Three weeks, three answers to the same question. The split is about who keeps the deposit and who gets to move it.
On August 28, the general manager of the Bank for International Settlements told an audience at Jackson Hole that tokenised deposits, not stablecoins, are the credible path for money at scale. On September 1, twenty-one banks, including Bank of America, Citi, Goldman Sachs and Wells Fargo, announced they would form a company to issue a dollar stablecoin. On September 10, U.S. Bank reported a completed test payment in its own dollar stablecoin, sent across the Atlantic.
Read those together, and the picture is not a market converging on one design. It is a market sorting itself by use case, with the largest bank in the United States refusing one of the two instruments while most of its peers quietly build both.
The two designs, stated plainly
A payment stablecoin is a liability of its issuer, backed by segregated reserves, and it travels to anyone who can hold it. Under the new American regime the issuer cannot pay holders interest, and reserves held as bank deposits are insured to the issuer rather than to the people holding tokens.
A deposit token is a digital representation of an existing bank deposit. It sits on the bank’s balance sheet, inside the two-tier system where commercial bank money and central bank money already operate, and it is generally limited to that bank’s clients.
JPMorgan’s version has been live since November 2025. JPMD is a USD deposit token issued through Kinexys on Base, available to institutional clients, described by the bank as a regulated dollar deposit with near-instant settlement around the clock. Coinbase’s Jesse Pollak framed the technical promise as “sub-second, sub-cent, 24/7 settlement.” The money never leaves the bank.
What the twenty-one banks decided
The consortium announced on September 1 is the larger commitment. Twenty-one institutions, among them Capital One, PNC, TD, Scotiabank, UBS, Santander, BBVA, Deutsche Bank, Commerzbank, Lloyds, Crédit Agricole, Rabobank and MUFG, plan to stand up a joint stablecoin company by the end of this year, issue a dollar token in the first half of 2027, and follow with other G7 currencies.
The reason they chose a stablecoin rather than extending their own deposit tokens is structural. A deposit token is a claim on one bank. Two banks’ deposit tokens are two different instruments, and a payment between them needs a bridge or a settlement arrangement. A jointly issued token is one instrument that every member can distribute, which is the only way to get fungibility across twenty-one balance sheets without building a network to reconcile twenty-one ledgers.
Industry analysis on the day made the same point about reach: a stablecoin is designed to “move between wallets, platforms and public blockchain networks without requiring every participant to maintain a direct account relationship with the issuing bank,” while a tokenized deposit “remains a commercial bank liability” that preserves the banking relationship.
The same analysis flagged what the consortium has not solved. Governance is the test, because success depends on who decides when twenty-one members’ commercial interests diverge.
What JPMorgan decided
JPMorgan is not in the group and has said it has no plans to issue a stablecoin. It has a deposit token instead, and its argument is visible in the design.
Keeping the instrument on the balance sheet means the deposit keeps funding the bank’s lending. It means the client relationship stays intact, because holding the token requires being a client. And it means the bank avoids the reserve model entirely: no segregated portfolio to manage, no issuance licence to obtain, no prohibition on paying interest, because a deposit can pay interest in the ordinary way.
What it gives up is reach. A token that only the bank’s clients can hold is a settlement tool for institutional flows, not a payment instrument for the open market.
Read more: GENIUS Act implementation, tracked rule by rule • New payment chains court banks with rails built for dollars
The shared rail both camps joined
Between the two announcements sits a third one that has drawn less attention and may matter more than either.
On June 5, seventeen banks committed to an on-chain money initiative operated by The Clearing House, built to clear and settle tokenized deposits between institutions and connect that activity to the RTP and CHIPS networks. The participants are Bank of America, BMO, BNY, Citi, Citizens, Fifth Third, HSBC, Huntington, J.P. Morgan, KeyBank, PNC, Regions, Santander, TD Bank U.S., Truist, U.S. Bank and Wells Fargo.
That list is the answer to the fungibility objection. A deposit token is a claim on one bank, and two banks’ tokens are two instruments, but a shared clearing layer is how the banking system has always made one bank’s deposit spendable at another. Max Neukirchen of J.P. Morgan put the reasoning in a sentence: “A regulated market-infrastructure solution for clearing and settling tokenized deposits is essential to keeping payments stable and effective.”
Now hold the two rosters side by side. Bank of America, Citi, Wells Fargo, PNC, TD and Santander appear in both the June deposit network and the September stablecoin consortium. JPMorgan, BNY, Truist, U.S. Bank, HSBC and the regional banks appear only in the first. Goldman Sachs, Capital One and the European members appear only in the second.
Read that way, the September announcement looks less like a schism and more like an extension. Most of the large American institutions had already agreed to settle deposits with each other. A subset then decided they also needed an instrument that could leave the club, because a deposit rail reaches only the banks connected to it.
U.S. Bank shows how untidy the positions are
One institution makes the point better than the rosters do.
U.S. Bank joined the deposit network in June. On September 10 it reported a completed test payment between North America and Europe using USBDC, a dollar stablecoin it has been developing with Stellar and PwC since late 2025. Jamie Walker, who leads digital assets and money movement at the bank, called the pilot “another step forward in our broader digital asset strategy” and framed it around solving client problems rather than choosing an instrument.
