Why Can't Two Banks' Deposit Tokens Talk to Each Other?
Because a deposit token is a claim on one named bank, and no shared ledger exists to settle that claim against another bank's. JPMorgan's Kinexys platform averages more than $7 billion in daily volume and has cleared $4 trillion since launch. Citi Token Services moves value between New York, London and Hong Kong. Neither can settle with the other. In April 2026 Nellie Liang, the former Treasury under secretary now at Brookings, stated the position directly: interbank settlement of tokenized deposits on private blockchains does not exist. JPMorgan, Bank of America, Citigroup and Wells Fargo are now building a shared network through The Clearing House, targeting the first half of 2027. This guide sets out why the gap exists, what the shared network has to solve that the individual platforms did not, and why the four bank blockchain consortia that came before all closed.
TL;DR — Key Takeaways
- ✓The Gap: Kinexys clears $7 billion a day and $4 trillion cumulatively. It cannot settle against Citi Token Services. Interbank settlement of tokenized deposits on private chains does not exist.
- ✓The Fix Under Construction: JPMorgan, BofA, Citi and Wells Fargo are building a shared network via The Clearing House for H1 2027, with more than a dozen banks expected.
- ✓The Precedent: we.trade (2022), Marco Polo (Feb 2023), Contour (late 2023) and the USDF Consortium all failed. The technology was never the binding constraint.
- ✓The Hard Part: Who carries credit exposure when a token sits between two banks overnight, and how it settles when Fedwire is closed.
- ✓The Pressure: $263 billion in stablecoins circulating and $226 billion in B2B stablecoin payments in 2025. Brian Moynihan has warned $6 trillion of deposits could migrate.
A Working System With No Exits
JPMorgan's Kinexys platform averages more than $7 billion in daily volume and has processed $4 trillion since launch. By any measure that is a functioning payment network. It also cannot send a dollar to a client of Citigroup.
This is the shape of the tokenized deposit market in 2026: several real, high-volume, internally complete networks, each closed at its own perimeter. A deposit token is a claim on the specific bank that issued it. Moving it to a client at another bank means extinguishing one bank's liability and creating another's, which is interbank settlement, and interbank settlement needs infrastructure that no individual bank can build alone.
Interbank settlement of tokenized deposits on private blockchains “does not exist.”
— Nellie Liang, former US Treasury Under Secretary for Domestic Finance, writing with colleagues at the Brookings Institution, April 2026
That sentence is the whole problem stated in six words. The rest of this article is about what it takes to make it untrue, and why the last four attempts did not manage it. The distinction between a deposit token and a stablecoin — a bank liability against a bearer instrument — is covered in our guide to how tokenized bank deposits work for institutions; this article assumes it.
What Each Bank Has Built Alone
Every major US deposit token platform is single-issuer by construction: the bank mints tokens against its own deposit liabilities and its own clients hold them. Kinexys is the largest at $7 billion daily and $4 trillion cleared. Citi Token Services runs real-time transfers across New York, London and Hong Kong. Wells Fargo announced its platform on 4 August 2026.
| Platform | Status | Scale or scope | Settles with other banks? |
|---|---|---|---|
| JPMorgan Kinexys (JPM Coin / JPMD) | Live | $7bn daily average; $4tn cleared to date | No |
| Citi Token Services | Live | New York, London, Hong Kong corridors | No |
| Wells Fargo tokenized deposits | Announced 4 Aug 2026; fall launch | Selected corporate clients, USD to GBP at launch | No |
| The Clearing House shared network | In construction | Target H1 2027; 12+ banks expected | That is the entire point |
Wells Fargo is the clearest illustration of the two-track reality. Mike Santomassimo, the bank's CFO, framed the proprietary launch around clients moving money “between accounts and across borders with greater ease and increased speed” — accurate, and scoped to Wells Fargo's own clients on Wells Fargo's own blockchain. The same bank is simultaneously co-building the shared network for H1 2027. A bank can ship the first without asking anyone. The second requires agreement from competitors.
The result today is that a corporate treasurer holding deposit tokens at two banks holds two unrelated instruments. They net against nothing. Where a payment has to cross the perimeter, it falls back to conventional rails, which is exactly the outcome tokenization was meant to improve on.
How Much of Dollar Settlement This Actually Represents
Kinexys at $7 billion a day is roughly 0.15% of Fedwire's $4.6 trillion daily average for 2025, and about 0.35% of the $2 trillion CHIPS averaged over the same year. The tokenized rails are operating at production volumes and are not yet a material share of dollar settlement.
This ratio is worth holding onto, because it cuts in both directions. It punctures the claim that deposit tokens have displaced anything. It also explains the urgency: the incumbent rails close. Fedwire does not run on a Saturday, and a settlement network that stops for the weekend is a poor foundation for a market that does not. Continuous settlement is the single capability the tokenized rails have that the incumbent ones do not, and it is worth very little while it only works inside one bank.
