A colleague forwarded me a revenue table last week. Three columns, no commentary. Circle's total revenue stood at $2.75 billion; reserve income accounted for $2.64 billion of it. Distribution and transaction costs absorbed $1.66 billion. I have spent twenty-eight years reading documents of this kind, and what held me was not the magnitude of the figures but their direction โ value is drifting away from the institution that issues the token and toward the institutions that touch the customer.
Ninety-six percent of a stablecoin issuer's top line now arrives as interest on short-dated US Treasuries and overnight repo. Roughly sixty percent of it departs again through distribution agreements. What survives is not a franchise. It is a pipe, and pipes collect a toll rather than a margin. That distinction sits beneath Citi's integration with Coinbase, and it is precisely the part the bull-market coverage keeps skipping.
For those arriving late: Citi now routes institutional payment flows through Coinbase's blockchain rails. Coinbase receives digital dollars, converts them automatically into fiat, and Citi โ acting as the bank of record โ completes settlement through Spring by Citi, a platform already operating across twenty-three markets and processing billions annually. The enterprise treasurer never touches a stablecoin. That design choice is deliberate, and it tells you more about the architecture than any whitepaper would.
The regulatory gravity well here is the GENIUS Act, signed in July 2025 and scheduled to take effect on 18 January 2027. Section 4(a)(11) forbids an issuer from paying interest directly to token holders. Read that clause slowly, because it does not abolish yield โ it relocates it. The drafters did not extinguish the economic incentive; they simply changed who is permitted to hand it over. And that, in turn, is why a 3.75% reward can exist at all in a supposedly non-interest-bearing instrument.
Trace the flow. Reserve interest accrues at the issuer, passes through a distribution partner, and arrives at the holder relabelled as a reward. The economics are identical to a deposit rate; the legal character is not. This is not a semantic trick so much as a compliance arbitrage, and its durability depends entirely on whether regulators accept the rename. A corporate treasurer's yield is unchanged. Only the noun on the statement has moved.
What fascinates me, having audited contracts since the 2017 ICO season, is how familiar this pattern is. I once refused to sign off on a reentrancy flaw in a $2 million raise, and the founders called me a blocker. The lesson I carried away was that code and accounting share one property: both reveal intent more honestly than marketing does. The same holds here. Follow the reserve income and you find the actual product.
Tether, by contrast, retains three to three-and-a-half cents on every dollar, because it distributes through nobody of comparable scale. Its margins look superior; its posture looks agile. But that advantage is purchased with regulatory distance and the absence of an accountable distribution chain. In a regime that increasingly rewards licensed distribution, Tether's efficiency and its vulnerability are one and the same asset โ and the moment compliance becomes the price of liquidity, the arithmetic inverts.
I have spent the last decade designing quadratic voting systems for small governance communities, and I learned the hard way โ after a $50,000 treasury drain that I still turn over in my mind โ that incentives are always captured by whoever controls the interface. Citi appears to understand this better than most. It is running a barbell: today it earns processing volume through Coinbase's existing rails, while simultaneously participating in a twenty-one-bank consortium that intends to launch its own dollar stablecoin on a public chain by early 2027. That is hedging, not conviction โ and hedging is what prudence looks like when the terminal form is genuinely unknown.
Notice, too, what is under-discussed. BlackRock administers the reserve fund behind the largest compliant stablecoin. It sits quietly inside the issuer's stack, inside competing consortiums, and inside the plumbing of every narrative at once. Infrastructure rarely announces itself.
Here is where I part ways with the consensus. Almost every analyst frames this as a contest over issuance โ who mints the dollar. That framing is comfortable and largely irrelevant. The decisive layer is the last mile: the conversion of tokenised dollars back into settlement dollars. Citi is not trying to be a stablecoin issuer. It is positioning itself as the throat through which every stablecoin, regardless of brand, must eventually pass. Whoever owns that throat owns the economics, because every model needs a fiat exit.
The blind spot in the bullish case is timing. A twenty-one-institution coalition spanning Goldman, Bank of America, Wells Fargo, Deutsche and UBS must align on technology, compliance and governance simultaneously โ and the 2027 horizon leaves almost no slack. Long before I cared about blockchains, I learned that multilateral committees move at the speed of their least convinced member. Citi's barbell exists precisely because its own strategists suspect the consortium slips.
There is a second, quieter risk. If the notice-and-comment process ultimately defines a reward as interest in substance, the entire 3.75% structure faces reconstruction. A statute that has not yet taken effect is already redesigning the business models of firms it has not yet regulated. Looming over all of it is the rate path itself: falling policy rates compress reserve income instantly, while distribution splits negotiated in a richer era tend to stay where they were written.
So watch the pipe, not the coin. If value keeps migrating toward the institutions that control settlement and distribution, then the interesting question is no longer which stablecoin wins the market โ it is whether any issuer can retain pricing power once the last mile belongs to someone else. Ask that, and the next earnings season reads very differently.