A stablecoin issuer should not also be the vault for its rivals' reserves

Two firms, one charter, opposite promises

Two companies filed the same application with the same federal regulator on the same Monday. Modern Treasury, a San Francisco payments infrastructure firm, told the world its proposed bank would not issue a stablecoin of its own. Rain, a New York stablecoin infrastructure platform, said its proposed bank would issue stablecoins — and would also hold and administer the reserves backing coins issued by other companies.

Same charter. Same label. Entirely different animal underneath.

A national trust charter from the Office of the Comptroller of the Currency, the federal regulator that licenses national banks, is the most conservative banking licence on offer. The holder may custody assets. It may not take deposits and may not make loans. No maturity transformation, no credit book, none of the classic ways a bank blows up. That conservatism is exactly why these charters are attractive, and why a trade group sued the OCC over them days before these two filings landed.

But the restriction that makes a trust charter safe — you only hold other people’s assets — is doing less work than it looks when the assets you hold are the cash backing someone else’s money, and you also print money of your own.

The problem is not custody. It is the seating plan

Under the GENIUS Act, a permitted stablecoin issuer must hold reserves against the coins it has issued. Those reserves have to sit somewhere. Rain proposes to be that somewhere — for itself, and for other issuers building on its infrastructure. Its own statement makes the commercial logic plain: firms using Rain want the assets behind their programs held in a federally supervised entity.

That is a reasonable thing for customers to want. It is also how you build a single point of failure.

A stablecoin run is not a credit event. It is an operations event. Holders redeem, the issuer instructs the reserve holder to liquidate and pay, and the only question that matters in the first six hours is how fast the instructions clear. If one trust bank is administering reserves for a half-dozen issuers while simultaneously defending its own coin’s peg, it is sitting at the controls of a queue in which it is also standing. Nobody needs to behave badly for that to go wrong. Operational capacity is finite, and the entity with its own liability on the line is the worst possible allocator of scarce settlement attention.

The custody-bank analogy — the big securities custodians have served competing asset managers for decades — breaks precisely here. A custodian holding equities for two rival fund houses does not issue a rival fund. Rain would be the vault and a competitor in the same building, and the thing being competed over is redemption at par.

What a good approval order would say

None of this is an argument against the charter. It is an argument that the approval should be conditional and the conditions should be public.

Three would do most of the work. First, legal and operational separation between issuing Rain’s own stablecoin and administering third-party reserves — different entities, different staff, different systems, not a Chinese wall described in a policy document. Second, disclosure of concentration: how many issuers, and what share of the reserve-administration book the largest represents. Regulators learned that lesson about funding concentration in 2023 and it applies with more force to an entity whose entire business is other people’s backing assets. Third, a redemption-operations plan that has actually been exercised before the doors open, with a published statement that it was.

The proposed chief executive of Rain’s bank has run finance at a bank-as-a-service lender, at Square’s industrial bank and at Green Dot. That is a serious résumé for exactly these questions, which is a reason to believe the firm could meet such conditions, not a reason to skip them.

The case for leaving it alone

The strongest counter-argument is that conditions of this kind are supervision dressed as chartering, and the OCC has no business designing a company’s org chart. Trust banks already owe fiduciary duties. Reserve assets held in trust are not the bank’s assets; they are segregated by law and remote from its own failure. Layering bespoke structural conditions onto a charter application, the argument runs, is how a licence becomes a negotiation, and how a regulator ends up picking winners.

There is also a blunter version: these reserves are sitting somewhere today, and that somewhere is less supervised than a national trust bank. Pushing them into the federal perimeter is a strict improvement even with the conflict attached.

Both are right, and neither reaches the conclusion. Segregation protects reserve assets from the bank’s insolvency. It does not govern whose redemption instruction gets processed first on a Tuesday when the peg is wobbling and the operations team is twelve people. And “better than the alternative” is an argument for approving the application, which the OCC should probably do. It is not an argument for approving it unconditionally.

The direction of travel

This is not a one-off. Anchorage Digital, the oldest holder of one of these charters, has just bought a payouts company to link issuance, custody and settlement in one stack. Vertical integration is the business model here, and the charter is the keystone. If the OCC writes the first conditions badly, or writes none, it sets the template for everything behind it.