you can think of treasury yields as something close to the baseline price of money

if you can earn 4.0% → 4.8% relatively safely from treasuries, investors are effectively demanding a higher return for holding that government debt

so few important implications follow:

- bond yields ↑ → existing bond prices ↓ as bond prices and yields move inversely

- borrowing costs tend to ↑. mortgages, corporate debt, and other financing become more expensive

- stocks can come under pressure, especially expensive growth/tech stocks, because future earnings are discounted at a higher rate

- the USD can strengthen, because higher U.S. yields can make dollar assets more attractive, although many other factors matter

- crypto can face pressure, because safe government bonds become relatively more attractive and financial conditions tighten

then why would yields go up?

there are several very different reasons, and the reason matters more than the move itself, suppose the 10-year yield rises from 4% to 4.5%. The market might be saying:

1. inflation is going to stay higher

investors demand more yield to compensate for inflation. this is generally negative for long-duration assets.

2. the Fed won’t cut rates as much as we expected.

expected future interest rates rise → treasury yields rise.

3. the economy is stronger than expected.

growth expectations improve, reducing the need for rate cuts. this can push yields higher while stocks may initially remain strong.

4. the government is issuing a lot of debt.

more Treasury supply can require higher yields to attract enough buyers.

5. investors don’t want to own long-term bonds at current prices

selling bonds pushes their prices down and their yields up.

image credits to @neilsethinew

@HyperliquidX is doing $800M in annualized revenue. @Pumpfun is doing $440M.

- both are extremely profitable businesses by crypto standards

- both use substantial portions of their economics to buy their own tokens

- both therefore have a relatively straightforward mechanism through which business activity creates recurring token demand

yet $HYPE trades around a $65B FDV while $PUMP trades around $1.4B.

even after adjusting for differences in growth, revenue quality, token supply and the exact percentage of economics accruing to holders, the valuation gap is so large that i don’t think buybacks or present-day revenue can explain it.

my view is that buybacks create buying pressure, but trust determines the valuation investors are willing to give the asset.

that distinction sounds semantic, but I think it explains a lot of crypto valuation.

the protocol is simply becoming the marginal buyer for an asset whose natural demand is insufficient.

this is why I increasingly think buybacks are an amplifier rather than the source of value itself.

and $HYPE vs $PUMP is probably one of the cleanest examples.

with most tokens, much of the economic relationship is social.

you are effectively underwriting the expectation that the team will continue making decisions that cause the success of the protocol and the success of the token to converge.

Hyperliquid has earned a very high degree of confidence in that relationship.

- perps are the starting point

- HIP-3 expands the universe of permissionless markets and gives Hyperliquid exposure to traditional financial assets

- HyperEVM creates another economic surface around the exchange

- priority fees allow Hyperliquid to monetize something closer to the HFT layer itself: traders paying for execution priority and earlier access to information, with those fees creating another source of $HYPE burn

- lastly, USDC reserve yield creates a revenue stream that is less directly tied to trading turnover

and AQA v2 potentially directs 90% of the yield generated on Hyperliquid’s USDC reserves toward HYPE purchases, with current estimates implying roughly $178M of incremental annual buying pressure.

Pump fun hasn’t, and that is the rerating.

high and higher.

Pumpfun paying KOLs $30k/month to trade 👀

I have my $PUMP thesis, so lets run it back.

Very simply, its cheap on PE vs $HYPE, revenue is consistent, 50% of revenue hardcoded to buybacks until April 2027, therefore up.

I’m not going to argue on the valuation.

I wrote a fairly aggressive critique of the PUMP a while back, it never changed what I think, being a -ev product and one with a good pmf.

The discount is real and I think there’s a trade here along with some friends who have posted their bullish takes on $PUMP, @Flowslikeosmo, @TheDeFinvestor, @sjdedic

And you have @blknoiz06 in it too

What I want to do here in this article is to show that through my analysis the bullish argument becomes weaker than they look.

1. The buyback mechanism is never an accelerant

This is the one that changed my mind and I’ve seen almost nobody state it.

The buyback is denominated in dollars. It’s 50% of revenue, roughly $700k/day. Revenue doesn’t move with the token price. The unlock overhang is denominated in tokens, a fixed count that vests on a schedule, whose dollar value scales linearly with price.

So you have a constant sitting against a line with positive slope.

Run the math.

• At today’s price, unlock sell pressure is ~$484k/day against ~$700k/day of buyback, a genuine net bid of ~$216k/day.

• But the crossover comes at +44.6%.

• Above that, the same schedule that supports you today becomes net supply, because the tokens vesting out are worth more while the dollars buying them back are not.

Let’s look at the unlocks:

• July 2026: ~82.5B PUMP unlocked — 50B team + 32.5B investors. This was the large 12-month cliff.

• After July: the remaining team/investor allocation vests over roughly 36 months, through mid-2029. This should be about 6.875B PUMP/month: ~4.167B team + ~2.708B investors.

• That means roughly 247.5B team/investor tokens remained after the July cliff alone.

• At the current PUMP price of approximately $0.0028, Team + Investors = Total: ≈ $19.25M/month, or roughly $633K/day

• There may also be community/ecosystem supply entering circulation, although the exact schedule there is less certain.

A reflexive mechanism has to strengthen into strength. This one weakens into strength and strengthens into weakness. It’s mean-reverting by construction. You can absolutely make money on that.

It’s a decent floor and a reason the thing hasn’t bled out.

There’s a second-order version that’s worse. The buyback is revenue-linked, and revenue is attention-linked. In a genuine risk-off, memecoin volume falls, revenue falls, the buyback shrinks, at exactly the moment you need it.

2. Pump’s revenue is consistent but demand isn’t.

The volatility argument here is that Pump’s daily revenue vol ranks top 20% of the top 100 apps, more stable than @Uniswap or @Polymarket.

$PUMP have lost market share since Robinhood chain launched.

Pump’s share of total launchpad fees went 80% → 27% in early July, then back to ~50% by mid-August. Two-thirds of market share lost in weeks to Pons and NOXA on Robinhood Chain, then half of it clawed back.

• Then @RobinhoodApp Chain created an entirely new competitive front. $NOXA generated nearly $12M of fees in roughly two weeks and actually exceeded $PUMP daily protocol fees for five consecutive days.

• NOXA deployed ~60,000 tokens, accounting for 65.8% of Robinhood Chain token launches during its active June 30–July 11 window.

• After NOXA stopped launching tokens, $PONS scaled extraordinarily quickly.

• On July 15, Robinhood Chain launchpad trading-volume share was NOXA 43%, Pons 37%, Flap 6%. The following day $PONS generated 11,547 of the chain’s 42,709 new tokens, or ~27%.

• By July 20, Pons had reached approximately 62% of Robinhood Chain launchpad trading volume and 51% of active launchpad wallets.

However, absolute fees still rose 30% to $31.8m over that window. The launchpad market itself grew 77% to $75.4m. The sector growth masked away a losing market share.

Structurally bullish, tactically cautious, slowly stack. A good take to consider by @gumsays on his bearish take on $PUMP

Rising revenue on falling share is often misread in growth investing.

3. Cost of defending volume

@Pumpfun has reportedly been paying traders $20,000 upfront plus $30,000 monthly to defect from FOMO, conditional on closing their FOMO accounts and trading exclusively on Pump.

Sit with what that implies about pricing power. If you have to rent your volume, your take rate is not defended by switching costs. It’s defended by cash.

4. Two things on the horizon

April 2027 is a cliff. The buyback commitment expires then. Whatever support it provides goes to zero on a known date. Markets front-run known terminal dates by roughly two quarters, which puts the anticipation window around Q4 2026 — i.e. inside most people’s holding period for this trade. Nobody I’ve read has priced what the asset is worth the day after.

FDV is $2.4bn against ~$1.1bn circulating. Roughly 53% of supply is still to come. The buyback math above assumes the current unlock rate; it gets harder.

Where I land

However all that said, I don’t think this is a short.

Revenue is real, product-market fit is real, the boost mechanism looks like a genuine improvement, and the market probably is still pricing the equity/token separation as a binary when it’s a spectrum

I think this is a long on memecoin attention, I am not leanging it as a long term cash-flowing business. The buyback thins out when attention dies.

If you’re right about the cycle you’ll make money and you’d have made more in something with cleaner beta to the same thesis and a tougher business model.

The stat I’d watch is the share of launchpad fees. Revenue can rise on a shrinking share in a growing market, and that’s the exact setup that precedes a violent repricing.

All that said, bagholders are simple, none of this changes would change the bull case, simply because, PUMP is cheap relative to HYPE on P/S, revenue has been unusually consistent, current revenue is accelerating, and 50% of revenue continues flowing into buybacks through April 2027.

