The OCC Chartered a Second Money Supply. The Reserve Is the Product.

Yesterday's final rule turns the GENIUS Act into an operating licence: permitted issuers, one-for-one reserves, criminal penalties from 18 January 2027. The interesting part is not the crypto. It is who must hold what, and who is forbidden from paying for it.

On 21 October the Office of the Comptroller of the Currency published the final rule that turns last year's stablecoin statute into something an institution can actually apply for. A new part of the rulebook, 12 CFR 15, sets the standards for who may issue a payment stablecoin, what must sit behind it, how redemption works, and what happens when an issuer fails. It takes effect on 18 January 2027, the same day the statute does, and from that date issuing one without permission carries criminal penalties. That sounds like plumbing. It is actually a chartering decision: the state is naming a second set of entities that may issue a dollar-denominated liability, and telling them exactly which government paper must back it. Treasury's own press release reaches for the National Currency Act of 1863, which is not the innocent comparison it looks like. This piece is about what that structure does to the money supply, to bank funding, and to the bill market that ends up holding the collateral.

What was actually published

The rule is Bulletin 2026-50, accompanied by a waiver guidance bulletin and a press release from the Comptroller. It finalises the framework the OCC proposed on 25 February 2026, published in the Federal Register on 2 March, after what the agency describes as hundreds of comments.

Substance sits in a new 12 CFR 15: permitted activities, reserve assets, redemption, risk management, audits and reporting, custody, applications and registrations, examination of foreign issuers, and revocation of an approval. It also reaches into four existing parts of the rulebook that govern bank capital, prompt corrective action, assessments, and practice and procedure, because an issuer of stablecoins now needs a capital and operational backstop defined in the same language as a bank.

Two details matter more than the rest. First, the rule deliberately leaves out the Bank Secrecy Act, anti-money-laundering, and sanctions requirements; those arrive in a separate rulemaking with Treasury. Second, an issuer with a substantially complete application pending on or before 18 January 2027 may ask for a waiver from parts of the statute or the rule. The OCC will only start accepting applications once the Office of Management and Budget approves the information collection behind the forms.

So the compliance clock starts on a date certain, while the enforcement architecture for illicit finance does not yet have a text, and the application window has no opening date at all. Both facts are load-bearing for what follows.

The 1863 comparison is the tell

Treasury's framing invokes the National Currency Act, which created the national banking system and, with it, a national market for the bonds that financed the Union. That Act did two things at once: it granted charters, and it told the chartered banks what they had to hold. The reserve requirement was not a safety add-on. It was the demand curve.

The GENIUS structure repeats the shape. A permitted issuer may take dollar deposits in the form of tokens only if it holds one-for-one backing in assets the statute permits: cash, deposits at a regulated institution, short-dated Treasury bills, repo on those bills, and government money-market funds. Every token issued is a claim on that pool, redeemable at par.

That is a narrow bank with a payments licence. The holder gets a claim that is always worth a dollar and never earns anything. The system gets a new, price-insensitive, rule-bound buyer of the shortest government paper, sized by how many dollars people want to hold in token form rather than in a bank account. Nobody has to like the arrangement for it to work; they only have to use it for payments.

The reserve is the product

Follow the collateral and the business model becomes visible. On the 19 October 2026 par curve the 3-month bill yields 4.27%, the 2-year 4.77% and the 10-year 5.15%. A reserve book restricted to the short end therefore earns roughly the bill yield on the whole float.

At that rate, every $100bn of tokens throws off about $4.3bn a year of reserve income before costs. The issuer keeps it, because the holder is promised par and paid nothing. The margin is not a fee for technology; it is the spread between what short government paper pays and what the token pays, which is zero.

This is why the fight over the rule was never really about cryptography. It is about who captures the carry on the safest collateral in the world, and whether that capture happens inside the deposit franchise or beside it.

Why the yield ban is the architecture

The statute forbids paying interest on a payment stablecoin. That single line does more work than any reserve rule. If a token paid a market rate, it would be a money-market fund with a payment app attached, and the competitive question would move straight onto bank deposits, which is a funding question for the entire credit system.

The prohibition keeps the token a medium of exchange rather than a store of value with a coupon, which is precisely the boundary that protects deposits. It also creates the obvious loophole: rewards routed through an affiliate or a third party, or a stablecoin bundled with free services that are worth something. A prohibition that can be met with a loyalty programme is not much of a prohibition.

Expect the first serious supervisory dispute to be about exactly that, not about reserves. The reserve rules are mechanical and auditable. The yield rules are economic and porous, and they are the ones that decide whether this becomes a payments layer or a deposit substitute wearing a different name.

The hole in the rule: AML arrives later

Deferring the Bank Secrecy Act, anti-money-laundering and sanctions provisions is defensible as sequencing, since the Treasury has to co-write them. It is also the part of the framework with the least room for a second draft. A payment rail that settles instantly, across borders, at any hour, is the highest-value laundering surface ever built, and the rule creates the licence before it creates the surveillance obligation.

Meanwhile the penalties for operating without permission begin on 18 January 2027. The result is a period in which the state is strict about who may issue and silent about how issuance must be watched. That is not a stable equilibrium. It is the window in which the compliance culture of the industry gets set.

The fiscal arithmetic that cancels itself

The popular case for stablecoins inside the government is that they manufacture demand for Treasury bills at the short end, where the sovereign borrows most. That case is weaker than it sounds, and the reason is seigniorage.

Dollars that move from bank deposits into tokens do add to bill demand, but dollars that move from physical currency into tokens remove a zero-interest liability from the Fed's balance sheet and replace it with an interest-bearing one. The net fiscal improvement can be close to nothing. The composition of the sovereign's funding changes; the consolidated position may barely move.

For a bank, though, the composition change is the whole story. Deposits that leave for a token leave the balance sheet, and the reserve income goes to the issuer instead of the lender. Community banks will feel this first, which is why their trade associations spent the comment period arguing about yield rather than about reserves.

What would falsify this

The claim is not that stablecoins take over the payment system, and not that the dollar is debased. It is narrower: that this rule creates a chartered, reserve-backed parallel money supply whose economics are the carry on short bills, whose boundary against deposits is the yield prohibition, and whose weakest joint is the deferred AML rulemaking.

Falsifier one: the OCC permits rewards that are economically equivalent to interest, through affiliates or bundled services. Then the boundary fails, and the deposit franchise is exposed on price rather than protected by statute.

Falsifier two: applications do not open before 18 January 2027 because the OMB approval slips. Then the compliance date arrives with no licence available, the waiver route becomes the whole market, and the framework starts life in grandfathering rather than in supervision.

Watch the first one. It is cheaper to observe than the second, and it tells you whether the state meant to build a payment rail or to re-charter banking.