A tokenised deposit and a stablecoin can travel on the same blockchain, settle a transaction within seconds and execute a programmed instruction. They do not represent the same debt. The first is a claim on a bank recorded as a token. The second is a claim against a specialised issuer, or a right defined by its terms, supported by a separate reserve. The rail may be identical, while the balance sheet, access and exit procedure are not.

The distinction returned to the foreground after a 28 August speech by Pablo Hernández de Cos, General Manager of the Bank for International Settlements. The BIS argues that tokenised deposits offer a more direct way to modernise payments without breaking their anchor in central bank money. It also acknowledges that no genuinely interoperable, multi-bank and cross-border tokenised-deposit ecosystem exists yet.

Marc Norat’s original contribution is to test that preference against two observable systems: JPM Coin, a dollar deposit issued by JPMorgan on Base, and USDC, Circle’s stablecoin across public networks. Comparing claims, access and data reveals a less binary trade-off than “banks versus crypto”. Tokenised deposits preserve the banking perimeter more effectively; stablecoins circulate much more broadly. Neither automatically makes separate forms of money interchangeable.

A token does not tell you who owes the money

A blockchain describes a state: an address controls a number of units and may transfer them if the contract rules allow it. The ledger alone does not establish the legal nature of those units.

With a tokenised deposit, the bank remains the debtor. JPMorgan describes JPM Coin, whose onchain ticker remains JPMD, as a dollar bank deposit available on Base to approved counterparties. A client funds a Blockchain Deposit Account, converts an amount into tokens and can move it back into that account. The funds stay within JPMorgan’s banking infrastructure. The central exposure is therefore to the bank, supplemented by the account rules, jurisdiction and deposit protection where it applies.

For USDC outside the European Economic Area, Circle’s terms describe a token issued by Circle Internet Financial and backed one-for-one by dollars or dollar assets held in segregated accounts. A holder does not own a deposit at each reserve bank. The holder has a right attached to USDC. Direct redemption for dollars requires an eligible Circle Mint account; someone without that account must first become eligible or use a market intermediary.

The difference is therefore less about the token’s appearance than four questions: who is the debtor, where is the corresponding asset, who can request conversion, and what mechanism applies if the responsible institution fails?

GatherHub’s claims map

Our comparison separates five properties often compressed into the phrase “digital money”.

| Property | JPM Coin tokenised deposit | USDC stablecoin | What blockchain does not decide | |---|---|---|---| | Primary debtor | JPMorgan Chase Bank | Circle, under its terms and applicable jurisdiction | Solvency and the resolution regime | | Economic backing | Bank balance sheet and deposit account | Segregated reserve of deposits and liquid assets | Availability of assets under stress | | Access | Institutional clients and approved counterparties | Broadly transferable; direct redemption subject to eligibility | Who can actually exit at par | | Circulation | Public Base network, permissioned transfers | Multiple public blockchains and self-custodied wallets | Equivalence across chains and bridged tokens | | Return to bank money | Through a Blockchain Deposit Account | Through Circle Mint or a secondary market | Timing, cost and liquidity at redemption |

This map avoids two shortcuts. A 100%-backed stablecoin is not a bank deposit: reserves, custody and redemption remain separate mechanisms, as our method for reading stablecoin reserves explains. Conversely, calling a token a “deposit” does not make it universally transferable. The user still needs admission by the bank, contract and network.

An internal payment is not interbank settlement

In the simplest case, two customers of one bank use the same system. The bank debits one claim and credits another on its own balance sheet. A token can automate timing, conditions and reconciliation. The transfer is fast partly because the debtor does not change.

The hard problem returns when the recipient uses another bank. If bank A issues one token and bank B another, someone must organise conversion and settle the resulting debt between them. The BIS proposes anchoring that settlement in central bank money, potentially in tokenised form. Without a common settlement asset and common rules, two deposits in the same currency can become separate walled gardens.