Not advice. I hold a position.

the economic regime has changed and, more importantly, along with who has the advantage within it

2021 was an unusually powerful environment for people whose primary asset was their ability to generate future income.

capital was cheap, credit creation was strong, asset turnover was high and risk appetite was everywhere.

today the hurdle rate is much higher, but the important part is that the economy did not completely reprice around it.

i think something similar is happening across the economy.

the standard of what we know = high rates → expensive money → less activity. Its true but lazy.

i wish to emphasize that the current rate policy doesn’t suppress the level of economic activity evenly, it aims to suppress movement and movement is disproportionately how young people earn.

four charts 👇

1. the most under-covered statistic of this cycle

my friend @rektdiomedes shares the thing almost nobody looking at labor economics is tracking,

worth reading

if you’re 45 and senior, your income comes from a salary that gets reviewed annually. if you’re 27, historically your income came from LEAVING, the switching premium was the escalator. You didn’t get rich getting raises but you got rich getting a new job.

that escalator has effectively stopped.

• let’s look at some stats, in 2022, switchers got ~18% wage growth against ~7% for stayers. an 11-point gap.

• by Q1 2026 that gap is 3 points, per Bank of America Institute, the smallest in seven years.

• ADP’s version puts the switching premium at 1.9 percentage points, a record low since they began tracking in 2020.

now sit with the second order implication, what you have used to built an entire generation’s career strategy is now negative EV.

BofA data shows job switching actually ticked up, 13.5% of workers in Q1 2026 vs 12.9% a year prior. more people moving but getting paid less to do it.

2. the transaction economy shrank

look at the occupations in that original list of the transaction economy: contractors, agents, mortgage brokers, cleaning companies, mechanics, car dealers, movers-adjacent trades.

nearly all of them are paid per transaction, and a large share are paid per housing transaction.

• existing home sales ran ~6.5m at peak,

• ~4.0m through the 2023–25 trough,

• 4.2m as of May 2026 against a long-run norm of 5m.

• research puts lock-in as preventing roughly 870,000 sales in 2026 alone.

the house is not the unit of economic activity, its the transaction. nobody in that chain gets paid because a house exists. they get paid when it changes hands, commission, origination fee, title, inspection, moving van, the new appliances, the contractor doing the pre-sale bathroom.

870,000 prevented sales from 870,000 bundles of downstream income that never got created. the housing industry is roughly 30% smaller than 2021 by transaction count.

3. the escape hatch also closed

to escape the matrix, people want to build something themselves

that’s what half the people in the in the 2021 cohort did

• drop in share of small-business applicants receiving all the financing they sought: 62% in 2019 → 41% in 2026.

• big banks approve 13–15% of applications.

• roughly 60% of applications from firms under two years old are declined outright

so you can’t get a raise by moving, and you escape by starting something, because the credit box closed at the same time and for the same reason. both exits are downstream of the same policy rate.

4. the aberration that makes the whole debate incoherent

unemployment at 4.2%, near historic lows, layoffs flat.

meanwhile quits stuck at 2.0%, flat for over two years and the hiring rate pinned at 3.3–3.4% since June 2024.

we are in the low hire, low fire economy, and i’d argue it’s the defining aberration of this cycle.

the unemployment rate answers “do you have a chair.” it says nothing about whether the music is playing, and essentially all of young people’s economic mobility, first job, better job, own business, first house, is a flow phenomenon.

the labor market is increasingly disconnected, offering mobility to the already-employed while the unemployed are increasingly isolated.

Its very impressive to see that crypto cards just hit ~$759M/month, up ~19% MoM.

But the more interesting number, imo, is that monthly inflows into stablecoin neobanks reportedly crossed $1B for the first time.

Because those two numbers measure very different things.

Card volume measures spending. Deposits measure financial relationships.

If I deposit $2,000 into a crypto neobank and spend $200 through its card, the $200 payment isn’t necessarily the valuable part.

The remaining $1,800 is.

It can potentially become deposits, collateral, yield-bearing assets, liquidity, or the basis for additional financial products.

That’s why I think we’re entering the second phase of the crypto-card market.

1. The market is already surprisingly concentrated

So RedotPay alone represents >50% of tracked volume.

July card volume reportedly looked roughly like:

RedotPay — $395.1M
ether.fi — $100.3M
KAST — $89.6M
Karta — $58.4M
Wirex — $25.0M
Tria — $16.0M
Plasma One — $15.2M
Others — ~$59.2M

And the top three represent roughly three quarters.

This doesn’t look like a market where 30 interchangeable crypto cards all win.

It looks more like the beginning of consolidation around a few distribution platforms.

But what’s interesting is that they’re increasingly competing on completely different things.

2. Crypto card is becoming a useless category

Put the major products side-by-side and they’re not really selling the same product anymore.

@RedotPay = distribution / accessibility

Its advantage appears to be scale.

RedotPay says its cards can be used across 130M+ merchants, and its ~$395M July volume is miles ahead of everyone else.

It isn’t necessarily trying to build the most sophisticated DeFi product.

It has become one of the clearest examples of crypto, where we see everyday payment infrastructure reaching meaningful scale.

@ether_fi cash = DeFi account + credit

Ether fi has probably the most interesting architecture to me.

If you still remember they started with liquid restaking, which gave it something most standalone crypto cards don’t have from day one: distribution, liquidity, and a base of users already keeping meaningful assets inside the ecosystem.

Cash builds on top of that rather than replacing it. Users can spend supported stablecoins directly or use ETH, BTC-related assets, stablecoins and other supported assets as collateral for card spending, so those assets can remain in the vault while the card becomes the spending interface.

That’s fundamentally different from the usual flow: crypto → sell → fiat → debit card

It’s closer to a flow like: crypto assets → yield/collateral → credit → payments

This is where the fat-app thesis starts to make sense:

Instead of being a single-purpose restaking protocol, ether.fi is stacking more financial services around the same user and the same capital. Restaking gets users in; Cash gives them more reasons to stay by letting them earn, borrow and spend without moving elsewhere.

So the card is really just the last 5% of a much larger financial stack. Distribution brings the user in, the broader product stack retains them, and each additional service captures more of their financial activity.

@KASTxyz = stablecoin neobank / global money account

It’s Revolut/Wise-style global banking.

KAST took almost the opposite route to ether.fi. It started with the card as the wedge, then expanded into USD accounts, salary deposits, transfers and savings.

The interesting part is its progression through: card → payments → salary/deposits → savings → broader financial services

Each step increases retention. A card user can easily switch providers; a user receiving their salary, holding savings and moving money through the same account is much harder to displace.

Stablecoins also give KAST a distribution advantage. Instead of rebuilding banking relationships market by market, it can use stablecoin rails as a common settlement layer while plugging into local payment and banking infrastructure where needed. That makes the addressable market much broader than crypto-native users, particularly for global workers, travelers and users in markets where access to dollar accounts is limited.

So KAST increasingly looks less like another crypto card and more like a stablecoin-native Revolut/Wise. The card is the acquisition wedge, global dollar access is the distribution advantage, and salary/deposits are the retention layer.

@BleapApp = consumer fintech / rewards / invisible crypto

Bleap is f by former Revolut employees, it started with the idea of building a bank account onchain, using a self-custodial wallet underneath but making the experience look increasingly like normal fintech.

The UX, the card its their GTM strength.

Cashback gives 1% base, higher rewards across everyday categories and up to 20% on selected subscriptions and gives mainstream users an immediate reason to try it, while zero Bleap card fees and FX markups reduce friction.

But the more interesting strategy is what happens after acquisition. Bleap is bundling card spending, fiat transfers, on/off ramps, direct debits and self-custody into one account, while abstracting away much of the crypto complexity.

So unlike ether.fi, which starts with crypto-native capital, or KAST, which starts with global dollar access, Bleap starts with familiar consumer fintech UX and pushes the blockchain into the background.

The long-term bet is that users don’t choose Bleap because it’s a better crypto card. They choose it because it’s a better everyday financial product where rewards acquire the user, UX retains them, and crypto becomes invisible infrastructure underneath.

@useTria = rewards + onchain account + highest spend per transaction

Tria is probably the outlier worth watching here. Tria is pushing harder on reward tiers.

Their base cashback ranges from 1.5% to 6%, depending on card tier, with additional staking-based boosts.

Paymentscan puts its average purchase at $180, the highest of the 16 cards tracked and more than 3x the $56 market average.

For comparison, ether.fi sits at $80, RedotPay at $55, while Tria is even above Plasma One at $143 and Avici at $170.

The interesting part is that Tria isn’t winning on raw transaction count. It seems to be attracting users who are comfortable making much larger purchases through an onchain card.

Pair that with its 1.5–6% base cashback tiers and staking boosts, and the strategy starts to look clear by using aggressive rewards to attract higher-value spenders, then pull more of their financial activity into the broader Tria ecosystem.

