All articles
Tokenized DepositsStablecoinsBlockchain PaymentsBanking Technology

Tokenized deposits vs stablecoins in 2026: how banks are putting money on a blockchain

Between September and late October 2026, banks completed live tokenized deposit payments on Swift's ledger, OCBC, VersaBank and SoFi launched tokenized deposit products, the OCC finalised its GENIUS Act stablecoin rule and an EY survey of 521 decision-makers found the market moving to implementation. What tokenized deposits are, how they differ from stablecoins, and what a bank technology leader should build now.

By Yassineh Pourebrahim · · 10 min read

For years, "money on a blockchain" meant stablecoins, issued outside the banking system. September and October 2026 tell a different story. In September, Citi, First Abu Dhabi Bank, OCBC, DBS, UOB and ANZ completed live payments with tokenized deposits on Swift's blockchain-based ledger. In October, OCBC launched a tokenized deposit solution for Ant International's cross-border treasury, VersaBank started the commercial phase of its tokenized deposit programme in the United States, and SoFi launched a tokenized deposit for consumers. In the same month the US Office of the Comptroller of the Currency (OCC) finalised its rule implementing the GENIUS Act for payment stablecoins, and EY-Parthenon published a survey showing that both corporates and banks are moving from experimentation to implementation.

This article explains what tokenized deposits are, how they differ from stablecoins, what is actually live, and how I would approach the topic as a technology leader in a bank or fintech.

Tokenized deposits vs stablecoins: the difference that matters

Both instruments are digital dollars (or euros, or Singapore dollars) recorded on a distributed ledger. The difference is whose liability they are.

  • A tokenized deposit is a normal bank deposit represented on a ledger. DBS describes tokenised deposits as "regular bank deposits in digital form", issued by regulated banks. OCBC calls them "digital representations of commercial bank deposits on a distributed ledger". They sit on the issuing bank's balance sheet, alongside its other deposits.
  • A payment stablecoin is a token issued by a separate issuer and backed by a reserve of assets. Under the US framework, only a permitted payment stablecoin issuer may issue one in the United States.

US regulators are drawing this line explicitly. In its final GENIUS Act rule, the OCC writes that the definition of "payment stablecoin" distinguishes a payment stablecoin "from a deposit, including a tokenized deposit". The FDIC's April 2026 proposal goes further on deposit insurance: it would provide that deposit insurance "does not depend upon the technology or recordkeeping" used to record a bank's deposit liabilities, while deposits held as reserves backing a payment stablecoin would not be insured to stablecoin holders on a pass-through basis.

The product launches reflect the same distinction. SoFi's disclosure states that SOFIE "is a tokenized deposit representing a deposit liability of SoFi Bank" and "is not a payment stablecoin". It pays a variable 4.40% APY (as of 20 October 2026) and is FDIC-insured up to the applicable limits. VersaBank describes its VUSD tokenized deposits as interest-paying and FDIC insurance-eligible.

Why central banks care: singleness, elasticity, integrity

The Bank for International Settlements set out the economic argument in its 2025 Annual Economic Report. It proposes three tests for money: singleness (every payment settles at par, so a dollar from one bank is worth a dollar from another), elasticity (the system can supply liquidity for large payments without gridlock) and integrity (protection against financial crime). The BIS concludes that stablecoins "fall short" on all three: as bearer instruments they can trade away from par, any new supply must be fully paid up front, and tokens can circulate to wallets whose holders have not been identified. Tokenised commercial bank money, by contrast, keeps the two-tier system: payments between customers of different banks still settle in central bank reserves, and the BIS notes that "no new credit exposures are created across institutions".

From a bank architecture point of view, that is the key insight. A tokenized deposit is not a new kind of money; it is a new way of recording and moving the money a bank already issues.

What went live in September and October 2026

Swift's ledger moves from design to live transactions. According to DBS, Swift announced in July 2026 that its blockchain-based ledger was ready for initial use, with 17 banks across six continents preparing to transact with tokenised deposits. On 2 September Citi announced live transactions with FAB and OCBC, as part of a controlled proof-of-concept phase running from July to December 2026. On 10 September DBS, OCBC and UOB completed the first live domestic Singapore dollar interbank transactions using tokenised deposits on the ledger. On 30 September ANZ completed a US dollar corporate treasury payment between Melbourne and New York with BHP, Swift and Citi.

The design is deliberately conservative. FAB explains that the tokenized deposits remained on the participating banks' balance sheets, that the Swift ledger coordinated payment commitments and recorded interbank liabilities "without taking custody of funds", and that interbank settlement was completed through established correspondent banking channels. Citi describes the same model: instant payment commitment through tokenized deposits, with existing settlement models, including real-time gross settlement (RTGS), used for final settlement. In the Singapore transactions, DBS says the ledger acted as an orchestration layer where obligations were matched and netted before final settlement through existing systems.

Corporate treasury is the first real use case. In the ANZ transaction, BHP used its existing ANZ account and banking channel while tokenised deposits and ledger technology "operated behind the scenes"; ANZ described the payments as "completely invisible to customers". OCBC's 7 October launch with Ant International targets the same need: SGD and USD tokenised deposits that Ant International can move between its Singapore and Malaysian entities 24/7, through its blockchain-based WhaleRTP platform, instead of waiting for batch cycles and cut-off times.

