Deposit Tokens vs. Stablecoins: JPMD and the Banks

Introduction

JPMorgan made JPMD available to institutional clients on November 12, 2025, describing it as the first bank-issued USD deposit token on a public blockchain. The instrument is a tokenized claim on a demand deposit at JPMorgan, which places it on the liability side of a fractional-reserve balance sheet that lends against the same dollars it owes. A payment stablecoin issued under the GENIUS Act is constructed the opposite way: backed one-for-one by segregated cash and short-dated government paper, with the issuer barred from paying interest or yield to holders for simply holding the token.

Most coverage treats the two as competing implementations of the same product, dollars that move on a blockchain. They are different instruments with different creditors, different failure pathways, and different regulators. A deposit token holder is a bank creditor whose recovery depends on receivership priority and insurance limits. A stablecoin holder is a redemption claimant against a segregated reserve pool whose bankruptcy remoteness has never been tested at scale in a US court.

The evidence boundary here is narrow. JPMorgan’s own announcement confirms the launch date, the institutional-client framing, and the deposit-token label, and almost nothing else. It contains no settlement architecture, no interest terms, no regulatory classification, and no transfer policy. Widely circulated details, that JPMD deployed on Base with Coinbase, Mastercard, and B2C2 among the first users, that Cash App extended USDC to roughly 60 million users, that the FDIC approved GENIUS prudential standards in April 2026, and that Bank of America, Citi, and Wells Fargo are weighing their own issuance, come from reporting we could not verify against a primary source. They appear below as reported claims, not established facts.

What can be analyzed rigorously is the structure. The balance-sheet mechanics determine which use cases each instrument can serve, suggesting that deposit tokens and stablecoins suit different operational requirements rather than compete as direct substitutes for a single market.

Two liabilities, two balance sheets

The decisive difference between a deposit token and a payment stablecoin is what the issuer does with the dollars behind it. A bank takes a deposit and lends most of it out, holding a fraction in reserves and liquid assets. The depositor’s claim is an unsecured senior obligation of the bank, enhanced by deposit insurance up to statutory limits and by the priority that insured and uninsured depositors hold ahead of general creditors in a US bank receivership. A permitted payment stablecoin issuer cannot do this. The statute requires the reserve to be held one-for-one against the outstanding token supply in cash and short-term government instruments, held apart from the issuer’s own assets, and the issuer cannot pay holders a return on their balance.

This produces a clean inversion of incentives. The stablecoin issuer earns the entire yield on the reserve and passes none of it to holders, so its profit scales with float and with interest rates, and its business is destroyed by neither credit losses nor a duration mismatch. Its risk is concentrated in custody, reserve quality, and the mechanics of redemption under stress. The bank earns a spread between what it pays depositors and what it earns on loans, so its profit depends on credit underwriting and on funding that stays put. Its risk is concentrated in the asset side of the balance sheet, which is exactly where a stablecoin issuer has no exposure.

Tokenization does not change either economics. It changes the speed and the hours at which the liability can move, and that turns out to matter more for the bank than for the issuer, for reasons taken up below.

The comparison across the dimensions that determine use-case fit looks like this.

The one genuinely confirmed cell in the JPMD column is the launch date. Everything else in that column is either a structural inference from what the word “deposit” means in US banking law or an open question that JPMorgan has not answered publicly.

The disclosure gap around JPMD

JPMorgan has disclosed less about JPMD than the volume of commentary implies. The newsroom post establishes that JPMD became available to institutional clients after a proof-of-concept phase and that the bank considers it the first bank-issued USD deposit token on a public chain. It does not describe how a transfer on a public ledger interacts with JPMorgan’s core banking system, and that interaction is the whole engineering problem.

Three architectures are consistent with the public framing, and they differ enormously in what a holder actually owns.

In the first, the public-chain transfer is the settlement event, and the core ledger is updated from it in near real time, with the token contract treated as an authoritative sub-ledger of deposit balances. In the second, the token is a representation that moves freely between whitelisted addresses while the core ledger is reconciled in batches on banking-day cycles, leaving the bank carrying an unreconciled position between the moment of transfer and the moment of booking. In the third, the token is a messaging and instruction layer over a permissioned database, where a transfer does not move a deposit at all until JPMorgan’s internal systems act on it.

Each carries different risks. Batched reconciliation creates a window in which the onchain state and the bank’s books disagree, which matters most during weekends and holidays when a counterparty might transfer a balance that the bank has not yet debited anywhere. A pure representation layer would mean the public-chain deployment is closer to an API than to settlement, and the comparison with stablecoins would be largely cosmetic. We have no basis in the sourced material to say which of these describes JPMD, and the distinction is the first thing a counterparty underwriting the instrument would need to resolve.

The same gap applies to transfer policy. If JPMD can only be held by addresses that JPMorgan has whitelisted, the token inherits the bank’s onboarding perimeter, and its public-chain deployment buys programmability and settlement hours rather than open transferability. That yields an operational product distinct from an open stablecoin.