The bigger question is whether those high-value transactions eventually translate into higher deposits and sticky balances. That matters much more than having the highest average swipe size today.

This is another example of card economics becoming a user-acquisition mechanism for a broader onchain financial account.

@Plasma = chain-native distribution + highest cashback paid

Plasma’s strategy was centered around stablecoin infrastructure, building a chain around moving and using digital dollars, then moved downstream with Plasma One to give that infrastructure a direct consumer endpoint.

Plasma already has an ecosystem of lending, DEX and yield applications from protocols including Aave, Uniswap, Fluid and Pendle. Then, the card extends that stack into real-world spending.

stablecoin → DeFi/yield → deposit → card → merchant

The GTM is also heavily subsidized. Since March, Plasma One has reportedly paid ~$866K in XPL cashback to 21.4K addresses, equivalent to ~2.7% of the $32.4M processed through the cards, plus another ~$141K in stablecoin incentives.

But that spend makes more strategic sense when viewed as ecosystem acquisition rather than card cashback. Plasma can subsidize the card to acquire users who may eventually hold stablecoins, use DeFi, generate transactions and keep more activity on Plasma.

So while KAST uses the card to build a neobank and ether.fi uses it to retain existing DeFi capital, Plasma uses the card to create distribution for the chain itself.

The card gets users in, incentives bootstrap behavior, and the broader ecosystem gives that activity somewhere to go. Plasma is essentially trying to close the loop between onchain stablecoin liquidity and everyday spending.

@OKX = exchange account → payment account

Exchange cards have another advantage entirely by relying on their existing distribution.

OKX already owns the customer relationship, liquidity and trading account.

Its card can turn assets already sitting inside that ecosystem into payment balances.

Depending on region and user tier, OKX currently advertises cashback starting around 2%, with substantially higher VIP rates, while its Singapore card lists zero issuance, recurring and crypto-to-fiat conversion fees for supported stablecoins.

For exchanges, cards aren’t necessarily standalone businesses.

They’re retention products.

Every additional reason not to withdraw assets makes the exchange account more valuable.

@krakenfx = simplicity

Kraken is approaching it much more like a conventional consumer card.

Its current European offering advertises:

• up to 2% cashback

• 600+ supported crypto/fiat currencies

• no transaction fees

• no FX fees from Kraken

• no account fee

Crypto increasingly becomes invisible infrastructure rather than the product itself.

Much like users don’t think about TCP/IP when browsing the internet or email protocols when sending a message, they shouldn’t need to think about stablecoins, wallets, chains, or settlement when paying for something. The winning products may simply feel like better financial accounts, with crypto quietly handling the infrastructure underneath.

3. So what is the crypto card playbook?

The real competition is increasingly happening around:

Distribution = acquiring users efficiently at scale
Deposits = becoming the place users actually keep stablecoins
Yield = generating returns on idle balances
Credit = turning crypto holdings into usable collateral
FX = reducing the cost of cross-border spending
Rewards = using incentives without destroying unit economics
On/off ramps = making money easy to move in and out
Reliability = delivering consistent acceptance and payment execution
Financial products per user = expanding the relationship beyond the card

That’s exactly how neobanks evolved.

4. A word of warning about the cashback system being the least defensible moat

There’s another reason I’m skeptical of ranking these cards purely by cashback from a long term perspective point of view.

Some require VIP tiers, monthly caps, depend on staking or token holdings, offer unusually high rates only for particular merchants and compensate for fees elsewhere.

And cashback itself is trivial for competitors to subsidize.

Competitors can subsidize higher cashback to acquire users, but unless the underlying economics support it, those rates eventually get capped, repriced, or removed.

Over time, the market should converge toward an equilibrium where sustainable rewards ≈ revenue generated per user minus the margin the platform needs to retain.

The real advantage, then, isn’t offering 5% instead of 3%. It’s building enough revenue around deposits, yield, credit, FX, and other financial services to sustain better rewards without subsidizing them indefinitely.

The more interesting metric would be look at net user economics = yield + cashback + credit value - FX - conversion fees - subscription/card fees

That’s what actually determines whether one account is better than another.

5. Privacy cards are effectively another market

Privacy cards sit in a different part of the market because they optimize around a different constraint about identity and regulatory friction rather than yield, rewards or deposits.

That distinction becomes more important as jurisdictions tighten crypto regulation.

@ahboyash talked about recent MAS regulations here:

Singapore is a good example. MAS regulates digital-payment-token services under its payments framework, with licensed providers subject to requirements around areas such as AML/CFT and customer safeguards. Major Payment Institutions providing DPT services operate inside that regulated perimeter.

This creates an interesting market split. Regulated stablecoin neobanks can compete on trust, deposits and deeper financial services, while privacy-focused products compete on minimizing the amount of identity and financial information exposed through the user experience.

The trade-off is that privacy, compliance, reliability and access are increasingly intertwined. As regulation tightens, the demand for privacy doesn’t necessarily disappear, it may become a more distinct product segment.

@Nikitont shared a good list of web3 private cards here:

6. The next phase of crypto card: July’s numbers matter because deposits and spending are growing together

The numbers now look something like:

~$759M monthly card spend
+19% MoM
>$1B monthly stablecoin-neobank inflows
~489K tracked addresses
~132M cumulative transactions

The first wave proved that crypto could be spent through cards. The next phase is about whether these platforms can become places where users actually keep capital.

If spending grows without deposits or retained balances, crypto cards remain payment products. If deposits, credit, yield and other financial activity grow with card volume, they start to look more like stablecoin-native financial accounts.

The card is only the interface. The real value is the financial ecosystem behind it.

the invisible economy thesis is directionally right, but it conflates three separate claims:

1. ai agents will become economic actors

2. they will need machine-native financial infrastructure

3. therefore crypto captures the value

the first two are increasingly plausible

the third is where this analysis gets interesting

agents probably do create demand for infrastructure humans weren’t designed to need like continuous settlement, programmable payments, machine-readable ownership, micro transactions and autonomous coordination

crypto is the best use case suited for this

but if we say that machines can’t use banks, that is too strong. they can operate through apis, corporate accounts and existing payment infrastructure. the real question is whether those systems remain competitive once agents transact with other agents at machine scale

more importantly, machine activity does not automatically translate into token value accrual

if trillions of dollars eventually move between agents, i would find the largest rent seeker. think like what the situational awareness setup adopts.

it could be l1s. but it could equally be stablecoin issuers, agent platforms, compute providers, proprietary data owners, identity networks or applications

and if blockspace becomes abundant and chains become increasingly interchangeable, the machine economy could explode while settlement itself becomes commoditized

eventually, ai creates a new class of economic participant, and those participants will eventually demand infrastructure designed for machines rather than humans

crypto currently looks like one of the strongest candidates for part of that stack

but identifying the infrastructure is easier than identifying where the economics accrue

the machine economy can be a $100 trillion idea while most of the tokens supposedly representing that thesis still go to zero

that’s the distinction investors should probably spend more time thinking about.

The administration can want lower rates, expensive dollar, all time high assets. Warsh can eventually cut. Bessent can lean on whatever levers Treasury has available.

But if the bond market looks at the fiscal situation + inflation + oil + Treasury issuance and is still demanding 5% on the long end, then cutting Fed Funds doesn’t necessarily give Trump the easing cycle he wants.

You could actually get the annoying scenario where the Fed cuts 100bps and the 10Y barely moves because the term premium keeps expanding.

And this is also why the Japan/Yen situation is more important than people think.

Japan is one of the biggest holders of Treasuries. You absolutely do not want a situation where defending the Yen forces Japan to become a seller of USTs at exactly the same time America is issuing enormous amounts of debt.

Bessent seems very aware of this.

The recent US/Japan intervention therefore wasn’t just about saving the Yen IMO. It was partly about protecting the global dollar system.

Basically:

Weak Yen → Japan needs dollars → potential Treasury selling → UST yields higher → US financial conditions tighten → housing/economy gets smoked.

And that’s probably the biggest risk to the bullish thesis IMO.

I agree that other energy resources will eventually replace oil, I’m also bearish oil longer term, but oil isn’t one market.

Transportation, petrochemicals, aviation and industrial demand are different from electricity generation. LNG can destroy gas/coal economics in certain places without necessarily destroying crude demand at the same speed.

And supply matters just as much.

On the topic of private credit, it may have delayed monetary tightening rather than escaped it. The Fed now puts the U.S. private-credit loan market at roughly $1.4T, around one-third of below-investment-grade corporate debt

The Fed says semi-liquid private-credit vehicles now hold about $425B of gross assets, and redemption requests accelerated in Q1 2026, with many managers using their roughly 5% redemption caps. At the same time, bank credit commitments to nonbank financial institutions reached $2.6T, with PE/BDCs/private credit the largest exposure category.