$263 billion in stablecoins circulating, and $226 billion in business-to-business stablecoin payments during 2025 — against bank deposit token networks that cannot pay each other.
— Market figures compiled in Forbes, “America's Biggest Banks Are Building One Deposit Token,” 28 July 2026
Stablecoins solved interoperability by not being bank liabilities at all. A bearer token on a public chain moves to anyone with a wallet, with no interbank settlement question because there is no bank on either side. Banks are unwilling to accept that trade-off — it is precisely the deposit franchise they are defending — so they have to solve the harder version. Brian Moynihan, Bank of America's chief executive, has warned that up to $6 trillion of deposits, around 30-35% of all US commercial bank deposits, could migrate if stablecoins are permitted to pay interest. Which US entities may issue those instruments at all is governed by the perimeter rules we cover in what counts as issuing a stablecoin in the US.
Four Bank Blockchain Consortia That Are No Longer Operating
Every previous attempt at a shared bank blockchain network has closed: we.trade in 2022, Marco Polo in February 2023, Contour in late 2023, and the USDF Consortium has shown minimal activity since 2024. In none of these cases did the software fail to work. The shared commercial model did.
| Venture | Backers | Purpose | Outcome |
|---|---|---|---|
| we.trade | HSBC, Deutsche Bank, Santander | Trade finance | Insolvent, 2022 |
| Marco Polo | BNY, Commerzbank | Trade finance | Insolvent, February 2023 |
| Contour | Nine banks | Digitizing letters of credit | Shut down, late 2023 |
| USDF Consortium | Community banks, formed 2022 | Bank-minted tokenized deposits | Minimal activity since 2024 |
The common failure pattern is worth naming precisely, because it is not a technology story. A shared network is only valuable once enough participants join that a member can reach most of its counterparties through it. Below that threshold each bank is paying to maintain a channel it barely uses, and the rational move is to deprioritise it. Every bank reasoning that way simultaneously means the threshold is never crossed. Contour digitized letters of credit competently and closed anyway.
The Clearing House project differs on two specifics that matter. It is operated by an entity the participating banks already own and already route payments through, rather than a new venture that must reach viability on its own balance sheet. And its four founding members are among the largest US banks by deposits, so the network reaches a substantial share of US corporate banking on day one rather than needing to accumulate it. Neither guarantees anything. Both address the mechanism that killed the previous four.
The Question Nobody Has Answered: Who Holds the Risk Overnight?
The central obstacle is credit exposure between banks. When a token representing a claim on Bank A moves to a client of Bank B on a Saturday, Bank B has extended value against a claim it cannot settle until Fedwire reopens. Someone carries that exposure across the gap, and the network cannot launch until the rule for who is written down.
This is the same structural issue as any settlement system, with one difference: continuous operation is the product. A network that only settles when the Federal Reserve is open has surrendered its main advantage over the rails it replaces. So the design has to either hold intraday and weekend credit exposure between competing banks, prefund positions at a capital cost, or wait for central bank money that settles continuously. Each option shifts cost somewhere a participant will argue about.
The three ways to close a weekend gap
- Extend credit. The receiving bank accepts an unsettled claim until the RTGS system reopens. Cheapest to build, and it puts interbank credit risk into a payment product that was sold as risk-reducing.
- Prefund. Each bank posts collateral or maintains a balance sized to its expected weekend flow. Removes the credit question and ties up capital continuously to serve intermittent volume.
- Settle in tokenized central bank money. Removes both problems and requires the central bank to operate continuously. Hong Kong has stated this as a planned upgrade path; the US has not.
David Watson, chief executive of The Clearing House, has called the project “a big move for the banks,” which is a fair description of a group of direct competitors agreeing on mutual credit terms. Whether that agreement holds through to H1 2027 is the thing to watch, and it is a governance question rather than an engineering one. The related problem of when a transfer becomes legally irrevocable is covered in when a tokenized transfer is actually final.
What a Tokenized Asset Issuer Should Do Before H1 2027
Treat the cash leg as bank-specific and design for it. A tokenized fund settling subscriptions in deposit tokens can only offer atomic delivery-versus-payment to investors who bank where those tokens are issued. Every other investor settles on conventional rails, which means two operational paths for one instrument.
This is the part most often skipped in tokenization design. The asset leg gets attention — the token standard, the transfer restrictions, the registry — while the cash leg is assumed to be solved because deposit tokens exist. They do exist, and they exist per bank. An issuer whose investors bank at five institutions has a fragmented cash leg regardless of how well the security token is engineered.
Practical checks before committing to a deposit token cash leg
- Map investor banking relationships first. If they concentrate at one institution, a single-bank deposit token works today. If they spread, plan for dual-path settlement into 2027.
