep collecting their receipts. When AI valuations compress because of a new open-source model, the genomics companies haven’t changed at all. Long term and diversified portfolios have natural shock absorbers — not because we’ve engineered a hedge, but because genuine diversification across timelines is structurally different (and additive) from diversification across sectors.

The 50-year wave

The most important frame I keep returning to is this: innovation’s really big waves play out over 50 years or more. Semiconductors and PCs. The internet. Mobile. Cloud. Each of these looked, at various moments, like they were “over” — the bubble had popped, the hype had died, the narrative had moved on. And each time, the underlying compounding continued quietly, indifferent to the panic.

As we’ve been maintaining for 20+ years now, we believe the UK, New Palo Alto, Europe and EMEA are still early in this curve. Not out of sentiment, but because the evidence points that way. The foundational infrastructure of deep tech were born here — including ASML, ARM, BioNTech, DeepMind and many others and now the innovation genie is out the bottle, the next generation of founders are building here as well - and in extraordinary numbers and at scale. The 50-year wave is real, and we are, relatively speaking, still close to the beginning.

The power law dynamics that govern returns aren’t a quirk of Silicon Valley — they’re a structural feature of how innovation scales. Research has shown that 50% of all public market wealth creation since 1926 came from just 90 companies out of 24,240. The same logic applies in private markets: a tiny fraction of companies generate virtually all the value and as Bessembinders more recent research has shown in the public markets as well - although we may be seeing our “attention-economy” move from just a Mag7 to a Mag10 with Anthropic, SpaceX/xAI and OpenAI waiting in the wings. While it’s great to be in the room (or cap table) when these outliers are at scale, the job of venture is to try to be in the room when they’re born — and to hold on long enough for the weighing machine to do its work.

VC has been the best-performing asset class of the last 40 years — not because of luck, but because it is the primary pipeline for the companies that eventually dominate every other index. The top five companies in the world by market cap today all started as venture-backed businesses that didn’t exist fifty years ago.

This is why panics don’t fundamentally change my posture. They may change what I do at the margin — harvesting a position that’s ripe, leaning into a dislocation, accelerating a conversation that was moving slowly. But they don’t alter the underlying thesis, because the underlying thesis operates on a different clock to the one markets run on.

So what can you do when it gets loud out there?

I lean back on the NASDAQ chart. I note where we are relative to prior peaks and troughs. And then try to ask a different question: have the companies we’ve backed actually changed? Their revenue, their team, their product, their customer relationships. Almost always, the answer is no. The market moved. The companies didn’t.

Amazon IPO’d in 1997 with $146m in revenue. In 2000, at the peak of the panic, its enterprise value had collapsed. By 2023 it was worth $1.3 trillion — a 20,000%+ return for anyone patient enough to hold. The patient holders captured that. The panicked sellers did not.

In the short run, the market is a voting machine. In the long run, it is a weighing machine. Ben Graham wrote that in 1949. Jeff Bezos was still quoting it fifty years later. I find myself returning to it every time the noise gets loud.

The voting machine is very noisy right now. The weighing machine is patient.

So are we.

Over the last 20 years we’ve collectively passed through crisis (GFC), divorce (Brexit), pandemic (COVID), war (Ukraine). By last year, even after a new government was elected on the basis of one word (Change) and the promise of a “decade of renewal” the sense of drift and exhaustion had become hard to ignore — inside and outside the innovation bubble alike.

The paper I ended up writing was really a reminder to myself that sentiment and structure are not the same thing. Sentiment may be low - very low; but structure is strong - very strong.

So if you’re interested in reading it and seeing if it brings in some sunshine, here you go “The UK’s 150 Year Moment - Seven Factual Reasons to be Cheerful” and if you’re interested in listening to some of the conversations I’ve started to have around the paper, here are some places to listen - feel free to do at 1.5x :)

• Rest is Money -

• Two Matts -

• Looking for Growth -

Looking forward to a optimistic 2026 and beyond - HNY