> ## Content Index
> Fetch the complete content index at: https://thepaymentsoperator.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# A Stablecoin Settles in Seconds. The OCC Just Priced the Exit at Two Days.
- URL: https://thepaymentsoperator.com/a-stablecoin-settles-in-seconds-the-occ-just-priced-the-exit-at-two-days/
- Published: 2026-03-20T03:30:00.000Z
- Updated: 2026-11-18T11:17:53.000Z
- Description: The OCC set stablecoin redemption at two business days, with a seven-day gate above 10% in twenty-four hours. That is not a deposit. It is a fund with a payment interface — and it changes what the instrument can sit underneath.
- Author: Armando

*The first federal regulator to write the GENIUS Act's rules did not write a payments rule. It wrote a liquidity policy, with a run gate buried in the middle of it.*

On 25 February 2026 the Office of the Comptroller of the Currency published a notice of proposed rulemaking to implement the GENIUS Act. It is the first federal regulator to propose substantially all of the rules the statute requires of it. The proposal creates a new Part 15 of the OCC's regulations and amends four existing ones — capital adequacy, prompt corrective action, fee assessments, and rules of practice and procedure. It asks 211 specific questions. Comments close on 1 May 2026.

Most of the commentary read it as a licensing document: here is how you become a permitted payment stablecoin issuer in the United States. That reading is correct and incomplete. The licensing provisions are the least surprising part of the text. What the OCC actually wrote is a liquidity policy — and the most consequential sentence in it is about how long you wait for your dollars.

## Two business days, unless everybody asks at once

The proposal sets "timely redemption" at two business days after a request. Then it adds the provision that will matter more than any other: if redemption requests exceed 10% of outstanding issuance value within a 24-hour window, the redemption period extends to seven calendar days, and no redemptions occur during that period without the OCC's approval.

Read those two sentences next to each other and the instrument comes into focus.

An on-chain transfer of a stablecoin settles in seconds and is final. The same unit, converted back into bank dollars, takes two business days — and takes seven calendar days precisely in the moment when the largest number of holders want out. The asymmetry is not a defect in the proposal. It is the proposal's honest description of what the thing has always been. What is new is that the asymmetry now has a number attached to it, and a regulator's signature under the number.

A deposit does not work this way. A deposit is a claim on a bank, payable on demand, standing behind a public insurance scheme and a central bank willing to lend against the bank's assets. There is no gate because the gate is unnecessary: the backstop is the gate.

A money market fund does work this way. Liquidity fees and redemption gates are the standard architecture for a pooled claim on short-dated assets held by a private manager with no public backstop. When the pool is stressed, you slow the exit rather than fire-sell the assets.

The OCC has written a money market fund with a payment interface. That is a defensible thing to write — arguably the only defensible thing, given that the GENIUS Act deliberately withheld both insurance and yield. But it settles a question a great deal of payments strategy has been pretending is open: whether a stablecoin can sit underneath an obligation that must be met in fiat on a known date. Under this proposal it cannot, because the date is not known. It is two days, or it is seven, and which one applies depends on the behaviour of strangers.

## The capital number that is not a number

There is no standardized minimum capital requirement in the proposal. Capital must be "commensurate with the level and nature of all risks" the issuer faces. Eligible capital is common equity tier 1 and additional tier 1 only — tier 2 instruments, subordinated debt among them, are excluded. A de novo issuer must hold at least $5 million during its first three years.

This is the most important thing in the document and it will receive the least attention, because it is not a figure, and figures are what get quoted.

A capital requirement with a ratio is a compliance test. You model it, you report it, you pass or fail, and the answer is the same for everyone. A capital requirement that is "commensurate" is a supervisory conversation. You cannot model it in advance; you arrive at it through a relationship with an examiner who has formed a view about your business.

Supervisory relationships are not distributed evenly. An institution that has held a national bank charter for a century knows which questions will be asked, has answered them before, and employs people whose job is knowing the answers. A new issuer does not. The practical effect of a rule without a ratio is that the cost of capital becomes a function of institutional familiarity, and institutional familiarity is the one input a new entrant cannot buy.