A lot of private borrowers are floating-rate and already have relatively weak debt-service capacity. The Fed notes interest coverage on riskier private firms has been deteriorating, while private-credit borrowers were coming in around ~2x coverage at issuance.

So far the system has absorbed that through extensions, restructurings and patient capital rather than immediate liquidation.

Private credit is now ~$1.4T and roughly a third of below-IG corporate debt. Meanwhile banks’ commitments to PE/BDCs/private credit grew ~17% last year.

It’s that higher-for-longer eventually forces lenders to stop extending and start rationing credit.

Then we will see: weak borrower → restructuring/default → private lender takes losses → underwriting tightens → PE refinancing/M&A slows → employment/capex weakens → broader economy rolls over.

We’re already seeing the first cracks where redemption requests at some private-credit vehicles jumped enough that managers started capping withdrawals, and the Fed now explicitly lists private credit among its major near-term financial-stability risks.

The irony is that private credit may have helped prevent a recession earlier by cushioning borrowers from public-market price discovery.

But in doing so, it may also have extended the cycle and created a larger lagged tightening impulse.

now we should start watching for a healthy rotation into the next generation of L1s and L2s.

because every new chain will become the next trench and new networks create a brief window where:

- liquidity is concentrated

- valuations start extremely low

- ecosystem leaders have not been established

@RobinhoodApp was the latest example.

$CASHCAT became its first major speculative runner. One wallet reportedly turned an $85 purchase into roughly $688K of realized profit while retaining another ~$1.2M on paper.

it reached a ~$105M valuation with only ~$6.6M of liquidity. a few early wallets won enormously because thousands of later buyers became their exit liquidity.

@arc may be the clearest next play.

it is @circle stablecoin-native Layer 1, built for payments, FX, tokenized assets and institutional settlement. Transaction fees are denominated in USDC, and the network integrates directly with Circle infrastructure such as CCTP and Gateway.

people paying $30–$40 for the promise of receiving $1 of Arc mainnet USDC.

Arc ecosystem watchlist

Infrastructure and analytics

@ArcDEXScan — Arc token discovery, market tracking and trading analytics

@jpeg_world_ — One-stop application for navigating parts of the Arc ecosystem

DEXs and liquidity

@TowerExchange — Stablecoin DEX aggregator designed for the Arc ecosystem.

@synthra_finance — DEX positioning itself to support Arc trading from day one.

@lunexfinance — Curve-style StableSwap AMM focused on low-slippage stablecoin trading.

Launchpads and trading terminals

@actfunxyz — Token launchpad already associated with the Arc testnet ecosystem.

@warponarc — Token launchpad and trading terminal.

@AstraPump — Launchpad and DEX offering simplified token creation and trading.

@arcfunapp — Early Arc-native token launchpad.

@ArcPump — Launchpad and DEX concept targeting Arc.

@UseCoinbarrel — Arc-focused token deployment and trading platform.

Payments and financial applications

@BlockradarHQ — Stablecoin wallet and payment infrastructure.

The alpha: Directly aligned with Arc’s intended payment and settlement use cases.

@kolan_xyz — Stablecoin payment infrastructure, previously associated with Hurupay.

@myazahq — Payment and financial infrastructure operating across emerging markets.

@ArrelTechnology — Payment infrastructure project.

@payritHQ — Stablecoin payment application.

@ViFi_Labs — Stablecoin and financial infrastructure.

@usesfxmoneyapp — Cross-border money and stablecoin payment application.

@flezpay — Stablecoin payment platform.

Privacy and infrastructure

@hinkal_protocol — Privacy and confidential settlement infrastructure.

AI and social infrastructure

@thearcosai — Autonomous AI project offering rewards for user interactions.

@Xylonet_ — Social liquidity and payment infrastructure.

Memecoins

$ARCANINE — Dog-themed token claiming early positioning within the Arc ecosystem.

@Brc_meme — Early Arc community memecoin.

@CatBatHatFatRat — Meme project based around the phrase “Cat Bat Hat Fat Rat.”

@pepeonarc — Arc-native Pepe derivative awaiting further confirmation.

@Buildog9 — Dog-themed memecoin openly positioned around community belief rather than utility.

@arc_bat — Early Arc-themed bat memecoin.

$NOAH — Early token appearing on Arc-related trackers.

I noted that nearly every standard market-level valuation metric tracked, currently shows the US market at or near the most overvalued readings in their respective datasets. Shiller CAPE, Tobin’s Q, trend-deviation, and others.

However such historical ceilings aren’t hard limits.

Japan’s market reached roughly 100x earnings at its bubble peak, and China has also reached comparable extremes, meaning a US Shiller CAPE in the mid-40s is “not a magic number” that precludes further expansion.

Two Markets in One

The core observation is that today’s overvaluation is concentrated, closely echoing the dot-com top, when a small number of mega-cap growth names traded at extreme multiples alongside a much larger universe of reasonably or cheaply priced smaller companies.

This bifurcation can be tracked through two long-running ratios:

• Equal-weight S&P 500 (RSP) vs. cap-weighted S&P 500: equal-weight has historically outperformed over the long run, except during large-cap growth booms (late 1990s, 2020, and 2024–2025), when the relationship reverses.

• S&P 100 (largest 100 names) vs. S&P 500: tracks the same phenomenon from the opposite direction and shows a similar pattern across the dot-com era, 2020, and the recent cycle.

Both ratios have begun to turn back in favor of the broader, more equal-weighted market over the past several months, though with significant volatility, including a sharp reversal coinciding with the recent Iran conflict.

This is suggesting that the market hasn’t really picked a direction yet, but is closer to the ceiling of large-growth dominance than the floor.

Illustrative depiction of the large-cap-growth vs. equal-weight performance cycle described by Carlisle, spanning roughly 2015 to the present.

How Wide Is the Value Spread?

Let’s look at some valuation-spread metrics (incorporating quality adjustments), the gap between the most expensive and most undervalued segments of the market currently sits around the:

• 90th–95th percentile when looking at value alone, and in the

• 85th–90th percentile range when quality is included.

These seemed to be comparable to the depths of the 2009 and 2020 drawdowns, even though the broader index has not experienced a comparable decline. This can be attributed to an underappreciated earnings recession in small- and mid-cap stocks from 2022 through late 2025, even as mega-cap earnings rose in a near-straight line over the same period.

Once the multiples get as stretched as they have for the Mag 7, it’s hard to see where else they can go. If you believe in mean reversion, then the smart bet is small and micro value, midcap value.

On AI, Productivity, and Who Captures the Value

The use of large language models personally has been eye-opening, analagous to having a fast, accurate, occasionally wrong junior analyst on call. I am skeptical that this necessarily translates into durable supernormal returns for today’s dominant AI model providers, for two reasons:

• GPUs appear to depreciate functionally much faster (roughly 5–7 years) than prior infrastructure booms like fiber-optic cable or rail (multiple decades), and

• it remains unclear whether one AI model will dominate versus multiple competing models converging in capability, which would compress returns to a more normal, competitive level.

It’s also entirely conceivable that all of this work goes into creating these incredible AI models and all of the value accrues to the consumer and not to the people who create these models. I would then compare this dynamic to how having a website was a meaningful differentiator in 2000 but is now simply table stakes.

On Financing & Mega-Cap Valuations

Though many would, but I do not subscribe to the idea that AI capex financing structure (cash flow vs. debt) is the key distinguishing feature of this cycle versus 2000. The dot-com top was also fundamentally a large-growth-multiple phenomenon spanning companies well outside pure tech, such as Walmart and GE being the old economy names that nonetheless carried extreme multiples at the time simply for being large and growing.

Some of today’s mega-cap valuations such as Apple at roughly 10x revenue, for instance, invite similar scrutiny to historical examples like Sun Microsystems’ Scott McNealy critique of paying 10x revenue without accounting for the full cost structure beneath it.

IPO Supply as a Cycle Signal

There is also unusual concentration of mega-IPOs in a single year. WIth such magnitude of recent and upcoming offerings is a classic late-cycle signal: when private-market sponsors offload growth assets to public markets en masse, it typically means the “marginal buyer” pool, public retail and institutional capital not previously exposed to the asset, has largely been exhausted, historically coinciding with cycle tops.

