The Rake

---

Every bull case for Coinbase eventually arrives at the same metaphor, because it is the only metaphor that makes the valuation work. You are not betting on crypto, the argument goes — you are betting on the exchange, the neutral middleman, the toll booth, the casino. Bitcoin can go up or down; the house collects its rake either way. This is the story that lets an investor who would never buy a volatile, unbacked digital token feel sophisticated buying the company that lets other people buy it. Picks and shovels. Selling jeans to the miners. The house always wins.

The first quarter of 2026 is the quarter that should have ended that story, and it is worth understanding exactly why. Coinbase reported revenue of $1.41 billion, down 30.5% year over year and well short of the $1.51 billion analysts expected. It posted a net loss of $394 million, or $1.49 a share. The cause was not exotic: crypto trading volume fell across the industry, Bitcoin and Ether retraced through the quarter, and spot trading on Coinbase's own platform dropped 37%. When the gamblers slowed down, the company's revenue collapsed in near-lockstep.

Now hold that against the metaphor. A real casino's economics are defensive — the house edge is a fixed percentage of whatever is wagered, and while a slow month means less profit, the casino does not swing to a loss because foot traffic dipped. Its costs are largely fixed against a rake that is structurally positive. Coinbase swung to a nine-figure loss in a single soft quarter. That is not the behavior of a house. It is the behavior of a business whose revenue is a direct, leveraged function of an activity it cannot control and does not generate — the speculative churn of a notoriously cyclical asset class. The casino metaphor was always backwards. Coinbase does not run the casino. Coinbase is the rake, and the rake is worth nothing when nobody is playing.

This essay is about the machinery underneath that loss — about the "diversification" the bulls now point to as the answer, why two of its three legs are not diversification at all but disguised bets on interest rates and on a revenue-sharing arrangement under active attack, and why the single most-cited bullish statistic of the quarter is, on inspection, a confession dressed as a triumph.

The denominator illusion

Begin with that single statistic, because it is everywhere in the bullish coverage and it is the cleanest piece of misdirection in the whole report.

The headline that every Coinbase bull repeated was that subscription and services revenue reached 44% of net revenue — a record mix. The framing is irresistible: Coinbase is becoming a stable, recurring-revenue software company, weaning itself off volatile trading fees, maturing into something durable. Forty-four percent recurring! A record!

Here is what that number actually is. Subscription and services revenue came in at about $584 million — itself down 16% from the prior quarter. Transaction revenue came in at $755.8 million — down 23% from the prior quarter. The reason the mix hit a record is not that subscription revenue soared; subscription fell too. It is that transaction revenue fell faster. When both halves shrink but one shrinks more, the ratio between them climbs, and you can issue a press release celebrating a "record mix" that was produced entirely by the thing going wrong. It is the financial equivalent of bragging that vegetables are a record share of your diet because you lost your appetite for meat even faster than you lost it for everything else. The "diversification" the market cheered was, in significant part, an arithmetic shadow cast by the trading collapse. A genuinely diversifying company shows the stable line growing while the volatile one swings. Coinbase showed both lines falling and called the resulting ratio progress.

This is the oldest trick in the forensic book: a ratio celebrated without its denominator. Always ask what moved — the part you're being shown, or the part you're being distracted from.

The interest-rate business wearing a crypto costume

Now the part that the diversification story most needs you not to examine: what is actually in that subscription-and-services line. Because the largest and fastest-growing piece of it is stablecoin revenue, and stablecoin revenue is not a crypto business. It is a bond fund.

Here is the mechanism. Circle issues the USDC stablecoin; each USDC is backed by a dollar of reserves, largely short-term U.S. Treasuries; those reserves earn interest. Through a revenue-sharing arrangement that has been called, accurately, "overly generous" to Coinbase, Coinbase captures roughly half of the total economics of USDC — and it earns on USDC held off its platform as well as on. In Q1 2026, stablecoin revenue contributed about $305 million, with average USDC held in Coinbase products hitting a record $19 billion and more than a quarter of all USDC in existence sitting on Coinbase. By some measures the company relies on USDC for roughly a fifth of its revenue.

