The venture does not have a name yet. It has 21 financial institutions, a plan to incorporate by the end of 2026, and a target of going to market in the first half of 2027.

What it does not have is agreement, and the disagreement runs deeper than the usual industry noise. In the space of a single week, the banking sector produced three incompatible answers to one question: how should regulated money move on a blockchain?

The sharpest signal is not what any of the three said. It is who signed which one.

Three architectures, built simultaneously, by overlapping groups of the same institutions. Fragmentation is the failure mode the referee just warned about, and the industry is constructing it on purpose.

The banks that signed

Bank of America, Citigroup, Capital One, Goldman Sachs, PNC Financial Services Group, and Wells Fargo anchor the US side. Alongside them sit TD, Scotiabank, UBS, Santander, BBVA, Deutsche Bank, Commerzbank, Lloyds, Crédit Agricole, Rabobank, and MUFG.

That is 17 named institutions out of 21, according to Payments Dive. The entity itself is to be announced in due course, which is a phrase that usually means the lawyers are still arguing about governance.

The coin will launch dollar denominated and expand into G7 currencies with an emphasis on euros. The stated scope covers wholesale, institutional, and retail markets, with cross-border payments and digital asset settlement as the headline use cases.

The pitch is bank-grade compliance, strong governance, distribution, and institutional risk management. Read that as: everything Tether and Circle do not have, offered by the institutions that already hold the customer relationships.

It is a credible pitch. It is also a defensive one, and the banks have been unusually open about why.

The clock they are running against

The GENIUS Act takes effect on January 18, 2027.

A venture that incorporates at the end of 2026 and reaches market in the first half of 2027 is not choosing that timing casually. It is choosing to be operational within weeks of the moment a federal framework for payment stablecoins becomes enforceable in the United States.

We looked at how that framework collides with existing anti-money-laundering obligations in our piece on the FDIC, the Bank Secrecy Act, and GENIUS. The short version is that the compliance burden lands on the issuer, and issuers with bank charters already carry most of it as a fixed cost.

For 21 banks, that is not a burden. It is a moat. Every crypto-native issuer has to build a compliance function from scratch to meet the same bar these institutions cleared decades ago.

Which explains the urgency but not the anxiety.

Six trillion dollars of anxiety

Brian Moynihan, chief executive of Bank of America, has warned that 30 to 35 percent of US commercial bank deposits could migrate to stablecoins. That is roughly $6 trillion.

He may be wrong about the size. He is not wrong about the mechanism.

A deposit is a bank's cheapest funding. Fractional reserve lending is built on the assumption that most of that money stays put. A dollar that leaves a checking account for a stablecoin wallet stops funding loans and starts funding somebody's Treasury bill portfolio. The bank loses the spread, and the depositor gains nothing except faster settlement.

So the banks have two options. Watch the deposit leave, or issue the thing it leaves for.

They picked the second, which is the correct commercial answer and does nothing to resolve the underlying question of whether a bank-issued stablecoin is a stablecoin at all or simply a deposit wearing different clothes.

The second architecture

Project Agorá thinks it is a deposit wearing different clothes, and Project Agorá has a working prototype.

Run jointly by the Bank for International Settlements and the Institute of International Finance, Agorá brings together seven central banks and more than 40 regulated financial institutions. It has demonstrated tokenized commercial bank deposits settling atomically against tokenized central bank reserves on a shared platform, across currencies and across jurisdictions. As of May 2026 it moved from experimentation toward real-value testing.

The design preserves the two-tier banking system. Commercial bank money and central bank money keep their existing relationship. Nothing leaves the regulated perimeter, and the settlement finality comes from the central bank, where it has always come from.

JPMorgan Chase is a participant in Agorá.

JPMorgan Chase is not among the 21.

That is the most informative fact in this entire week of announcements. The largest bank in the United States, the one furthest along in tokenized deposits with its own production infrastructure, looked at a 21-member stablecoin consortium and passed. It is already running the alternative.

Joel Hugentobler, cryptocurrency analyst at Javelin Strategy & Research, framed the distinction carefully: "Stablecoins and tokenized deposits may look similar on the surface, but they represent very different models of money."

The third architecture

While one group of banks builds a coin and another builds tokenized deposits, Bottomline did something less ambitious and possibly more useful.

On September 3 it partnered with Chainlink to give more than 600 bank customers access to blockchain-based cross-border settlement. Bottomline serves over 600 banks, 1,200 financial institutions, and 10,000 businesses, and processes more than $16 trillion in payments a year.

The clever part is what it does not require. Banks keep sending payment instructions in ISO 20022, the messaging standard they already use. Chainlink's Cross-Chain Interoperability Protocol moves tokenized value underneath, coordinated by the Chainlink Runtime Environment. No core system rebuild. No new coin. No consortium governance to negotiate.

Cross-border payments today still take days and cost 5 percent or more per transfer. A route that attacks that number without asking a bank to replace anything has a lower activation energy than either of the other two architectures.

This is what we meant when we argued that the settlement layer is where the real competition happens. The winner does not have to be the best money. It has to be the money that requires the least change from everyone else.