If we reference, the base-rate research (Michael Mauboussin’s growth-rate analysis) shows that some leading AI model companies’ growth has already exceeded what historical base rates would suggest as sustainable and a pattern he finds both remarkable and, by definition, unlikely to persist indefinitely

my view is that defi lending:
- moves from pooled money markets into modular credit infrastructure, as institutions enter and it becomes too big to fail
- creation of specialized credit markets, risks contained, composability increases
- upside is its growing loan base and fee activity eventually accrue into $MORPHO token value

overview

- morpho is modular lending infrastructure, not just another money market.
- aave is closer to a pooled lending protocol.
- morpho is closer to a credit engine where different markets can be created with custom collateral, loan assets, LLTVs, oracles, and risk curators.
- my bet is that defi lending moves from generalized pools into specialized credit markets: stablecoin lending, ETH-backed borrowing, BTC collateral, RWA collateral, institutional vaults, and fixed-rate lending.

key numbers

- market cap: ~$1.23b
- fdv: ~$1.90b
- active loans: ~$3.61b
- market cap / active loans: ~0.34x
- 30d fees: ~$21.2m
- annualized fees: ~$215.7m
- gross protocol revenue:
q2 2026: $51.0m
q1 2026: $44.7m
q4 2025: $59.3m

thesis

- the main thesis is that morpho can become the backend credit layer for onchain finance.
- lending is unlikely to stay as one generic pool.
- different collateral types need different risk parameters.
- morpho’s isolated market design fits this better than a single shared-risk model
- morpho’s edge is distribution through infrastructure. wallets, exchanges, fintech apps, vault curators, and RWA platforms can build lending products on top of morpho without building the full credit stack themselves.
- this gives morpho a broader market than a normal lending frontend. it can grow even if users do not directly interact with the morpho app.
- the industry setup supports this. stablecoins, tokenized treasuries, BTC-backed borrowing, and institutional onchain credit all need lending infrastructure.
-@Morpho is positioned to be one of the neutral layers underneath that.

bull case

- active loans continue growing from ~$3.6b toward $10b+, and@RobinhoodApp integration should fuel that growth
- morpho v2 expands the market from variable-rate defi lending into fixed-term credit.
- more vault curators and institutional products use morpho as backend infrastructure.
- RWA collateral and tokenized treasuries become more useful inside morpho markets.
- @base and other chains drive more embedded lending distribution.
- governance activates fees or creates a clearer economic link between protocol usage and MORPHO.

bear case

- protocol growth does not translate into token value.
- this is the biggest issue. morpho can be useful while $MORPHO still underperforms.
- $AAVE, $SPK, $FLUID, $EUL and other lending protocols copy modular vault/market designs.
- curator risk increases as more third-party vaults launch.
- lending margins compress as credit markets become more competitive.
- fixed-rate lending adoption may be slower than expected.
- RWA/institutional adoption may move to permissioned systems instead of open defi protocols.
- risk-off markets reduce borrowing demand and increase collateral/liquidation risk.

No sector in history has ever been this large:

The Information Technology sector now accounts for a record 39% of the S&P 500’s total market cap.

This percentage has more than doubled since the 2020 pandemic.

This figure is also now above the 2000 Dot-Com Bubble peak of ~33% and the ~31% peak reached by the Energy sector in the 1980s.

Including internet retailers and digital media platforms such as Amazon, $AMZN, and Netflix, $NFLX, tech now accounts for a record 50% of the S&P 500’s market value.

By comparison, this same group accounted for just 29% of the index at the Dot-Com Bubble peak.

Tech is all that matters now.

[ you get 12% more points boosts and instant upgrade to bronze tier if you trade perps here. ]

Two Different Lenses on the Same Boom

Jim Chanos, the veteran short-seller known for positions against Chinese real estate, Wirecard, and Enron, compared the decade before Netscape’s 1995 launch to the decade after, US economic growth and corporate profit growth rates were essentially unchanged, despite the internet’s obvious transformative effect on individual winners and losers. I believe he meant that AI may reshuffle which companies win without necessarily lifting aggregate growth or profitability.

Val Zlatev, who runs a multi-billion dollar long-short fund focused on semiconductors and hardware, countered from the micro level: across a roughly 500-company hardware universe he tracks, headcount has barely grown (or has declined) over the last three to four years while operating profits have risen sharply, evidence, in his view, that AI’s productivity effect on the technology sector itself is already real and measurable, even before it diffuses to the broader economy.

The Accounting Disconnect

Chanos’s central thesis says that there is a disconnect in profitability accounting between companies selling AI infrastructure (chips, data center equipment, construction), which recognize revenue and profit immediately, and the hyperscalers spending those dollars, who capitalize the costs rather than expensing them.

Much of current hyperscaler capex is sitting in “construction in progress,” not yet depreciated, with Chanos using a conservative 10-year useful life assumption for GPUs in his own modeling.

Figure 4. Illustrative depiction of the revenue/profit recognition gap discussed: equipment sellers recognize revenue immediately; hyperscalers capitalize and defer.

The 2000–2001 Telecom Parallel

I noticed that there is a direct historical parallel to the fiber/telecom capex boom of the late 1990s. S&P 500 operating profits rose roughly 30% from mid-1998 to mid-2000 on a similar capex-driven earnings surge, then fell approximately 40% from mid-2000 to mid-2001 as order books collapsed, a decline comparable in magnitude to the drop seen during the 2008–09 financial crisis, despite 2001 being a mild recession.

There is also a widely repeated but apparently fabricated statistic at the time, that internet traffic was doubling every quarter, which helped justify the over-ordering of network equipment; later analysis found traffic was actually doubling roughly annually.

Illustrative reconstruction of the S&P 500 operating profit cycle Chanos referenced from 1998 to 2001.

However, in today’s environment, that can be tracked in near-real time via open-source token-usage data, rather than relying on a corporate CFO’s quarterly characterization and that current GPU rental prices for chips as old as six to eight years have been rising sharply since around January, reversing a prior year-over-year decline of 20–30%, evidence that token demand is currently outstripping supply rather than being manufactured by a false narrative.

Where Chanos Is Actually Short

Chanos is not actually short the chipmakers themselves. His short exposure targets business models he views as structurally low-return regardless of the AI growth outcome: Bitcoin miners that pivoted to data center hosting, and “Neocloud” GPU rental businesses such as CoreWeave.

These companies are effectively equipment-leasing/finance businesses wrapped in technology branding, and even under generous best-case assumptions (including a 10-year chip life), the unrounded returns on capital pencil out in the mid-single digits.

You want to belong what the chips produce, not where the chips reside.

However, a select number of Neoclouds, such as CoreWeave’s reported software/optimization layers and Nimbus’s roughly 50–60% revenue exposure to (higher-margin) inference rather than 100% hyperscaler-contracted capacity.

The broader “landlord” category lacks differentiated technology.

Data Center Capex: Deceleration, Not Decline

I have began to notice a slowing rate of change in hyperscaler capex growth. They are still expanding in absolute dollars, but decelerating from roughly 80% year-over-year growth toward the mid-40% range, per data cited from JPMorgan.

Illustrative reconstruction of hyperscaler capex deceleration discussed across sources, based on cited JPMorgan data center capex research.

Memory: A Genuinely Different Cycle?

On DRAM and NAND memory, historically among the most brutally cyclical, commoditized corners of semiconductors, this cycle differs structurally.

DRAM and NAND prices have risen four-to-fivefold, driven by the shift from simple chatbots to reasoning models (requiring more token storage), expanding context windows, and the recent emergence of AI agents, which consume memory more intensively than prior workloads.

Even with unlimited clean-room capacity, equipment makers such as ASML and Applied Materials are physically constrained to roughly 30–35% annual bit-supply growth, and memory manufacturers, characterized as a much more cautious, cyclically-scarred 60-to-70-year-old leadership cohort than Silicon Valleym under-invested in new clean-room capacity during the 2023–2024 downturn, leaving the industry unable to quickly meet the current surge in demand.

The consequence flagged is that memory now represents roughly 50% of bill-of-materials cost for PCs and smartphones, up from roughly 20% historically, squeezing device-maker margins at sub-5–6% operating margins and forcing price increases on consumers. PC and smartphone unit volumes were cited as down in the mid-teens percentage range this year as a result, an unusually sharp decline for typically flat, slow-growth categories.

Chanos has never made a single dollar shorting DRAM companies over a 40-year career, the sector’s history of repeated boom-bust mistiming, and he generally avoids pure-play memory and semiconductor-cyclical shorts as a result, though he does see a valuation gap worth noting between legacy CPU names trading at elevated revenue multiples in increasingly competitive markets versus dominant, oligopolistic AI infrastructure names that are comparatively cheaper despite stronger market positions.

Valuation Context

However I push back the common critique that the current AI-adjacent valuations that broadly resemble 1999–2000.

Chanos noted that 1999–2000-era extremes (e.g., Cisco at roughly 160x earnings) are not matched by today’s leading AI infrastructure names. Nvidia at roughly 15x estimated 2027 earnings and Broadcom at roughly 12x estimated 2028 earnings as comparatively reasonable, even as semiconductor capital-equipment suppliers (ASML, Lam Research) trade at richer multiples (~35x) for similar growth rates, given their structurally capped ~30% growth ceiling.