I do not read this as a flaw in the drafting. Standardized ratios for an industry with four years of history and no failure data would be guesswork dressed as rigour. But the consequence is predictable, and anyone building a business plan around a US stablecoin licence should write it into the plan: the number you will be held to is not published, and the people who will find it easiest to discover are the ones who were already in the room.

The operational backstop sits beside the capital provision and deserves more credit than it will get. An issuer must hold assets equal to twelve months of total expenses, in specified high-quality liquid assets, kept separate from reserves.

Look at what that assumes. It assumes an issuer's most likely failure mode is not a reserve shortfall. It is running out of money to operate. You can be perfectly backed at 1:1 and still fail because you cannot pay the firm that examines your reserve reports, the custodian that holds your Treasuries, the engineers who keep the mint and burn functions alive. A reserve pool is not a payroll.

That is an operations insight written into a capital rule, and it is more honest than most capital rules. Every payments regulator in this hemisphere should read the provision and then ask why it does not apply to acquirers, who also hold other people's money in transit and also die of operations rather than of balance sheets.

## Five hundred million dollars, in increments of two hundred and fifty thousand

An issuer with at least $25 billion outstanding must hold 0.5% of reserves in fully insured deposits, capped at $500 million. Federal deposit insurance currently covers $250,000 per depositor, per insured institution.

Do the division. Five hundred million dollars of fully insured deposits, at $250,000 of coverage per institution, requires on the order of two thousand insured depository relationships.

Nobody has two thousand bank relationships. The requirement will be met the way large insured balances are always met: through a deposit placement network that spreads one balance across hundreds of institutions and reports back a single number. Those networks exist, they work, and they are almost certainly what the OCC has in mind.

That is the point. A provision written to reduce the concentration of an issuer's reserves is discharged, in practice, by introducing a dependency on third-party infrastructure that the provision does not name, does not supervise, and does not require to be redundant. The largest issuers in the market would all be routing the same requirement through the same small set of intermediaries. The rule diversifies the deposits and concentrates the plumbing.

I have made this mistake myself at smaller scale, in a different domain, and the shape is always the same: a control is specified in the unit the regulator can measure and discharged in the unit the operator can actually execute, and nobody owns the gap between them. The question for the comment file is not whether placement networks are acceptable. It is who examines them.

## Yield, and the cousins who pay it

The GENIUS Act's prohibition on paying interest or yield to stablecoin holders is restated in the proposal. Then the OCC adds something the statute did not have: a rebuttable presumption that arrangements between an issuer, its affiliates, or "related third parties" which pass yield through to holders violate the prohibition.

When the GENIUS Act passed in July 2025, I argued that a dollar which pays nothing was the design rather than the flaw — that the yield ban existed to stop a tokenised dollar from draining the deposit base that funds bank lending. The structural gap in that design was never the issuer. It was the distributor. The issuer pays nothing, and the exchange or the wallet pays a "reward" out of its own revenue, and the holder receives yield on a dollar that legally bears none.

"Related third parties," coupled with a rebuttable presumption, is the OCC closing that gap. The mechanism is what matters. A presumption means the burden sits on you. You are not accused of passing yield; you are assumed to be passing yield, and you have to show otherwise.

For anyone whose distribution economics include a share of reserve income, this is a repricing event. The question is no longer whether the arrangement is legal. It is whether you can demonstrate to a supervisor, with documents, that you are not a related third party — and proving a negative to an examiner is a cost line with no ceiling.

## Diversification, in two flavours

The proposal offers two versions of the same reserve diversification requirements. Option A presents them as a safe harbour. Option B makes them mandatory. The figures are identical: daily liquidity of at least 10%, with no more than 50% of that daily liquidity at a single institution; weekly liquidity of at least 30%; no more than 40% of reserves at any one institution overall; and a weighted average maturity of no more than twenty days.

Those are money market fund numbers. Anyone who has read Rule 2a-7 will recognise the furniture.