Interest, liquidity rules, and the treasury calculation

The widely repeated claim that interest-bearing capability gives deposit tokens a decisive advantage in corporate treasury is weaker than it appears, and prudential liquidity rules are the reason. A bank can pay interest on a deposit, and a GENIUS stablecoin issuer cannot pay yield on a token. On that basis a corporate treasurer choosing between an idle stablecoin balance and an idle deposit-token balance should prefer the one that earns. Whether JPMD actually pays anything is not disclosed in the material available to us, and the inference that it must rests on the word “deposit” rather than on a published term sheet.

The structural problem sits one layer down. Bank liquidity regulation does not treat all deposits alike. Wholesale deposits from financial and corporate counterparties attract assumed run-off rates that depend on how operationally entangled the depositor is with the bank and how quickly the balance can leave. A deposit that can be moved in seconds, at any hour, by a counterparty with no operational dependency on the bank, is close to the regulatory archetype of unstable wholesale funding. The faster and more frictionless the token makes the exit, the more high-quality liquid assets the bank must hold against the balance, and the less of the spread survives to be shared with the depositor.

This creates a tension inside the product. The feature that makes a deposit token attractive as a settlement instrument, instant movement outside banking hours, degrades its value as a funding instrument. A bank can resolve the tension by paying little or nothing on tokenized balances and positioning them as working capital that passes through rather than sits, or by paying competitively and accepting a worse liquidity profile on funding it can lose at machine speed. The first option narrows the yield advantage over stablecoins to near zero. The second is expensive.

There is a further reason to doubt that yield decides the corporate case. Large corporate balances sit far above deposit insurance limits already. The practical comparison balances uninsured bank credit exposure against a claim on a segregated Treasury portfolio. Treasurers who hold money market funds rather than bank deposits have already made that trade, and they did so without a blockchain. A tokenized deposit that pays below the money market rate and carries single-bank credit exposure is a worse store of value than what they hold now, whatever its settlement properties.

Where the deposit token has a genuine advantage is in the balance that must be live for payment purposes anyway. If a firm’s payables run through JPMorgan, keeping that working balance in a form that can settle onchain at 3 a.m. against a counterparty in another time zone is an operational improvement over wire cutoffs, and the yield question barely enters. The corporate case for deposit tokens centers on float mechanics rather than interest.

Whether a deposit token can leave its issuer

A deposit token’s reach is bounded by its issuer’s client list in a way that a stablecoin’s is not, and no public source establishes that JPMD has escaped that bound. When a JPMD holder transfers to another JPMorgan client, nothing leaves the bank. The deposit claim changes owner, aggregate deposits are unchanged, and the bank’s funding is undisturbed. When the recipient is not a JPMorgan client, the bank faces a choice it cannot avoid: onboard the recipient as a depositor, or settle the obligation out through conventional interbank rails and lose the deposit. There is no third option, because a deposit claim on JPMorgan held by someone JPMorgan has no relationship with is an awkward object in both law and compliance.

This is the structural reason deposit tokens tend toward closed loops. The issuer has a direct economic interest in keeping the balance inside its own perimeter, and a regulatory obligation to know who holds its liabilities. A stablecoin issuer has the opposite incentive. Every new holder is more float and more reserve income, and the issuer is largely indifferent to who passes the token to whom between mint and redemption.

If the reporting that Bank of America, Citi, and Wells Fargo are examining their own deposit tokens proves accurate, the fragmentation problem becomes concrete. Four bank tokens, each denominated in dollars, each a claim on a different balance sheet with different credit quality, do not aggregate into a payment network. They are four closed loops, and moving between them requires either mutual acceptance of each other’s credit or a settlement layer that nets positions and discharges them in central bank money. That is correspondent banking with new plumbing, and it reintroduces precisely the friction and intraday credit exposure that a single widely-accepted liability avoids.

The counterargument is that banks have solved multilateral settlement before and can standardize again, through a consortium structure, a shared netting venue, or eventual settlement in a wholesale central bank instrument. That is plausible and it is also slow. In the interval, a stablecoin’s main competitive property is its fungibility: one issuer’s liability that every venue, wallet, and counterparty treats as the same asset regardless of who holds it.

Failure asymmetry: resolution versus redemption

Deposit token holders and stablecoin holders both face untested recovery processes, but the uncertainty sits in different places, and the deposit-token uncertainty is sharper than the insurance label suggests. In a US bank failure, the FDIC is appointed receiver, insured depositors are made whole up to the statutory limit, uninsured depositors hold a priority claim on the receivership estate ahead of general unsecured creditors, and recovery above the insurance cap depends on what the asset book is worth. That framework is well-established for account holders of record.

A tokenized deposit transferred to a party the bank did not onboard complicates the first step of that process. Deposit insurance determination depends on identifying the depositor and the ownership capacity in which the balance is held, which the FDIC resolves from the bank’s own records. If a token has moved through addresses the bank cannot map to a verified account holder, the receiver has to decide whether the current holder is a depositor at all, a general creditor, or someone holding an instrument with no deposit status. Nothing in the sourced material tells us how JPMorgan has structured JPMD against this problem, and the cleanest solution, restricting transfers to whitelisted and fully onboarded clients, is also the solution that most limits the instrument’s usefulness.