This is where the AI bubble comparison becomes more complicated. On the surface, the setup does look and feel like the dot-com bubble: one dominant technology narrative, extreme index concentration, massive capex buildout, investor belief in a new economic paradigm, and a market increasingly willing to underwrite years of future growth upfront. Semis have become the picks-and-shovels trade of the AI cycle in the same way networking, fiber, servers, and internet infrastructure became the backbone trade of the late 1990s. The chart pattern, the concentration, and the reflexive enthusiasm all rhyme with 1999.

My Take

All that said, I remain long broad semiconductor exposure (with call options on Marvell and Photronics, and naming Nvidia as the highest-conviction idea), while separately flagging Meta’s AI strategy as a name that I am bearish on, though not currently short.

My overall idea is that even if this is ultimately a bubble, being outright short semiconductors may not be a risk worth taking while the boom continues, though we do see that a genuine accounting disconnect exists between semiconductor profits and model-company losses that, in my view, is unsustainable over a multi-year horizon.

What a day in markets.

Earlier this week, the market was staring at a potential Korea-led VaR shock. KOSPI had sold off aggressively, volatility was exploding, and the pressure was concentrated in the same memory and AI infrastructure names that had led the entire rally. That made the setup dangerous.

That was the key risk going into $MU earnings.

MU had become the US confirmation vehicle for the global memory trade.

Instead, MU delivered a monster print.

Revenue came in around $41B, with earnings far ahead of expectations. More importantly, the quality of the print supported the core memory thesis: AI demand is not slowing, memory supply remains tight, and customers are increasingly willing to lock in supply through long-term commitments. That is exactly what the bulls needed to see after the Korea-led selloff.

This changes the setup into July.

My prior view was that if MU reported well, it would create a strong technical segue into July, especially because institutions that had de-risked during the Korea shock would need to re-risk back into the move. That is now the base case. The selling in semis earlier this week increasingly looks like a positioning reset rather than the start of a fundamental breakdown.

That does not mean the trade becomes risk-free. The memory complex has already moved a lot. Some names are still extended, and the first knee-jerk after earnings may be violent in both directions. But the fundamental signal is now stronger than the technical fear from earlier in the week.

[ you get 12% more points boosts and instant upgrade to bronze tier if you trade perps here. ]

The next part of the trade may also broaden.

After a large memory move, I would expect investors to start looking beyond the obvious HBM and DRAM leaders. The next leg may come from neoclouds, memory pooling, CXL, interconnects and AI infrastructure bottlenecks. If MU confirms that memory demand is tight, the market will naturally ask: who else benefits from the same constraint?

That keeps names like $ALAB, $CRDO and $PENG relevant. It also strengthens the case for neoclouds like $NBIS, especially if compute demand continues to accelerate into inference-heavy workloads.

One of the more interesting side notes is the report that Anthropic may have been cash-flow positive in 2Q. But if true, it is important.

The reason is simple: it suggests premium inference pricing can work.

I would not have expected frontier labs to show this level of monetization this early. The common view was that model providers would burn heavily for much longer because inference costs would overwhelm revenue. If a leading lab can generate positive cash flow, even temporarily, it changes the market’s perception of AI unit economics.

That is bullish for compute.

If inference can be monetized profitably, labs will not reduce compute demand. They will scale it. More usage, more agents, more coding workflows and more enterprise deployment all mean more demand for GPUs, memory, networking, storage and neocloud capacity.

This is why the AI infrastructure thesis remains intact.

The market spent the last few days worrying about a global margin call. MU reminded investors that the fundamental cycle is still very real. The Korea shock was painful, but it may have cleared out some of the weakest positioning before the next leg higher.

My view remains the same.

Memory remains the core of the current AI trade, but after this move, the better risk-reward may shift toward the second-order beneficiaries: neoclouds, memory pooling, interconnects and infrastructure names that solve the next bottlenecks.

Good luck.

Executive Summary

• The Fed’s first meeting under Chair Kevin Warsh delivered a sharply hawkish dot plot, even as oil prices fell ~30% from the prior meeting, and as peak hawkishness that will unwind as disinflation data confirms.

• Warsh’s real agenda may be less about near-term rate path and more about flattening the long end of the curve to support AI-related corporate bond issuance, housing affordability, and Treasury debt management, a shift from balance-sheet-driven to price-driven monetary transmission.

• On AI capex: short-seller Jim Chanos is net long the semiconductor index but flags a profitability accounting disconnect, chip/equipment sellers recognize revenue immediately while hyperscalers capitalize the same spend, a dynamic that preceded the 2000–2001 telecom profit collapse.

• Memory (DRAM/NAND) bull Val Zlatev counters that physical capacity constraints (wafer fab equipment growth capped near 30%/year) make this cycle structurally different from prior memory gluts, even as PC/smartphone unit volumes fall double digits on cost pass-through.

• Value investor Tobias Carlisle sees small/micro-cap value at a 90th-plus percentile valuation discount to large-cap growth, comparable to 2009 and 2020 lows, setting up a potential decade-long mean-reversion rotation, as occurred 2000–2015.

• The resilient nominal S&P 500 is partly an illusion: real (dollar-devaluation-adjusted) corporate profits have been flat since 2021 even as the index posts record nominal highs, with the gap explained by an accelerating decline in dollar purchasing power.

Reading the First Warsh-Era FOMC Meeting

The Core Tension: Hawkish Dots, Falling Oil

The headline surprise was a hawkish dot-plot pivot arriving at the same time oil prices have fallen dramatically from the prior two meetings, a combination that I think is pretty incoherent.

The chart below is an illustrative reconstruction of the dynamic discussed: a notable widening between the most hawkish and most dovish committee dots, even as the median moved higher.

The hawkish pivot comes at a time when oil prices are down dramatically from their last two meetings.

The working theory is that the hawkish dots may have come disproportionately from non-voting regional Fed presidents using their only tool; the dot, to push back on a new chair perceived as upending the institution, rather than reflecting the committee’s genuine reaction function. Inflation swaps were noted to have round-tripped back to pre-meeting levels within a day, and the 2-year yield, after spiking toward 4.20%, gave back gains almost immediately.

Peak Hawkishness Thesis

Several inflation components were cited as already rolling over or distorted by base effects:

• shelter/owner’s equivalent rent inflation is seen as artificially elevated due to a base effect from a zero-mark during the prior government shutdown;

• tariff-driven goods inflation is seen as fading as tariffs are unwound;

• and one-year inflation swaps have reportedly returned to pre-war levels, tracking oil.

My overall view is that April/May/June’s hot CPI prints and the hawkish dot plot represent a lagging reaction to a backdrop that is already disinflating.

Illustrative real-yield spread (2Y nominal minus 2Y breakeven) used as a proxy for tightening global liquidity conditions.

Why this matters for liquidity

A rising spread between nominal yields and breakeven inflation implies investors are increasingly compensated in real terms for holding bonds, a reversal from 2021, when real yields were negative.

When this spread widens, it means real interest rates are rising. That usually means, money is becoming more expensive after adjusting for inflation.

We can link this to dollar strength, a multi-decade-base breakout in USD/JPY, and a chain of effects they expect to run from FX volatility to rate volatility to credit spreads to equity multiple compression, though they noted high-yield credit spreads have, so far, barely moved.

The Real Agenda is Flattening the Long End

A recurring theme was that Warsh’s substantive priority may not be the short-term rate path at all, but rather using rate policy (rather than balance sheet tools) to bring down long-end yields. The rationale offered:

• Mortgage affordability: most US mortgages price off the 30-year/long end; lower long rates could unlock a generational housing turnover as boomers sell into better affordability.

• AI buildout financing: hyperscalers and AI infrastructure players (large corporate and even SpaceX bond issuance) increasingly fund capex via debt issuance priced off the long end, not Fed funds.

• Treasury debt management: with an estimated 85% of Treasury issuance concentrated on the front end, a flatter curve could allow Treasury Secretary Bessent to term out debt and reduce interest expense.

Market Structure

It’s easy to be hawkish when stocks are at all-time highs and everything’s rosy but it’s a different story when you have inflation still well above target and equities are down 10–15% and you have to act in that regard too.

The balance sheet is singled out as most consequential, notably, the yield curve has so far moved in the opposite direction (flattening) from what a balance-sheet-shrinkage stance would imply, suggesting other forces are dominating in the near term.

On positioning and seasonality: record short positioning in SOFR futures from the fund category was cited as a setup for a potential short squeeze if hikes fail to materialize. Separately, quadruple witching, summer seasonality, elevated household cash on the sidelines, and a forthcoming Jackson Hole symposium in August were all flagged as factors likely to produce a low-volatility “grind” through summer, with the first real test of the new policy framework more likely in the fall.