The choice between the two options is the real decision in this section, and it is not a technicality. A safe harbour means the figures are one way to comply and you may do something else if you can defend it — which makes reserve management a treasury function, exercised by people who form views about duration. Mandatory means the figures are the only way to comply, which makes reserve management a compliance function, exercised by people who check boxes. Same numbers, different department, different hiring plan.

The OCC also asks, separately, whether Treasury securities with up to two years of remaining maturity should qualify as reserves, against the statute's 93-day limit. Of the 211 questions, that is the one that moves the economics most. Reserve income is substantially a function of duration, and the gap between 93 days and two years is the difference between a business that earns a thin spread and one that earns a real one. Everyone who intends to issue should answer that question in the comment file. Most will not.

## The ten-billion-dollar trapdoor

A state-regulated issuer that crosses $10 billion in outstanding issuance must notify the OCC in writing within five days, request a waiver within 240 days if it wants one, and complete its transition to federal regulation within 360 days. If the state regime has been certified as substantially similar, a presumption of approval applies to the waiver.

This is a growth ceiling wearing the costume of a threshold. An issuer approaching $10 billion has to decide whether to slow its own growth or begin a federal process on a clock it did not choose. Anyone who has run a licensing migration knows what 360 days means in practice: it means you start eighteen months early, because the work that takes 360 days is the work after the decision, and the decision itself takes a board.

The federal application timeline, by contrast, is unusually generous. The OCC has 30 days to say whether an application is substantially complete, and a complete application is deemed approved 120 days after receipt unless the OCC denies it. Deemed approval is rare in bank licensing and worth noticing. It converts an open-ended wait into a deadline the regulator itself must meet.

For issuers outside the United States, approval requires a comparability determination from the Treasury Department about the home regime, after which approval occurs on the thirtieth day following receipt unless the OCC rejects the application. That is the provision this region should read first.

## From a Panamá desk

Panamá uses the dollar and does not issue it. A United States stablecoin regime therefore reaches us without any Panamanian instrument being involved at any point. What circulates here, if a tokenised dollar circulates here, is a claim on a US issuer governed by US rules, held by people under Panamanian law, in a country whose regulator has no seat at the table where the claim's terms are set.

The foreign-issuer route requires Treasury to find the home regime comparable. For a jurisdiction with no stablecoin regime at all, that is not a high bar — it is a bar that does not exist yet. There is nothing to compare. So the practical path for a regional institution is not comparability. It is a US licence, with the undisclosed capital number and the supervisory relationship that implies, or distribution of somebody else's coin.

And distribution is precisely what the yield presumption just made more expensive.

The two-day redemption clock lands here in a particular way. In a dollarized economy the case for stablecoins has never been that they are cheaper than dollars. We already have dollars; they are in the ATMs. The case has always been speed — that a tokenised dollar moves faster than the correspondent chain, which in this region is slow, expensive, and thinning as de-risking removes relationships.

That case survives this proposal intact in one direction and dies in the other. Money arrives instantly. Money leaves on a two-day clock with a seven-day gate behind it. For a treasury function that is not a cash instrument, and it never was one; the proposal has simply made the point unavoidable. Any balance sheet in this region carrying stablecoin balances under cash and equivalents should move them to the line where short-dated funds sit, and should do it before an auditor does it for them.

## My judgment

The OCC did not regulate stablecoins as payments. It regulated them as a fund with a payment interface, and I think that is the correct characterisation of what the GENIUS Act created. A statute that withholds both deposit insurance and yield has not built a deposit. It has built a share in a pool, and pools get gates.

But the characterisation has a consequence the payments industry has not absorbed. An instrument with an uncertain exit date cannot sit underneath an obligation with a certain settlement date. That rules out a long list of uses people have been pitching for four years: stablecoins as the settlement leg for card schemes, as the funding asset for same-day payouts, as the float behind an instant-transfer promise. Each of those requires knowing, today, that the fiat will be there tomorrow. Two days or seven, at the regulator's discretion, is not an answer a settlement system can build on.

What stablecoins are good at remains what they were always good at: moving value across borders faster than correspondent banking, for counterparties who can tolerate uncertainty at the exit. That is a real and large business. It is a smaller business than the one the industry has been describing.

