Banks Are Building Their Own Stablecoin Again

Citi, Goldman Sachs, and a group of major banks are building a stablecoin consortium. Their goal: faster settlement, programmable transactions, cross-border payments. All things that USDC and USDT already do, today, at scale, with $250 billion in reserves earning 4.5% yields.

The banks are not building a stablecoin because stablecoins don't exist. They're building one because the existing stablecoins aren't theirs.

The pattern

This is the third time banks have responded to an outside payment innovation by building a controlled version of it.

Venmo and Cash App made peer-to-peer payments mainstream. The banks responded with Zelle, a system that only works between participating banks and has no buyer protection. Zelle processed $806 billion in 2023. It works. It also exists because banks wanted P2P payment data flowing through their pipes instead of PayPal's.

FedNow and RTP gave the U.S. real-time payments. But the banks had already built RTP through The Clearing House, a bank-owned utility, to make sure they controlled the real-time rail before the Fed could offer one directly. When FedNow launched in 2023, it was entering a market the banks had already structured around their own infrastructure.

Now stablecoins. Circle and Tether built the market. $250 billion in circulation. B2B cross-border settlement at 30 to 80 basis points compared to 200 to 400 through correspondent banking. The technology works. The unit economics work. The banks watched it happen. And now they want their version.

What the consortium actually solves

The consortium solves a specific problem, but the problem is the banks', not the market's.

Regulated banks can't hold USDC on their balance sheet without regulatory complexity. Capital treatment is unclear. The compliance overhead is real. A bank-issued stablecoin, backed by deposits at member institutions, settles within the consortium's own regulatory framework. The banks understand that framework because they wrote it.

This is a legitimate constraint. Banks operate under Basel III capital requirements, FDIC oversight, and AML obligations that make holding third-party stablecoin reserves genuinely complicated. Building their own version lets them tokenize deposits rather than hold someone else's liability.

The BIS has been saying this for two years: tokenized bank deposits may play a larger role than third-party stablecoins in regulated finance. The consortium is the banks taking that guidance and turning it into infrastructure.

What it doesn't solve

Interoperability. The reason USDC works for cross-border settlement is that it runs on public rails. Any counterparty, any country, any time. A bank consortium stablecoin works between consortium members. If your counterparty's bank isn't in the club, you're back to SWIFT.

This is the Zelle pattern again. Zelle works great if both parties bank at a Zelle member institution. If one doesn't, the transaction fails. The same structural limitation will apply to a bank consortium stablecoin. The settlement is fast, but only inside the walls.

Revolut launched its own euro stablecoin (EURR) in late August. Circle has USDC and EURC on multiple chains. The stablecoin market is fracturing into proprietary issuances rather than converging on shared infrastructure. Each new entrant makes the interoperability problem worse, not better.

Why this still matters

The consortium doesn't need to win the stablecoin market to succeed. It needs to give member banks a regulatory-compliant way to settle tokenized transactions between themselves. If Citi and Goldman can settle a securities trade in minutes using a shared stablecoin instead of waiting for T+1 through DTCC, the consortium pays for itself on settlement efficiency alone.

The banks have never competed on openness. They compete on control. A stablecoin they control, operating under their regulatory framework, settling between their own books, is exactly the product they'd build. The fact that the open market already solved the same problem for everyone else is beside the point.