Crypto & MicroStrategy

I also want to discuss MicroStrategy ($MSTR), arguing the company is not insolvent from a going-concern basis but is testing the market’s patience: continued share issuance to buy Bitcoin, rather than building cash to cover dividend and debt service obligations, has pushed preferred-equity yields higher and left the capital structure trading at distressed levels. The fix is straightforward, raise cash to demonstrate liquidity coverage, but management has so far chosen the opposite path.

Bitcoin’s recent underperformance is a function of capital rotating toward more productive AI infrastructure investment, a reversal of the post-2008 “There Is No Alternative” (TINA) dynamic that favored non-yielding stores of value when corporate capital investment was scarce.

the first is revenue

the second is volume

OpenAI, Anthropic and Google still dominate revenue because enterprises buy trust, support, security, procurement, and integration. that part of the market moves slowly and is heavily shaped by cloud GTM, solution architects, and existing enterprise relationships

but OpenRouter is showing the leading edge of volume

by may 2026, chinese open-weight models were roughly 61% of tokens consumed on OpenRouter. four of the top five models were Chinese. Llama, once the default open-weight leader, has effectively disappeared from the top rankings

the cost-performance center of gravity for open-weight inference has shifted east

[ you get 12% more points boosts and instant upgrade to bronze tier if you trade perps here. ]

DeepSeek-V4-Pro pricing at roughly 12x below GPT-5.5 makes the point clear. the mass market does not always pay for the absolute frontier. it pays for the cheapest model that is good enough for the task

openRouter usage shifted heavily toward code, with programming rising from roughly 11% of usage at the start of 2025 to more than 50% by mid-2026. coding is high-volume, repeatable, and price-sensitive. that is exactly where cheap, capable open-weight models compound

premium reasoning is a high-margin niche. cheap, open, good-enough inference is becoming the volume layer

if the best open weights are increasingly Chinese, large enterprises will hesitate. export controls, procurement risk, data sensitivity, and political optics matter far more to a Fortune 500 buyer than to a startup trying to cut inference cost

so the market likely bifurcates, where startups chase performance per dollar, while enterprises stay with approved Western vendors longer:

- that delay creates a temporary distortion, not a permanent moat

- the real investment implication is that margin is moving away from the model layer

- it accrues above the model through distribution, workflow ownership, and application lock-in

- it accrues below the model through cloud, inference routing, optimization, and compute infrastructure

the model itself still matters, but in the volume tier it is becoming increasingly replaceable

[ you get 12% more points boosts and instant upgrade to bronze tier if you trade perps here. ]

Surfcash is the strongest play on crypto-native local payments and wallet-history-based credit, basically, your wallet becomes your bank.

think of it like a crypto-native local banking, where you spend stablecoins locally, pay through QR rails, convert to cash, and use wallet history as credit history.

here is how they compare to some of Solana’s neobanks by use case

tier S

@surfcashx — Best for crypto-native local payments and wallet-based credit

tier A

@JupiterExchange — Best Solana DeFi super-app

@Revolut — Best mainstream fintech bridge

@KASTxyz — Best global stablecoin card

tier B

@AviciMoney — Best self-custody card neobank

@offgridcash — Best frontier/off-grid banking angle

@xplaceapp — Emerging consumer finance/app layer

so i compared it with the chart of $MU, that also explains why $HYPE is starting to track US equity momentum more closely.

if semis, memory, AI infrastructure, and pre-IPO narratives remain strong, Hyperliquid benefits from the same risk appetite. more attention means more activity. more activity means more fees. more fees strengthen the token narrative.

the CFTC chair mentioning Hyperliquid has effectively pushed perpetual futures and onchain derivatives into the mainstream policy conversation

these products are increasingly being discussed in the same breath as the future structure of US derivatives markets.

the bull case for $HYPE is that it has become one of the cleanest live examples of what a 24/7, onchain, capital-efficient derivatives venue can look like.

the counterargument is that NYSE, CME and other major exchanges can eventually move toward 24/7, 365 trading too. technically, they can. but they have not, because legacy markets are constrained by regulation, settlement rails, intermediaries, clearing, market structure, and institutional habits.

that is the important part.

the US is clearly trying to onshore markets that previously lived offshore. if crypto perps become a regulated product category in the US, the market will start repricing the entire sector.

the immediate risk is competition from regulated venues. but the second-order effect is more interesting: regulatory attention validates the market structure itself.

this is where $HYPE becomes different from most crypto tokens.

its value is tied to exchange activity, revenue, buybacks, attention, and now potentially regulatory narrative:

- its the verticalization of crypto perps, prediction markets, pre-IPO speculation, equity beta, and 24/7 trading infrastructure.

- it is a bet that the future of speculation moves toward faster, global, always-on venues, where users want exposure to crypto, equities, private markets, event contracts, and macro narratives in one place.

the risk is that regulation cuts both ways.

if the CFTC opens the door for US-regulated perps, Hyperliquid gets validation, but it also faces a much more serious competitive landscape from compliant US venues.

whether @HyperliquidX can stay product-wise ahead, even as the regulated world starts copying the market it helped prove

Yields are lower across the curve, with the 10Y around 4.47% and the 30Y still below the 5% level. That matters because when long-end yields stop pushing higher, the pressure on long-duration growth stocks starts to fade. That gives AI, semis, memory, and high-beta tech more room to breathe.

With the rally overnight, the selling pressure that dominated earlier in the week started to fade, and once that supply was absorbed, the higher-quality AI and semiconductor names bounced quickly.

The market was not breaking because the AI story died. It was breaking because positioning was crowded, liquidity was thin, and selling flows had control. Once those flows tapered off, the market had room to squeeze.

Semiconductor

The semiconductor rebound was broad, but the leadership was still selective.

$NBIS remains one of the cleaner names in the AI infrastructure trade. The stock has had “plot armour” because investors are treating more like a direct AI compute / GPU infrastructure beneficiary. The Nasdaq-100 inclusion catalyst also matters because index inclusion can create mechanical demand and improve institutional visibility. That gives NBIS a stronger support narrative than most of the smaller AI infrastructure names.

$ALAB also participated strongly. This is one of the more important AI connectivity names because the market sees Astera as a beneficiary of the buildout in AI servers, data centers, and high-speed interconnect. When investors go back into “AI infrastructure picks,” ALAB tends to respond quickly because it sits close to the hardware bottleneck layer. The stock is also expensive, so it works best when yields are calm and risk appetite is improving.

Read more

the macro backdrop is not meaningfully different from yesterday.

US yields remain elevated but stable, with the 10Y around 4.53% and the 30Y near 5.01%. for now, the bond market is not sending a fresh shock signal.

that matters because the equity move still looks more flow-driven than fundamentally driven.

[ you get 12% more points boosts and instant upgrade to bronze tier if you trade perps here. ]

√ asia

turning to asia, the key market i am watching is KOSPI. asia has opened weaker, but not in a disorderly way.

KOSPI is down around 3%, which is softer than usual but still better than what the EWY move was implying overnight. Nikkei is also only down roughly 1.2%. given the scale of pressure in US equities, especially in high-beta and crowded longs, asia’s response so far has been more controlled than feared.

KOSPI has become one of the most important risk barometers in this AI-led market because of its heavy exposure to semiconductors, memory, and korean retail leverage. it is no longer trading like a normal domestic equity index. in this cycle, it behaves more like a leveraged expression of the global AI and memory trade.

the recent move in KOSPI has also been extremely stretched. the index had rallied hard into this pullback, driven by AI optimism, memory recovery, and strong participation from local retail investors. that made the market vulnerable to a sharp air pocket once US tech started unwinding. so weakness should be seen more as a positioning reset after a huge move.

the korean market has been more resilient than expected given the leverage unwind in the US. korean equities are typically more sensitive to retail leverage, so the fact that KOSPI is not breaking down more aggressively suggests the forced-selling impulse is contained for now. retail may have survived another day.

Kioxia also continues to trade well. its strength is consistent with the broader memory theme, even though not every memory-linked stock is behaving equally. SNDK initially showed strong relative performance in the US session before fading some of its gains, which suggests investors still want memory exposure, but they are not willing to chase indiscriminately.

so for now, i see KOSPI as a key confirmation market. if it stabilizes from here, it supports the view that this is still a controlled positioning unwind. if it fails and starts leading asia lower, then the AI risk unwind probably has further to go.

√ US

the key takeaway from the US session was the late recovery in semiconductors. names like SMH, MU, and NVDA staged a strong rebound into the close and nearly finished at the highs of the day. that was not necessarily a sign that the market suddenly became fundamentally bullish again.

earlier in the session, the downside move likely created a large expected sell imbalance into the close. but because a meaningful portion of the drawdown was recovered before the closing auction, there was less actual selling left to execute. funds that had pre-sold into the expected close imbalance then had to buy back or reduce hedges, which created a net bid into the end of the session.