The second judgment is about the capital provision, and I will restate it because it is the part that gets lost. A rule with no ratio favours whoever already has the relationship. That is not an accusation; it is arithmetic about how supervisory discretion distributes. Expect the first wave of licensed issuers to be subsidiaries of institutions that have held national charters for decades, and expect the commentary to attribute that to their superior risk management rather than to their familiarity with the examiner.

The third is the one I would act on immediately if I sat on a regulator's side of the table in this region. The twelve-month operational backstop is the best single idea in two hundred pages. It encodes something those of us who run payments operations know and rarely get to say in a capital document: the money is almost never the thing that fails. The thing that fails is the operation around the money. Any jurisdiction writing rules for acquirers, wallets or remittance firms should lift that provision, change the noun, and ship it.

## Three questions for the people who own this

**1.** If your treasury policy lists stablecoin balances under cash and equivalents, what happens to that classification when the exit is two business days with a seven-day gate above 10% in twenty-four hours? Who signs the revised policy, and what does the reclassification do to your liquidity coverage?

**2.** If any part of your distribution economics comes from reserve income shared with you by an issuer, can you demonstrate to a supervisor — with documents, today — that you are not a related third party? Who holds that evidence, and has anyone read it since the deal was signed?

**3.** If your institution would need a Treasury comparability determination to issue, and your jurisdiction has no stablecoin regime to compare, what is the actual plan? A US licence with an undisclosed capital number, a partnership on repriced terms, or nothing — and if it is nothing, who has said so out loud?

Reply to this email with a line or two. The sharpest answers will shape the next issue.

## Sources

**Office of the Comptroller of the Currency, 25 February 2026** — Bulletin 2026-3, "GENIUS Act Regulations: Notice of Proposed Rulemaking." Source for the date; the creation of a new 12 CFR Part 15 covering activities, reserve assets, redemption, risk management, audits and supervision, custody, applications and registrations, examination of foreign issuers, revocation of approval, and a capital and operational backstop; the amendments to 12 CFR 3, 6, 8 and 19; the scope of covered entities; the fact that Bank Secrecy Act, anti-money laundering and OFAC requirements are reserved for a separate joint rulemaking with the Treasury Department; and the confirmation that the GENIUS Act was enacted on 18 July 2025, with an effective date the earlier of eighteen months after enactment or 120 days after final implementing regulations.

**Sullivan & Cromwell LLP, 11 March 2026** — "GENIUS Act Implementation: OCC Issues Proposed Rules." Source for the detail the bulletin page does not carry: that the OCC is the first federal regulator to propose substantially all of its required rules; the 211 questions; the 1 May 2026 comment deadline; the rebuttable presumption covering affiliates and related third parties on pass-through yield; capital as CET1 and additional tier 1 only, with no standardized minimum and a $5 million floor for a de novo issuer's first three years; the twelve-month operational backstop; the two-business-day redemption standard and the extension to seven calendar days above 10% of outstanding value in twenty-four hours; the Option A and Option B diversification figures; the 0.5% and $500 million insured-deposit requirement above $25 billion outstanding; the $10 billion state-to-federal transition with its five-day, 240-day and 360-day clocks; the 30-day completeness and 120-day deemed-approval federal application clocks; the foreign-issuer comparability path and its thirty-day approval; and the bank-parent deconsolidation and CET1 deduction treatment.

**A note on what is mine, and what is not.** The facts above belong to the proposal. The readings do not. That a 10%-in-twenty-four-hours gate makes the instrument a fund rather than a deposit; that $500 million in fully insured deposits at $250,000 of coverage per institution implies a dependence on deposit placement infrastructure the rule never names; that a capital standard without a ratio advantages incumbents by distributing supervisory familiarity rather than risk; that the operational backstop belongs in acquiring rules — those are mine, and are arguments rather than findings. I have read the OCC's bulletin and a law firm's analysis of the full text. I have not read all two hundred-odd pages in the Federal Register, and where the proposal's own language differs from the summaries, the proposal governs. Nothing here reflects the position, portfolio, clients or vendors of my employer.