COCompany report
Coinbase Global, Inc. COIN
The bet you're really making is that people keep buying and selling crypto through Coinbase, and increasingly leave their cash sitting in its dollar-coin and staking accounts, where Coinbase earns a slice whether or not the market is hot. Underneath that, you're betting that when the buying and selling dries up, as it has the last three quarters, the money Coinbase makes on all that parked cash is enough to hold the business up. Right now it is going the wrong way: the money coming in fell 18.5% from a year ago, and the company has lost money three quarters running as crypto cooled. You pay about 41 times what it earned in its last full good year, 2025, and on the last twelve months there is nothing to put a multiple on, because it lost close to a billion dollars, its profits swinging from huge to negative inside a single year.
Key data
COIN · price with moving averages
Source: market data.
The business
Coinbase runs the largest US crypto exchange. Retail customers buy and sell bitcoin, ether and hundreds of smaller tokens and pay a fee on each transaction, richest on small retail orders, and that fee is the historic engine. Institutions transact too, at thinner fees, for the volume. The second engine, which management is pushing hard, is everything that is not a transaction: interest earned on the USDC dollar-coin reserves, a revenue share with Circle, whose 2025 IPO handed Coinbase a large one-time gain, plus rewards from staking customers' tokens, custody, and interest on customer cash. The moat is regulatory trust and liquidity, Coinbase being the compliant, public, audited venue cautious institutions and retail use, and liquidity begets liquidity. What the customer actually touches is an app where a $1,000 bitcoin buy quietly costs a few dollars in spread and fee. In 2025 the company spent $742M on acquisitions, with regulatory filings expected to detail the targets.
The numbers
Revenue peaked in Q3 2025 at $1.87B, declined 5% to $1.78B in Q4, then fell harder to $1.41B in Q1 2026 and $1.22B in Q2 2026, down 18.5% from a year ago and 13.7% from the prior quarter.
| Quarter | Revenue | Net income | Diluted EPS |
|---|---|---|---|
| Q2 2025 | $1.50B | $1.43B | $5.14 |
| Q3 2025 | $1.87B | $0.43B | $1.50 |
| Q4 2025 | $1.78B | −$0.67B | −$2.49 |
| Q1 2026 | $1.41B | −$0.39B | −$1.49 |
| Q2 2026 | $1.22B | −$0.36B | −$1.36 |
Net income tells a starker story, from a $1.43B profit in Q2 2025, itself flattered by the one-time Circle gain, to three losses running. Those losses are not all operating: Coinbase now holds crypto on its own balance sheet and marks it down as prices fall, so a chunk of the red ink is non-cash. Strip it and the operating business is near breakeven, not deeply underwater. Nothing new has printed since late August, so the three-quarter slide is still the last word, and board member Fred Wilson has kept trimming into it.
| Year | Revenue | Net income | Diluted EPS |
|---|---|---|---|
| 2021 | $7.84B | $3.62B | $14.50 |
| 2022 | $3.19B | −$2.62B | −$11.83 |
| 2023 | $3.11B | $0.09B | $0.37 |
| 2024 | $6.56B | $2.58B | $9.48 |
| 2025 | $7.18B | $1.26B | $4.45 |
| 2026, 1H to Jun | $2.63B | −$0.75B | −$2.85 |
The whipsaw is the whole history. The 2021 bull market threw off $3.6B of profit, the 2022 crash a $2.6B loss, the 2024 rally $2.6B again. Revenue compounded 31% a year from the 2022 trough, but diluted shares grew 29% over the same stretch, from 222M to 287M, so per share the picture is far tamer. Stock pay runs about 17% of revenue, and that is the tension under the reported cash: 2025 free cash flow was $2.4B, a 34% margin that looks strong, yet much of what keeps it that clean is handed to employees in stock that dilutes owners rather than in cash. The market prices the reported losses and the cooling volumes; what it may underweight is that even mid-winter this business throws off real cash and its non-transaction revenue is becoming a floor a bull market is not required to fill. The print that settles it is next quarter's split between transaction and subscription revenue.