That is the honest state of most large bank strategies right now. The public arguments are about which design is correct. The budgets are being split across both, because nobody wants to explain in 2028 why the bank sat out the one that worked.
The central bank case for deposits
The most complete argument for the deposit side came from the BIS.
Speaking at Jackson Hole on August 28, Pablo Hernández de Cos applied two tests. The first is singleness: every instrument denominated in a currency should be redeemable at par into central bank money, with finality. Stablecoins fail it in his reading, because when one person holds one token and a counterparty accepts only another, nothing enforces par between them, and secondary market deviations happen.
The second is interoperability and integrity. His line on that has been quoted widely since: “even the ‘same’ stablecoin on different chains is not interoperable without risky or costly workarounds.” He paired it with the difficulty of enforcing anti-money-laundering rules on pseudonymous public networks.
Tokenised deposits, in this view, get the technology without breaking the structure that makes bank money work, because they stay inside a supervised system where deposits and credit remain linked.
The counter-case, which the banks are funding
The rebuttal is simpler and commercial. Money that can only move between clients of one institution is not money for payments. It is an internal settlement upgrade.
Every use case the card and payments industry cares about involves someone outside the bank: a merchant acquirer in another country, a wallet holder with no banking relationship, a program manager settling with a network. A stablecoin reaches all of them. A deposit token reaches whoever the bank already onboarded.
The regulatory ground has also shifted in the stablecoin design’s favour. The GENIUS Act created a federal licence for issuance, which means a consortium can issue a dollar token as a regulated product instead of an experiment. And the FDIC’s proposed standards state that tokenized deposits are not payment stablecoins, which keeps deposit tokens outside the new regime and its interest prohibition, but also outside the distribution rights the regime confers.
Note that the two answers are not exclusive. Carl Slabicki, who runs commercial global payments at BNY, described the work as running in parallel: upgrading the existing rails to operate around the clock while “double-clicking and focusing on building a tokenized strategy for accounts, deposits” and the interoperability between them. Several consortium members already run deposit token projects, and the framing on announcement day was that banks are building both.
What actually changes on the balance sheet
For a treasury team, the choice comes down to where the money sits and what it is allowed to earn.
With a deposit token, the balance remains a deposit. It funds lending, it can pay interest to the depositor, and it carries deposit insurance up to the usual limits. The bank keeps both the funding and the relationship.
With a stablecoin, the customer’s dollars leave the deposit and become reserves held against the token: currency, short-dated Treasury bills, money market funds, deposits at banks. Some of that money returns to the banking system as wholesale deposits, and it does not necessarily return to the bank the customer left. The issuer earns the reserve income and cannot pass it to holders.
That is the whole split in two sentences. A deposit token keeps the funding and the ability to pay for it. A stablecoin trades both away for reach.
How big the deposit question is
The numbers being argued over are large enough to explain the urgency, and they are projections rather than measurements.
Treasury research cited in the congressional debate identifies about $6.6 trillion of transactional deposits at risk if stablecoins scale with yield attached. Citigroup has projected that stablecoins could displace between $182 billion and $908 billion of bank deposits by 2030, and Federal Reserve estimates in the same discussion put the potential reduction in bank lending anywhere from $65 billion to $1.26 trillion.
The range is the honest part. Nobody knows, and the spread between the low and high estimates is wide enough to support any strategy a bank already wanted to pursue. What is not in dispute is the direction: a dollar held as a token is a dollar not held as a deposit at a particular bank.
What this means downstream
For card programs, wallets and payment companies, the practical consequence is about who can be paid.
A program funded by a deposit token can only serve customers of the issuing bank, which makes it a strong fit for corporate treasury flows and a poor one for consumer products. A program funded by a stablecoin can serve anyone with a wallet, and the choice of token then becomes a question about the issuer’s licence, reserves and redemption terms.
There is also a marketing line worth getting right. Deposit tokens carry deposit insurance in the ordinary way because they are deposits. Stablecoin balances do not, and the FDIC has said so in its proposal. Programs that blur the two on a product page are creating a problem they do not need.
The chain choice sits alongside all of this, since both designs have to run somewhere, and the banks have been picking networks with as much care as instruments, a subject covered in our piece on the new payment chains. The route where a network itself offers banks a minting service is covered separately in our story on Visa’s minting service.
What to watch into 2027
Four markers will tell you which way this is settling.
Whether the consortium’s operating company is actually formed by the end of December, since a missed date on the first milestone says something about the governance problem. Whether any member defects or any large holdout joins, particularly JPMorgan. Whether a deposit token gets used between banks rather than inside one, which would answer the fungibility objection without a consortium. And whether the agencies clarify how tokenized deposits are supervised once the stablecoin rules are final, because right now the deposit route is defined mostly by what it is not.
The honest summary for anyone choosing a side today is that both designs work and they solve different problems. Banks that care most about keeping funding will build deposit tokens. Banks that care most about reaching customers they do not have will issue stablecoins. Most of the large ones have concluded they need to do both, which is expensive, and is also the only answer that does not require them to guess correctly.