The referee objects

On August 31, at the Jackson Hole Economic Symposium, Pablo Hernández de Cos, general manager of the Bank for International Settlements, questioned whether stablecoins can work at scale at all.

His objections were specific. Redeemability, meaning whether holders can reliably get their dollar back. Interoperability, meaning whether one stablecoin can be exchanged cleanly for another. Financial integrity, meaning the regulatory gaps in blockchain infrastructure relative to banking. Dollar concentration, meaning the sovereignty problem created when the world's tokenized money is denominated in one country's currency. And fragmentation, meaning the conversion inefficiencies and fee exposure that appear when there are many stablecoins rather than one.

He argued for tokenized deposits instead.

He also runs the institution behind Project Agorá. That does not make him wrong. It does mean the most prominent public critique of the bank stablecoin consortium came from the sponsor of its main competitor, which is worth holding in mind while reading it.

Take his last objection seriously though, because the week proved it. Fragmentation is not a hypothetical risk that might emerge if the market goes badly. It is the observable current state, produced by the same institutions, on purpose, in parallel.

Coexistence on paper, friction in practice

Simon Taylor has argued in Fintech Brainfood that the tokenized deposit versus stablecoin fight is a distraction. His framing is that banks multiply money through lending while stablecoins move it across borders without permission, and that these are complementary functions rather than competing ones. "This isn't either/or," he wrote. "It's and."

We think the economic logic is right and the operational conclusion is too generous.

Coexistence is easy to assert at the level of function. Of course a corporate treasurer wants cheap credit from a tokenized deposit and instant global settlement from a stablecoin. Nobody disputes that both jobs exist.

The difficulty is that coexistence has to be built by someone, and this week nobody built it. The consortium is building a coin. Agorá is building tokenized deposits. Bottomline and Chainlink are building a bridge that presumes neither. Three groups, overlapping membership, no shared conversion layer, no agreed standard for moving value between the models.

Under the MM Trust Layer Model, settlement trust requires that a recipient know what they are receiving and what it can be redeemed for. A world with a bank consortium coin, tokenized deposits at a dozen institutions, several crypto-native stablecoins, and a Chainlink bridge underneath is a world where that question has a different answer depending on which counterparty you asked.

That is the fragmentation Hernández de Cos described. Taylor's "and" is correct as a destination. It is not a description of anything that currently exists.

The questions the consortium has not answered

Go back to the language in the announcement. The venture promises a safe, robust, and trusted solution combining bank-grade compliance, strong governance, distribution, and institutional risk management.

Every noun in that sentence is a promise about governance. None of them is a specification.

What backs the coin, and where do the reserves sit? Cash, Treasury bills, and at which of the 21 institutions? Reserve composition is the entire credit story, and a coin backed by deposits held at its own owner banks has a circularity problem worth explaining before launch rather than after.

Who redeems, at par, on demand, and within what window? Redeemability was the first objection Pablo Hernández de Cos raised, and it is the one that decides whether this is money or a claim on a committee.

If it breaks the peg, whose balance sheet absorbs the loss? The GENIUS Act places obligations on the issuer. The issuer here is a new entity jointly owned by institutions that compete with each other in every other market they operate in. Loss allocation among 21 competing owners is the kind of clause that gets negotiated for a year.

And what does the coin do that Tether and Circle do not already do? The honest answer is compliance and distribution, which are real assets. But liquidity is a network effect, and network effects do not transfer with a press release. A dollar that 21 banks will honor and few venues will quote is a slower dollar than the one already trading.

None of this is fatal. Consortium infrastructure gets built in payments regularly, and some of it works extremely well. It is simply the hard part, and it is the part where a venture with 17 named members and no announced name is furthest from done.

The technology was never the constraint here. Twenty-one legal departments are.

What to watch

Three things will tell you which architecture is winning, and none of them is a press release.

The first is whether the consortium ships a name and a governance structure by the end of 2026. Twenty-one institutions agreeing on reserve management, redemption rights, and who takes the loss in a depeg is a harder problem than the technology. If the entity slips past incorporation, the coin is a negotiating position rather than a product.

The second is whether Agorá's real-value testing produces a live cross-border corridor. A prototype that settles tokenized deposits against tokenized reserves atomically is impressive and has been impressive for a while. Regulatory frameworks are still developing, and implementation timelines remain uncertain, which is the polite formulation for not yet.

The third is conversion. The moment somebody has to move value from the consortium coin into a tokenized deposit and finds there is no clean path, the fragmentation stops being a theoretical objection raised at Jackson Hole and starts being a reconciliation problem on a Tuesday morning.

We have watched this pattern before in how crypto market structure rules reshape settlement and in the card-issuance layer being built on stablecoin rails. Infrastructure arrives before the interoperability that makes it usable. It always has.

The banks have decided that being late to tokenized money is more dangerous than being fragmented in it.

They are probably right. That does not make the fragmentation any less expensive.

Sources

If 21 banks cannot agree on one architecture for tokenized money, what makes anyone think the market will settle on one for them?

Charlie Major is a Product Development Manager at Mastercard. The views and opinions expressed in Major Matters are his own and do not represent those of Mastercard.