√ AI, earnings, IPO

within AI and semis, the dispersion remains important. memory, nuclear, neoclouds, and several adjacent AI infrastructure themes continue to show signs of stress, with many names making new local lows. my base case is that these weaker pockets remain challenged through the month. they are likely to underperform both on rallies and on further drawdowns because capital is becoming more selective.

that said, NBIS stands out. the stock tested the low 200s area and held up better than many other high-beta AI names. if the broader market stabilizes, NBIS could be one of the names that sharply outperforms on the next risk-on bounce. the setup is not risk-free, but the relative strength is notable.

the next major event is Oracle earnings. ORCL is important because it sits at the intersection of cloud, AI infrastructure, and capex expectations. the market may use this print as an opportunity to reassess the company’s AI backlog, cloud growth, and financing needs. the key risk is not just whether numbers beat or miss, but whether investors become concerned that Oracle needs to raise capital to fund its AI infrastructure commitments.

on the AI demand side, the release of the Mythos/Fable model is also worth watching. early usage appears to be meaningfully heavier than Opus in terms of token consumption, potentially around 2–3x. if that trend holds, it reinforces the idea that frontier models are still becoming more compute-intensive, not less. that is structurally supportive for compute demand, but it also raises the question of whether the economics can scale smoothly.

SPCX IPO concerns have also been hanging over the market. my view is that the market may have over-discounted this risk. if investors have already positioned defensively around the IPO overhang, then the removal or digestion of that supply could become a positive catalyst rather than a continued headwind. sometimes the event people fear becomes fuel for a rally once the positioning is already too negative.

√ gold, bitcoin

the two assets that still concern me are Gold and Bitcoin.

Gold has been surprisingly weak. what stands out is that it has not performed well in either risk-on or risk-off conditions. that usually means the market is treating it less like a safe haven and more like a source of liquidity or an asset being de-risked. if XAU breaks the prior 4,120 area, the next major support zone looks closer to the 3,800–4,000 range. until Gold can reclaim momentum, it remains vulnerable.

Bitcoin is also not acting well. it continues to hover around the 60k area and has struggled to reclaim the range cleanly. i would not be aggressive shorting here because the risk of a violent squeeze, especially if equities stabilize. but historically, when BTC keeps pressing against the underside of a range without reclaiming it, the eventual resolution is often lower.

semis recovered because of flows, the fundamentals did not change. asia is holding better than expected, which reduces immediate systemic concern. however, weakness in Gold and Bitcoin suggests liquidity conditions are still not fully healthy.

for now, i would avoid treating every bounce as confirmation that risk is back. strong relative names can still work, but weak themes making new lows should continue to fade until they prove otherwise.

the reason it matters is that Japan has been a major cheap-funding source for global risk assets. when yen funding becomes unstable, it can affect Nasdaq, AI stocks, crypto, credit, EM assets, and even gold.

you can understand this as confluences for your trades/positions.

where yen is now 👇

[ you get 12% more points boosts and instant upgrade to bronze tier if you trade perps here. ]

—

USD/JPY is again near the 160 zone, which markets treat as Japan’s unofficial intervention line. Japan already intervened after USD/JPY hit around 160.7, pushing it back toward 155, but the yen later weakened again. that tells you intervention is losing durability unless it is backed by real BOJ tightening.

the BOJ is now expected to raise rates from 0.75% to 1.0% at its jun 16 meeting, which would be the highest policy rate level since 1995. markets are pricing a high probability of this move.

at the same time, Japan’s 10-year JGB yield has risen to around 2.7%+, a very high level by Japan’s modern standards. that matters because rising domestic yields make it more attractive for Japanese capital to stay home instead of buying US assets.

japan is trapped between three bad choices:

1. raise rates to defend the yen

this helps the yen, but pressures JGBs, banks, government debt servicing, and Japanese equities.

2. keep rates too low

this supports Japan’s bond market, but risks more yen weakness and imported inflation.

3. intervene in FX

this can slow the move, but unless policy changes, the market eventually tests them again.

the bigger issue is that Japan is moving from a stable cheap-funding regime into a more unstable tightening/intervention regime.

this affects global markets because of how the yen carry trade works:

→ borrow yen cheaply

→ convert to USD

→ buy higher-return assets.

those assets can include US tech, Nasdaq, credit, private credit, emerging markets, crypto, and high-beta trades.

when yen weakens slowly, the carry trade works. But when the yen suddenly strengthens because of BOJ hikes or intervention, leveraged investors may need to unwind: sell risk assets, buy back yen, reduce exposure.

that is why a yen shock can look like a global risk-off event as I analyse the market impact by asset

us equities / Nasdaq / AI stocks

Most vulnerable. AI has become the market’s main growth narrative, but it is also duration-heavy and liquidity-sensitive. If JGB yields rise and US yields stay high, valuation pressure increases. AI and semiconductor stocks already saw sharp stress recently as rate fears hit the market.

crypto

crypto usually behaves like high-beta liquidity. if carry unwinds, crypto can get hit faster than equities because leverage is higher and liquidity is thinner. BTC may hold better than alts, but alts/perps would likely be more exposed.

gold

mixed. If yen stress causes global risk-off, gold can benefit as a safe haven. but if the shock comes with USD strength and rising real yields, gold can initially sell off. Gold likes crisis, but dislikes strong dollar + higher yields.

nikkei

very vulnerable if the yen strengthens sharply. a weak yen helps Japanese exporters; a sudden yen reversal hurts earnings expectations and also pressures leveraged domestic positioning.

us Treasuries

complicated. if Japanese investors repatriate capital, that can pressure US Treasuries. but if global risk-off gets serious, Treasuries may still catch safe-haven flows. the direction depends on whether the dominant fear is “liquidity unwind” or “recession/risk-off.”

my take

if BOJ hikes aggressively or coordinates intervention and USD/JPY drops fast, that can trigger carry-trade unwind pressure. that is bearish for Nasdaq, AI, crypto, and high-beta risk.

so the yen is currently a macro volatility trigger, not just an FX pair.

here are the key levels/events:

USD/JPY 160 — intervention danger zone.

USD/JPY 155 — if it breaks lower fast, carry unwind risk rises.

Japan 10Y JGB yield above 2.7–2.8% — shows domestic bond stress.

BOJ June 16 meeting — hike to 1% is the big event.

Nasdaq/SOX reaction — semiconductors are the cleanest AI-liquidity signal.

DXY + US 10Y yield — if both rise with USD/JPY, pressure on risk assets increases.

yen weakness has been helping global risk for years, but yen instability is now becoming dangerous. i

- if Japan tightens too slowly, yen crisis risk grows.

- if Japan tightens too fast, carry unwind risk grows.

- either way, the market is becoming more fragile.

short usdjpy

my original target was $100/point.

after running the comp against lighter’s tge, i think that was too conservative.

i’ve moved my target closer to $185/point and even that may not fully price in what variational is becoming.

[ you get 12% more points boosts and instant upgrade to bronze tier if you trade perps here. ]

the easy comp is $LIT launched at $3b+ fdv with a 25% airdrop.

but the problem with comparing variational to lighter is that variational is not really playing the same game.

- @Lighter_xyz is a perp dex.

- @variational_io is trying to become an onchain global brokerage layer, with perps as only one part of the stack.

that changes the valuation framework.

most perp dexs are fighting for the same crypto-native leverage traders:

- variational’s real edge is its RFQ liquidity model.

- orderbook dexs need native liquidity for every market. every new market creates fragmentation, market maker cost, and incentive spend.

- variational uses an rfq model where liquidity can be routed through aggregated sources across cexs, dexs, and tradfi dealers.

that is why they can list hundreds of markets without needing to bootstrap every orderbook from zero.

this is the part most people are missing.

rwa markets are higher-value assets, with a larger addressable market, wider user base, and potentially better monetization than crypto perps alone.

so if rwa volume ramps, revenue should ramp with it and that leads to the second part of the thesis.

variational already has around $3mil fedcoins / usdc sitting in its onchain treasury wallet.

to my knowledge, this is one of the first dexs building a real treasury before TGE, with docs indicating a minimum 30% of revenue used for buybacks.

are people really going to instantly dump their airdrop if there is potentially a $10m–$15m twap buyback coming at any time?

maybe some will.

but that is a very different setup from the usual tge, where the only bid is retail exit liquidity.

this is why i think $100/point is no longer the bull case, it may be the base case.

if the market prices variational like a normal perp dex, $100/point makes sense.

my target is now closer to $185/point.

and if the rwa thesis actually works, billions.