Experiments demonstrate both feasibility and limitation. In September 2023, Citi announced a tokenised-deposit pilot for moving liquidity between its own branches on a private blockchain controlled by Citi. JPMorgan now markets Blockchain Deposit Accounts in eight currencies, but JPM Coin on Base is currently dollar-denominated and limited to its approved institutional clients. These products reduce cut-off times inside a perimeter; they do not yet form open multi-bank money.

“24/7” must also be read layer by layer. JPMorgan says transfers on its network operate continuously, while disclosing a three-hour interruption each Saturday for movements between traditional accounts and Blockchain Deposit Accounts. The ledger can remain open while an entry or exit ramp still follows a legacy calendar.

JPMD makes the scale gap visible

JPM Coin introduces an important change: a regulated bank deposit circulates on Base, a public blockchain, rather than only on the bank’s private ledger. Public does not mean permissionless. JPMorgan says that only vetted and approved counterparties can transact.

The exact contract linked by JPMorgan supports a minimal independent check. At Base block 50,986,259, timestamped 7 September at 06:17:45 UTC, the standard `totalSupply()` call returned 100,100,000 raw units with two decimals, equal to 1,001,000 JPMD. That figure measures the public token’s supply, not JPMorgan’s total deposits or all activity across Kinexys.

The distinction matters. Kinexys’s general page claims more than $7 billion in average daily transaction volume during 2025 and more than $3 trillion since inception, but those proprietary numbers cover several platform services. Attributing them to JPMD would be misleading. Ledger Insights independently confirms the November 2025 production launch and initial transactions involving B2C2, Coinbase and Mastercard, without publishing current token-specific volume.

The tokenised deposit has therefore become observable without becoming open. Its contract and supply can be checked; that does not disclose active client numbers, settled volume, concentration of economic holders or individual account terms.

USDC gains reach by separating itself from a bank balance sheet

The difference in size is striking, although supply must not be confused with use. DeFiLlama tracked $74.706 billion of USDC in circulation on 7 September 2026. That is roughly 74,600 times the nominal JPMD supply observed that morning. USDC runs on multiple networks and can pass between wallets without making every holder a direct Circle customer.

That openness moves the controls elsewhere. Circle publishes reserve composition weekly and obtains monthly assurance from an accounting firm. Its terms nevertheless distinguish Circle Mint users, who can issue and redeem, from other holders who are not Circle customers. Secondary markets connect these two populations at a price that normally remains near one dollar but is not technically enforced in every trade.

The BIS uses an exchange from USDT into USDC to explain the problem: two dollar-denominated tokens may require a sale and a purchase, with deviations from par possible under stress. Fragmentation also exists across blockchains. The same commercial name may refer to a native issue or a bridged token whose risk depends on another contract.

Stablecoins have thus achieved distribution that tokenised deposits still lack. In return, holders must inspect the reserve, intermediaries, redemption right and exact chain. Our recent analysis of stablecoin group rules also shows that licensing the issuer does not always encompass every activity conducted by its affiliates.

The BIS preference is a programme, not an observed outcome

The BIS advances three arguments for tokenised deposits: par conversion through central bank money, preservation of bank funding and identity controls inside a supervised perimeter. These advantages are credible, but they describe a complete architecture that still has to be built.

The speech acknowledges the obstacles: non-interoperable permissioned platforms, potential dominance by large banks, high costs for smaller institutions, access governance and potentially faster outflows during a panic. A continuously available deposit can accelerate a liquidity flight as effectively as it accelerates a useful payment.

Stablecoins and tokenised deposits are therefore not interchangeable versions of one product. The former currently prioritise reach and composability; the latter prioritise legal continuity with a bank. Marc Norat’s conclusion is that the real test is not token speed but continuity of the claim when a payment changes bank, network or jurisdiction.

Three milestones will show whether the model advances: a live settlement across several banks using central bank money, statistics that clearly separate JPMD from aggregate Kinexys activity, and conversion procedures that work during market stress. Until then, blockchain modernises circulation. It does not yet make every form of digital money equivalent.