There is a second feature of resolution that corporate holders should price. A failed bank can be closed on a Friday with balances unavailable or partially available until the receiver acts, and uninsured portions may take time to resolve or may be haircut. A deposit token that settles continuously on a public chain does not change that. Receivership is not a 24/7 process, and the onchain representation of a frozen deposit claim is a frozen deposit claim.

A stablecoin failure runs differently. The holder’s exposure is to the reserve: whether the assets exist in the represented quantity, whether they are genuinely segregated from the issuer’s estate, whether the custodian performs, and whether redemption can be processed at scale during a run. Statutory reserve requirements and segregation reduce the probability that the assets are missing, but they do not eliminate the gap between a legal claim on segregated assets and prompt cash in hand during an insolvency. Bankruptcy courts have not adjudicated a large permitted stablecoin issuer failure, so the speed and completeness of recovery is an assumption rather than a precedent.

The deposit token offers a better-documented recovery framework with an unresolved question about whether a transferred token holder sits inside it, while the stablecoin offers a structurally simpler claim whose enforcement in a real insolvency is untested. Neither instrument eliminates counterparty risk. The deposit token concentrates exposure in one bank’s credit; the stablecoin concentrates it in one issuer’s operational and legal integrity.

Retail reach and DeFi collateral

Deposit tokens are structurally disadvantaged in both consumer payments and protocol collateral, and the reason in both cases is the transfer restriction that makes them safe for the bank. Reporting that Block began a phased USDC rollout across Cash App’s roughly 60 million users, which we could not verify from an accessible source, would describe a distribution event rather than a technical one. Consumer stablecoin adoption at that scale runs through an app’s existing KYC relationship, and the token’s value to the consumer is that it can be sent to anyone the app can reach. A deposit token that requires the recipient to be an onboarded client of the issuing bank cannot replicate that, and JPMorgan has given no indication it wants to. The November 2025 announcement is explicitly institutional.

The DeFi collateral case is harder for deposit tokens than retail payments. A lending protocol accepting collateral needs three properties: the asset can be seized and transferred to an arbitrary liquidator without permission, its value can be marked continuously, and its solvency is not correlated with the conditions that trigger liquidations. A whitelisted deposit token fails the first outright. If only approved addresses can receive the token, a liquidation engine cannot guarantee delivery to whoever shows up with a bid, and a protocol that cannot liquidate cannot lend safely against the asset. A freeze capability, which a bank issuing its own deposit liability will almost certainly retain, makes the problem worse, because collateral can be immobilized in the middle of an auction.

The third property is also weaker for deposit tokens than for fully reserved instruments. A claim on a fractional-reserve bank is a claim on a credit portfolio, and bank credit stress is correlated with the broad risk-off conditions that produce mass liquidations. A fully reserved stablecoin’s backing is short government paper, which is the asset that tends to appreciate in those conditions. For trust-minimized protocols, the full-reserve model is the better collateral precisely because it strips out the bank counterparty.

This cuts against any reading in which deposit tokens straightforwardly displace stablecoins. The reintermediation thesis, that banks are using public chains to reinsert themselves where crypto removed them, only holds where they can actually operate, and the permissioning that makes a deposit token acceptable to a prudential regulator is the same property that keeps it out of permissionless collateral markets.

What would settle the comparison

The reintermediation question is weaker than it is usually posed, because corporate treasury was never disintermediated. Large firms kept their operating cash at banks throughout the stablecoin era. If JPMD and its peers mainly tokenize relationships that never left bank custody, they extend bank intermediation into new settlement hours rather than reclaim ground. The claim becomes interesting only if deposit tokens start displacing stablecoins in places stablecoins currently own, and the structural obstacles above suggest that requires either open transferability the banks have not offered or an interbank layer that does not yet exist publicly.

Several observable conditions would move the analysis. The first is whether JPMD balances appear in the hands of entities that are not JPMorgan clients. A deposit token that circulates only among onboarded institutions is a cash management product with better hours. A deposit token that a non-client can receive and hold raises immediate questions about deposit status and insurance determination that JPMorgan would have to answer.

The second is disclosure of interest terms. If JPMD pays a competitive rate, the liquidity treatment of the underlying balances becomes the thing to watch, because paying for funding that can exit in seconds is an expensive position for a bank to hold at scale. If it pays nothing, the most-cited advantage of deposit tokens over GENIUS stablecoins disappears and the comparison collapses into settlement convenience versus transferability.

The third is how prudential regulators classify tokenized deposits relative to nonbank stablecoin issuance. Reporting places FDIC action on GENIUS prudential standards in April 2026; we could not access a source to confirm it or to read what the standards say about bank-issued deposit tokens. Whichever instrument carries the lighter capital and liquidity charge for an equivalent dollar of liability will attract issuance, and the two regimes were not designed against each other.

The fourth is interbank mobility. Until a JPMD balance can discharge an obligation to a client of a different bank without routing through conventional correspondent rails, the deposit token is a closed loop with a public-chain interface, and the question of whether it competes with stablecoins does not really arise outside JPMorgan’s own client network.

References

• JPM Coin (JPMD) USD Deposit Token Available for Institutional Clients, J.P. Morgan