Management
Founder Brian Armstrong still runs it, took $9.7M in 2025, under 1% of the year's profit, and holds a founder's stake, so his incentives sit with the price. Capital allocation is where the record bites: Coinbase repurchased $790M of stock in 2025 at an average near $279, and the stock now trades at $184, so that buyback is down about a third, cash spent near the top of the cycle. Insiders have bought nothing in twelve months against 39 sales worth $9.2M, small beside a $49B market value but one-directional, led by longtime board member Fred Wilson. The balance sheet is a real strength: about $34 a share in cash, more cash than debt, so the company can bleed for a while without stress.
How it fails or surprises you
The winter deepens. Transaction revenue is the swing, and it has fallen to $1.22B in the latest quarter, down 18.5% from a year ago. If token prices keep sliding, retail activity thins, fees compress, and the marks on Coinbase's own crypto widen the reported loss further. The print that shows it first is next quarter's transaction revenue against this quarter's $1.22B.
The cash is cleaner on paper than in fact. The read leans on a 34% free-cash-flow margin in 2025, yet the company lost money three quarters running and diluted owners 29% in three years. If the next cash-flow statement shows stock pay still near 17% of revenue funding that "free" cash, the owner's real return is far thinner than the headline. What proves the read wrong is stock comp holding high while the share count keeps climbing.
The recurring base compounds anyway (right tail). Held to the same bar: if subscription-and-services revenue, the dollar-coin interest, staking and custody, keeps growing while transactions fall, Coinbase quietly becomes infrastructure that earns through the cycle, and the market is paying today for a volume bet, not that. Rate cuts would trim the dollar-coin interest, but circulation growth and any stablecoin legislation could more than offset. The print is subscription revenue holding or rising as transaction revenue falls.
Closing thoughts
At $184, about 3.7 times book and with no trailing profit to price, you are paying for two things at once: the next crypto up-cycle and the transition to recurring revenue. The distribution is fat on both ends. In a real rally this business prints violently, Q2 2025's $1.43B quarter is the proof, and the stock re-rates fast off a base 54% below its high. In a long cold spell the losses persist and the buyback-plus-dilution keeps bleeding per-share value. The payoff depends on crypto prices turning back up and how long that takes, not on any single quarter's numbers, so what matters is whether Coinbase survives the wait, and it does: net cash, $34 a share in the bank, capital-light, it outlasts the winter. The left tail here is dilution and drift, not ruin.
The bet is still that people keep buying and selling crypto through Coinbase and increasingly park their dollars in its dollar-coin and staking accounts, where it earns a slice in any weather. What breaks it is the cold case where volumes stay down and interest rates fall together, starving both engines at once. The two numbers that tell you first, before the share price does, are transaction revenue and subscription-and-services revenue in the next report: if the first keeps falling while the second climbs, the transition is real; if both fall, the winter is winning.
Methodology
Figures are drawn from Coinbase's SEC filings through the June 2026 quarter (filed 2026-07-30) and a market-data feed, with the filing taken over the vendor where the two differ. Revenue, net income and diluted EPS are as reported; Q4 2025 revenue is derived as fiscal 2025 revenue less the sum of Q1–Q3 2025; the 2026, 1H row sums the two filed 2026 quarters and is not a full year. GAAP net losses and positive free cash flow diverge because Coinbase marks crypto held on its own balance sheet through income and adds back non-cash stock compensation; both effects are noted rather than netted. Price, the 52-week range and valuation multiples are vendor market data as of 2026-09-06; trailing P/E is not meaningful on a trailing-twelve-month net loss. Documentation prepared with AI assistance. Not investment advice.
Fact check: 4 numerical errors corrected (revenue peak timing, sequential revenue pattern, FY2025 FCF margin corrected from 48% to 34%, QoQ decline from 14% to 13.7%). Bundle financials reconciled to FMP. Qualitative claims (board member status, acquisition target name) not web-verified due to tool access. Final analysis verified as of Sep 6, 2026.
Bid Cap
Daily ideas, a 390-name database, and a model long/short book from an investor who mostly covers financials. $70 a month or $700 a year.
Subscribe on Substack