Strip away the crypto vocabulary and look at what that income is. It is net interest earned on a pile of Treasury bills. It rises when interest rates are high and the USDC float is large; it falls when the Federal Reserve cuts rates, because the yield on the reserves is the entire source of the revenue. Coinbase's great "diversification" away from the cyclicality of crypto trading is, in its largest part, a move into the cyclicality of interest rates. Trading revenue is pro-cyclical with crypto speculation. Stablecoin revenue is pro-cyclical with rates. The company has not escaped the boom-bust; it has bolted a second, different boom-bust onto the first and asked the market to call the combination "stability." In a world of falling rates and quiet crypto, both engines throttle down at once — which is, broadly, the direction the most recent quarter pointed.

The generous arrangement that everyone now wants a piece of

And the stablecoin leg has a second, sharper problem, which is that the very generosity that makes it so profitable for Coinbase is an open invitation for everyone else to demand the same deal.

Consider what happened with Hyperliquid. The trading protocol struck a USDC arrangement under which it would capture as much as 90% of the reserve income generated by USDC deposits on its platform — a deal analysts estimated could shift roughly $160 million of revenue away from Coinbase and Circle and into Hyperliquid's ecosystem. Read that carefully, because it is the future of the entire stablecoin-yield business written in a single transaction. The reason Coinbase earns so much from USDC is that, historically, it kept the lion's share of the reserve yield and the platforms holding the stablecoins did not demand a cut. The moment a large platform does demand a cut — and gets 90% — the economics of every other distribution relationship are repriced in the demander's favor. Coinbase's stablecoin windfall depends on a status quo in which it keeps most of the yield. Hyperliquid just proved that status quo is negotiable, and once a thing is negotiable, the negotiation only runs one direction.

So the "durable, diversified" leg of the Coinbase story turns out to rest on two assumptions that are both under pressure: that interest rates stay high enough to make the float lucrative, and that no one else extracts the revenue share that Coinbase itself enjoys. Neither is durable. Both are simply current.

The single number the whole thing tracks

Step back from the line items and ask the question the casino metaphor is designed to suppress: what does Coinbase's revenue actually correlate with? Strip away the segments, the subscription mix, the staking, the custody, and you are left with one stubborn truth — Coinbase's fortunes rise and fall with the price of Bitcoin, because the price of Bitcoin is what governs both halves of the business at once.

When Bitcoin is rising, retail piles in, trading volume surges, transaction fees swell, and the speculative froth pulls more dollars onto the platform, fattening the USDC float and the interest income with it. When Bitcoin falls, the gamblers retreat, volume dries up, fees shrink, and the float deflates. The two engines the company describes as "diversified" are in fact wired to the same ignition switch. This is why the Q1 collapse was so clean: Bitcoin and Ether retraced, and every revenue line that matters moved the same direction at the same time. A genuinely diversified financial company has engines that are at least partly uncorrelated — when one zigs, another zags, and the blend is smoother than the parts. Coinbase's engines are correlated to a single underlying, which means the diversification reduces almost none of the variance that matters. You can call a portfolio of four bets on the same horse "diversified" because it has four tickets. It is still one race.

This is the deepest problem with the toll-booth thesis. A toll booth's revenue depends on traffic, not on the price of cars. Coinbase's revenue depends on the price of the very speculative asset its customers are trading — which makes it not a toll booth on crypto but a leveraged, fee-skimming proxy for crypto itself, sold at a software multiple to people who think they have bought something steadier than the coin. They have not. They have bought the coin's volatility with extra steps and a worse tax treatment.

The cycle Coinbase has already lived twice

There is a reason this quarter rhymes, and it is that Coinbase has run this exact arc before, in public, where anyone could have watched.

In the 2021 bull market, Coinbase went public into euphoria, posted enormous trading revenue as retail mania crested, and was valued as a generational franchise. In the 2022 bust, volume evaporated, the revenue that had looked like a franchise revealed itself as a cyclical fee stream, the company swung to heavy losses, and it cut thousands of jobs to survive. Then crypto recovered, the volume came back, and the franchise story was dusted off and resold as though the bust had been an aberration rather than the other half of the company's permanent heartbeat. The 2026 slowdown is not a new and surprising event. It is the third act of a play the company has already staged twice, and the only thing that changes between performances is the vocabulary used to explain why this cycle is different and the diversification is real this time.

The pattern is the tell. A business that booms with speculation and busts with its absence, repeatedly, across multiple cycles, is not a toll booth that occasionally has a slow month. It is a cyclical, and cyclicals are worth cyclical multiples — low at the top when the earnings look great, high at the bottom when they look terrible. The danger in Coinbase has always been that the market values it as a secular growth compounder near the peak of each cycle, paying a franchise price for what is structurally a leveraged bet on retail risk appetite. The casino story is the instrument that enables the mispricing, because it reframes a cyclical as an annuity.

