Coinbase Didn't Diversify Away From the Crypto Cycle. It Just Found a New Door Into It.

In 2020, ninety-six cents of every dollar Coinbase made came from one thing: a fee on a trade.¹⁰ When crypto went up, people traded, and Coinbase printed money. When crypto went down, the trades stopped, and so did the revenue. It was the most cyclical business in finance dressed up as a tech company - a casino that only earned when the floor was packed. So Coinbase set out to fix it, and by 2024 it could point to a real number: transaction fees were down to about 63% of revenue.³ The story wrote itself - the company had diversified. The trouble is what filled the gap.

The official story is that Coinbase built a stable, recurring revenue base alongside the trading desk and broke its dependence on the crypto cycle. Look at what actually grew, and a different story appears. The single largest new leg isn't a subscription product at all - it's the yield on the cash sitting behind a stablecoin, paid out under a deal with one company, and exposed to a rate cycle and a draft law that could end it. Coinbase didn't escape the cycle. It changed which door the cycle walks in through.

The shift was real - and the engine behind it is borrowed

Start with the part that's genuinely true, because the thesis only matters if the diversification looks convincing first. Coinbase's subscription and services revenue grew from $1.4 billion in 2023 to roughly $2.3–2.4 billion in 2024 - a 64% jump in a single year.²³ That's not cosmetic. It's a different shape of business than the one that IPO'd. But peel the label off that segment and the largest single sub-component is stablecoin revenue: Coinbase's cut of the yield earned on the reserves backing USDC.³ And here is where 'recurring' starts to wobble. That income is not Coinbase's to set. It flows from a revenue-sharing agreement with Circle, the company that actually issues USDC. In 2024, Circle paid Coinbase $907.9 million in distribution costs under that deal - 54% of Circle's entire revenue.⁹ The biggest pillar of Coinbase's diversification is a number printed on someone else's invoice.

Why does this matter? Because a diversified business is supposed to add a revenue stream that doesn't move with the first one. Stablecoin reserve yield fails that test twice over. It moves with interest rates - the reserves earn what Treasuries pay, so when central banks cut, the yield shrinks. And it moves with crypto adoption - USDC balances swell when people are active in crypto and drain when they leave. Coinbase swapped a revenue line that rises and falls with trading volume for one that rises and falls with rates and adoption. The volatility didn't leave. It just put on a suit and started calling itself recurring.

The quarter that gave the game away

If the diversification were real, a bad crypto quarter would hurt the trading line and the new lines would hold the ship steady. Q1 2026 ran the experiment. Trading volume fell to $202 billion - half what it was a year earlier - and overall revenue dropped 31% year-on-year to $1.41 billion.⁵⁷ So far, so cyclical. The tell is what happened to the supposedly stable segment: subscription and services revenue fell 13.5%, and blockchain rewards - the ETH staking line - were nearly halved to $101.8 million.⁵ The 'decoupled' base recoupled the moment the market turned. Coinbase posted a $394 million net loss in the quarter, against a year where it had earned $2.58 billion in net income.⁵¹ That is not the swing of a diversified company. That is the swing of a cyclical one that found new ways to be cyclical.

Management's preferred metric is product count: twelve distinct products each generating over $100 million in annualized revenue, up from two in 2019.⁷ It's a real achievement, and it's also a sleight of hand. Twelve products that all swell and shrink with the same crypto cycle are not twelve sources of stability - they're twelve expressions of one risk. The same quarter that boasted twelve $100M products also missed earnings by a mile, posting a loss of $1.49 per share against an expected profit.⁷ Counting the products doesn't break the correlation between them.

One law that could hit both legs at once

The sharpest version of the risk is regulatory, and it's specific. The CLARITY Act, now in the U.S. Senate, would bar stablecoin issuers from paying interest directly to holders.⁶ That strikes straight at the yield-and-rewards machinery underneath Coinbase's stablecoin revenue. The strange part - and Brian Armstrong wrote as much on X - is that a rewards ban might paradoxically make Coinbase more profitable in the near term, because Coinbase pays out large amounts of USDC rewards to its own customers, and stopping those payouts keeps more of the yield in-house.⁶ That loophole is genuinely a hedge. But it also reveals how fragile the foundation is: a single bill, aimed at a single mechanism, can reshape the largest growth leg of the business overnight. You don't have that kind of exposure to a law when you're actually diversified. You have it when one borrowed engine is doing most of the work.

Isn't this just a company doing diversification right?

The fair objection is that this is too harsh - that Coinbase is building exactly the kind of business it should, and that no diversification is perfectly uncorrelated. That's true, and worth granting fully. Coinbase isn't standing still: it has staking businesses with $15.2 billion staked by individuals and $8.1 billion by institutions, and it built Base, a layer-2 network that cut median transaction fees by more than 90% to sub-one-cent.⁸ These are real assets that compound with adoption rather than with any single trade. And moving from ~96% to ~63% trading dependence is a structural change that took real work.¹⁰³ The honest counter, though, is the one the Q1 2026 numbers settle: the test of diversification is not whether the mix shifted but whether the new mix breaks when the old one does. It didn't pass. Subscription revenue fell, staking rewards halved, and the company lost money in the same quarter trading collapsed.⁵ A more uncorrelated leg - say, custody fees indifferent to price, or software that bills the same in a bull or bear market - would have held. The reserve-yield engine, tied to rates and adoption and one partner, did not.

Coinbase set out to stop being a casino that only earns when the floor is full. It did real work, and the revenue mix genuinely changed. But the new chips it stacked - reserve yield it doesn't price, owed by a single partner, exposed to a bill in Congress - turned out to ride the same currents as the old ones. The cycle didn't leave the building. It learned to use the side entrance. And the real measure of whether a company has diversified isn't the chart of its revenue mix in a good year. It's whether the mix still looks diversified in the year the music stops.