Seven cents. That is what Circle keeps from every dollar of revenue it books on USDC. Not seventy. Seven. The company published the split on October 16 โ a blog post, not a filing โ carving its H1 revenue into three buckets: 59% to distribution partners and transaction costs, 34% reinvested into licenses, banking, liquidity, compliance and new products, and 7% net profit. On roughly $1.4 billion of revenue, that is about $98 million of net income. The "money printer" that retail traders imagine when they say "stablecoin issuer" produces a net margin thinner than a grocery chain.
That single number is the anomaly worth chasing. If Circle is telling the truth, the entire mental model of stablecoin issuance as a high-margin rent machine is wrong. And if it is not telling the whole truth, the gap is hiding inside the 59% it declined to split. When code speaks, we listen for the discrepancies. Here the "code" is an income statement, and the discrepancy is structural.
USDC is a fiat-backed centralized stablecoin โ one dollar in reserves, one token out, no hard cap, no governance token. Its "token economics" are not really token economics at all; they are the equity economics of Circle Internet Group (NYSE: CRCL), a listed company with SEC disclosure obligations. That distinction matters. Tether publishes attestations from an offshore structure. Circle files with a regulator. This 59/34/7 split is the first time a major issuer has voluntarily opened its P&L on distribution economics โ and it did so on a blog, which is a deliberate choice of format.
The scale is not trivial. USDC cleared more than $36 trillion in on-chain volume in H1 and minted or redeemed over $315 billion, with redemptions settling in as long as two business days. In June, USDC's share of stablecoin transaction volume reached roughly 77%, up from 60% a year earlier. By volume, USDC already dominates. By market capitalization it still trails USDT. That divergence โ volume leadership without supply leadership โ is the first clue that this is a settlement instrument, not a savings instrument.
I have watched this pattern before. In 2024, when I built the ETF-flow model against long-term holder supply, the market kept assuming inflows drove price. They did not. They drove a reduction in exchange float. The lesson then is the lesson now: the headline metric people trade on is usually not the metric that explains the mechanism.
Read the 59/34/7 split as a cash-flow diagram, not a press release.
Revenue is 100% reserve yield โ interest on Treasuries. It is not protocol-native income; it is macro rent. Circle earns it because short-term rates are high, not because USDC does anything clever. That is the first structural fact the disclosure quietly admits: the issuer's top line is a bet on the Federal Reserve, not on product adoption.
The 59% is the second fact, and the more damaging one. Distribution partners and transaction costs together take $819 million. Circle did not split the two. That single omission is where the narrative gets soft. If most of the 59% is partner revenue share โ Coinbase, Binance โ then Circle is paying enormous rent for distribution. If a large slice is actually transaction cost โ gas subsidies, mint and redeem infrastructure โ then the "we are so generous to partners" story is inflated and the real giveaway is smaller than advertised. We cannot tell which, and that is the point. A disclosure that refuses to split its largest line is a disclosure engineered for a headline, not for an audit.
Then there is the direction of the split. Circle states that partner revenue share rises as partner-held balances rise. That is a scale diseconomy dressed as growth. Every dollar of new USDC that arrives through a large distributor may carry a higher marginal cost than the last. Growth and margin point in opposite directions until fixed compliance costs are absorbed.
The competitive frame sharpens this. Open USD โ backed by Coinbase, Visa and Stripe โ reportedly routes 100% of reserve yield to partners. Circle's rebuttal is pointed: a stablecoin that distributes everything and funds nothing. Who pays for compliance, banking relationships and settlement liquidity in a zero-retention model? Fair question. But it is also a confession that Circle cannot match the price. When your competitor's offer is "all the yield," your defense has to be "we do the boring, expensive things." That is a real moat โ licenses, bank accounts, a listed balance sheet โ but it is a cost moat, not a margin moat.
And the moat is being tested from inside the castle. Coinbase holds roughly $20 billion of USDC and is Circle's largest distribution partner. Coinbase also backs Open USD. The largest channel is hedging both sides of the table. Binance, meanwhile, is buying $100 million of Circle equity under a five-year agreement. Read those two facts together: the exchanges have figured out that the issuer is the least powerful party in the value chain, and they are pricing that power accordingly. Circle's answer is vertical integration โ Arc, its own settlement chain; Circle Mint; Circle Wallets with free instant redemption. Own the rail, stop renting the road.
The consensus reading of this disclosure is "Circle is honest, therefore Circle is safe." I would invert it. The honesty is selective, and the most flattering number โ the 77% volume share โ is the one I trust least as a durability signal.
Transaction volume is not stock. USDC's 77% share tells you it is used for high-frequency settlement; it does not tell you the volume is sticky. A meaningful part of that share growth likely came from Binance migrating flows from USDT to USDC โ a single large client re-routing. Concentration masquerading as market share is the oldest trap in on-chain analytics. I learned that lesson on BAYC in 2021, when 40% of a "community" of 10,000 wallets turned out to be fifteen bots. Volume without holder diversity is the same illusion in a different ledger.
There is also a data-integrity flag I cannot ignore. The source material dates the figures to "H1 2026" while citing an October 16 post. H1 numbers are typically disclosed around August, not mid-October. Either the period label is wrong, or the blog is a retrospective. I flag this because a forensic read treats timestamp conflicts as signal, not noise. When code speaks, we listen for the discrepancies โ and a disclosure whose own dates do not reconcile deserves a discount until the original is checked.
Finally, the podcast title buried at the end of the release is the real tell: "Why Agents Will Ditch Stablecoins That Don't Pay Yield." That is Circle pre-writing its own next threat. The moment regulators permit yield-bearing stablecoins, a model that pays partners and pays holders nothing becomes structurally obsolete. A claim is a hypothesis until the ledger agrees โ and this ledger has not yet shown us the part that matters most.
Circle has not revealed a money printer. It has revealed a rate-sensitive, distribution-dependent, thin-margin settlement utility that happens to be listed. Price CRCL like a payment network exposed to the policy rate, not like a crypto protocol riding a narrative.
The signal to watch next is not the 7%. It is the 59% under a rate cut. If the Fed eases and partner share stays rigid, net income compresses toward zero โ and Circle will have to choose between cutting partners and breaking the "generous issuer" story that is its entire defense against Open USD. Watch Coinbase's USDC-versus-Open-USD custody mix. Watch whether Arc actually internalizes distribution. And watch the yield-bearing stablecoin legislation. The seven-cent confession is not the end of the story. It is the first footnote in a margin war that has barely started.