The competition is coming for the fee, too

One more pressure deserves naming, because it bears on the durability of even the trading franchise. The take rate — the percentage Coinbase skims from each retail trade — has long been generous by the standards of traditional brokerage, a legacy of crypto's early, under-competed years. That premium is exactly the kind of thing competition exists to destroy.

Robinhood and a widening field of brokerages and exchanges now offer crypto trading at thinner margins, courting the same retail flow with lower fees and slicker funnels. The history of every brokerage product in the modern era — equities, options, FX — is the same one-way ratchet: fees compress toward zero as competitors arrive and customers learn what they are paying. Coinbase's retail take rate sits at the top of a slope that has only ever pointed downward in every adjacent market. The bull case implicitly assumes Coinbase holds its premium pricing on the strength of trust and compliance. That is worth something — being the legitimate, regulated venue is a real moat against the fly-by-night exchanges — but it is not worth a permanent fee premium against legitimate, regulated, lower-cost competitors. So the trading leg faces volume that is cyclical and a take rate that is structurally compressing, while the stablecoin leg faces rate sensitivity and revenue-share clawbacks. There is no leg of this business that is not leaning into a headwind it cannot fully control.

What's genuinely working, in fairness

It would be dishonest to leave it there, because Coinbase is not Lucid and it is not a dying company. Several things in the quarter are genuinely, durably good, and a forensic read that ignored them would be propaganda.

The company posted its thirteenth consecutive quarter of positive adjusted EBITDA — $303 million, even in a brutal trading environment. That streak is real and it matters: it shows a cost structure that can stay profitable on an adjusted basis through a downturn, which most of the crypto industry cannot claim. Its crypto trading market share reached an all-time high of 8.6%, meaning that even as the pie shrank, Coinbase took a larger slice — it is winning its category while the category contracts. Regulatory clarity in the United States has, on balance, moved in its favor, and a more rationalized rulebook is a genuine long-term asset for the most compliant large exchange. And the subscription line, denominator illusion notwithstanding, does contain real recurring revenue — custody, staking, and services that are stickier than trading fees. Coinbase is the best-run, best-capitalized, most legitimate company in its sector. None of that is in dispute.

There is even a real long-horizon bull case that does not depend on the casino fiction at all: if stablecoins become genuine payment rails, if tokenized assets migrate on-chain at scale, if Coinbase's infrastructure becomes the regulated backbone of a larger digital-asset economy, then the company grows into something structurally bigger and less purely speculative than it is today. That is a coherent thesis. But notice that it is a venture thesis — a bet on a future build-out that may or may not arrive — not the "you're buying the house, the house always wins" pitch that is used to justify the stock right now, today, at this multiple. The honest long-term bull and the lazy casino bull are not the same argument, and only one of them survives contact with the income statement.

What is in dispute is the casino story and the price it justifies. "Best house in a cyclical, rate-sensitive, revenue-share-threatened business, with a real option on becoming payment infrastructure someday" is a fair description and a reasonable thing to own at the right multiple. "The toll booth that wins whether crypto rises or falls" is a fairy tale, and the $394 million loss is what the fairy tale costs when the tide goes out.

The kicker

The genius of the casino metaphor is that it lets you feel like the house while you are, in fact, the player — because owning the exchange feels one level removed from owning the coin, one rung up the ladder of sophistication, picks-and-shovels instead of gold. But the exchange's revenue is the gambling, repackaged. When the gamblers bet less, Coinbase earns less, and when they stop, it loses money, because there is no rake without a table and no table without players who believe the number goes up. Coinbase has spent two years telling the market it has diversified beyond that truth. Its largest diversification is a bet on interest rates; its proudest statistic is a side effect of the trading collapse; and its most generous profit stream is the one every counterparty has just learned to claw back. Each of those, on its own, is survivable. The problem is that they all point the same way, and they all point the way the gamblers were already heading: toward the exits. A business with one cyclical engine is a cyclical. A business with three engines wired to the same cycle is a cyclical that has spent two years and a great deal of investor-relations ingenuity convincing the market it is something else.

The house always wins, they told you, so you bought the house. But the house posted a loss the first quarter the gamblers grew quiet — and a thing that loses money when its customers go home was never the house at all. It was just the rake, and the rake is only ever as full as the table.