- Define the fallback explicitly. Document what happens when the cash leg cannot settle on-chain, rather than discovering it during a redemption window.
- Do not price atomicity you cannot deliver. If DvP only holds for a subset of investors, the offering documents should say so.
- Watch the H1 2027 date without depending on it. Four predecessors did not ship. Build the assumption that the shared network arrives late, and treat early arrival as upside.
The wider point is that tokenizing an asset does not tokenize the money that buys it. Those are two separate infrastructure programmes on two separate timelines, and only one of them is under the issuer's control. For the broader architecture this sits inside, see our institutional guide to RWA tokenization.
Who This Matters To, and Who Can Ignore It
It matters most to anyone settling value across institutional boundaries on a continuous schedule. It matters least to a single-bank corporate treasury, which already gets most of the benefit from a proprietary platform available this year.
| Situation | Affected by the gap? | Why |
|---|---|---|
| Corporate treasury, one primary bank | Barely | Internal transfers stay inside the perimeter; a proprietary platform covers it now |
| Corporate treasury, multi-bank | Directly | Balances at two banks are two unrelated instruments that do not net |
| Tokenized fund issuer, concentrated investors | Manageable | One bank relationship can carry the whole cash leg |
| Tokenized fund issuer, dispersed investors | Directly | Dual settlement paths required until a shared network exists |
| Cross-border payments at scale | Directly | Correspondent chains span institutions by definition |
The honest summary of the 2026 position: deposit tokens work, they work well, and they work in one bank at a time. That is a real capability and a smaller one than the category name suggests. The gap between those two statements closes in H1 2027 if the four banks hold together — and four previous consortia did not.
Frequently Asked Questions
Why can't a JPMorgan deposit token settle against a Citi deposit token?
Because a deposit token is a claim on one specific bank, and moving it to a client at another bank means one bank's liability has to be extinguished and another's created. That is interbank settlement, and it needs a shared ledger plus an agreed rule for who carries the credit exposure in between. Nellie Liang of Brookings put it plainly in April 2026: interbank settlement of tokenized deposits on private blockchains does not exist. Each bank's network is internally complete and externally closed.
What is The Clearing House building, and when?
JPMorgan, Bank of America, Citigroup and Wells Fargo are building a shared tokenized deposit network operated by The Clearing House, the payments infrastructure company those banks already own, targeting the first half of 2027. More than a dozen banks are expected to join. The initial users are multinational corporations, and the stated use cases are programmable treasury, real-time liquidity management and cross-border payments.
Haven't banks tried shared blockchain networks before?
Four times, and all four are gone. we.trade, backed by HSBC, Deutsche Bank and Santander, entered insolvency in 2022. Marco Polo, backed by BNY and Commerzbank, went insolvent in February 2023. Contour, a nine-bank venture digitizing letters of credit, shut down in late 2023. The USDF Consortium of community banks has shown minimal activity since 2024. The technology worked in each case; the shared commercial model did not.
How big are bank token networks compared with existing payment rails?
Small. JPMorgan's Kinexys averages more than $7 billion in daily volume and has cleared $4 trillion since launch, which is the largest bank token network by a wide margin. CHIPS averaged about $2 trillion a day in 2025 and Fedwire about $4.6 trillion. Kinexys at $7 billion daily is roughly 0.15% of Fedwire's daily flow. The tokenized rails are real and operating, and they are not yet a material share of dollar settlement.
Why is Wells Fargo building its own platform and a shared one at the same time?
Because the two solve different problems on different timelines. Wells Fargo announced a proprietary tokenized deposit platform on 4 August 2026 for a fall launch to selected corporate and commercial clients, initially USD to GBP only, expanding through 2027. That serves its own clients moving money between their own accounts, which needs no external agreement. The Clearing House network, targeted at H1 2027, is what lets those clients pay someone at another bank. Owning the first does not deliver the second.
What does this mean for a tokenized asset issuer?
That the cash leg of a delivery-versus-payment trade is bank-specific until a shared network exists. If an issuer settles subscriptions and redemptions in deposit tokens, every investor must bank where those tokens are issued, or the settlement drops back to conventional rails and the atomicity is lost. Until H1 2027 at the earliest, a multi-bank investor base means a multi-rail cash leg, and that should be designed for rather than assumed away.
Related Articles
How Do Tokenized Bank Deposits Work for Institutions?
What a deposit token is and how it differs from a stablecoin.
What Does Atomic Settlement Actually Change?
Why the cash leg determines whether DvP is real.
How Does a Cross-Border Tokenized Redemption Settle?
The multi-bank path a redemption actually takes.
When Is a Tokenized Transfer Actually Final?
The legal test that sits underneath any settlement network.