Retail and US commercial launches. VersaBank announced on 15 October that it had started the commercial phase of its tokenized deposit programme in the United States, beginning with its own VUSD tokens, and that its technology can "freeze, seize and burn" tokens that regulators determine are being used illegally. On 22 October SoFi launched SOFIE, which customers can convert 1:1 into SoFiUSD, the stablecoin SoFi Bank launched in December 2025, to buy crypto or move funds on blockchain rails 24/7. SoFi's model is a useful signal: a single bank offering a deposit token for holding and a stablecoin for moving.

The regulatory picture in the United States

The GENIUS Act was enacted on 18 July 2025 and takes effect on 18 January 2027. The OCC's final rule, announced on 21 October 2026 and published in the Federal Register on 23 October, takes effect on the same date. It creates a new 12 CFR Part 15 covering reserve assets, redemption, risk management, custody, capital and operational backstops for issuers under OCC jurisdiction. Rules on the Bank Secrecy Act, anti-money laundering and sanctions will come in a separate rulemaking with the Treasury. The FDIC published its own proposal for the institutions it supervises in April 2026, with a comment period that closed on 9 June; among other things, it would generally require stablecoin issuers to redeem within two business days.

For a bank, the practical consequence is that there are now two clearly separated regulatory routes for on-chain dollars: a deposit that happens to be tokenized, and a stablecoin issued under a dedicated regime, often through a subsidiary.

What the market says: multi-rail, embedded, partnered

EY-Parthenon's October 2026 survey of 521 decision-makers (383 corporates and 138 financial institutions, surveyed in July and August) gives the clearest view of demand:

  • Adoption is real but early. 18% of corporates have used stablecoins and 15% have used tokenized deposits. Among financial institutions, 69% already offer or plan to launch stablecoin services within two years, and 49% of banks offer or plan tokenized deposits over the same period.
  • The future is multi-rail. Among respondents interested in blockchain payments, 45% of corporates and 53% of financial institutions favour both stablecoins and tokenized deposits. Stablecoins are still the preferred standalone choice, which EY reads as a sign that they are currently seen as the more mature instrument.
  • Integration is the new bottleneck. Regulatory uncertainty is still the top barrier (76%), but limited ecosystem adoption (65%) and lack of ERP/TMS integration (54%) now rank close behind. 74% of corporates prefer blockchain payments embedded in existing interfaces, and 63% say they are likely to adopt them through their current bank or payments provider.
  • Banks will partner rather than build. Only 9% of financial institutions offering or planning stablecoin services plan to build entirely in-house. Competitive differentiation is now the top driver (66%, up from 52%), ahead of revenue generation (41%, down from 59%).
  • Programmable payments are next. 54% of financial institutions already offer programmable payments or plan to within two years.

A playbook for bank technology leaders

Here is how I would translate these signals into a technology plan.

  1. Choose your role before choosing a chain. EY frames the options as issuer, enabler, servicer, integrator or a combination. A bank with large corporate treasury clients may start as a tokenized deposit issuer on a shared network; a smaller bank may integrate someone else's rails. The role determines the architecture, the licences and the partners.
  2. Keep the core ledger as the book of record. In every live case above, tokens stay on the bank's balance sheet and final settlement uses existing channels. Treat the token ledger as an extension of the core banking ledger, with real-time, automated reconciliation between the two. This is fundamentally a data problem: the same discipline of one trusted, reconciled view that I described for a unified data warehouse for brokers applies to balances that now exist in two systems.
  3. Make it invisible to the client. The ANZ–BHP design and EY's 74% figure point the same way: corporates want the benefits in their existing accounts, channels, ERP and treasury systems. Budget for APIs, ISO 20022 mapping and treasury-system connectors, not only for the ledger.
  4. Design for 24/7 operations. Always-on payment commitments are the whole point, but most banks' liquidity management, fraud operations, support and change windows still assume business hours. Weekend commitments need limits, monitoring and on-call ownership before the first live client.
  5. Build compliance into the token. The BIS integrity test, VersaBank's freeze capability and the OCC's separate AML rulemaking all lead to the same requirement: identified holders, the ability to stop or reverse flows when required, and screening that runs at payment time rather than after the fact.
  6. Abstract the rail. With most interested institutions planning for both stablecoins and tokenized deposits, a payments orchestration layer that can route across RTGS, card, instant payments, tokenized deposits and stablecoins protects you from betting on a single network.
  7. Treat keys as critical infrastructure. Token transfers are authorised by digital signatures, so key custody, HSM integration and crypto-agility are core requirements. In Post-quantum cryptography in 2026 I explain why the signature layer is exactly where banks need to plan their quantum-safe migration first.
  8. Prepare for programmable money and machine customers. Conditional payments, automatic sweeps and agent-initiated purchases all need money that software can move within strict, auditable limits. The trust stack I described in Agentic payments in 2026 (agent identity, signed intent and scoped credentials) is a natural fit for tokenized money.

Closing thought

The debate used to be framed as banks versus stablecoins. In 2026 the more useful framing is which liability you hold and which rail moves it. Tokenized deposits let banks offer 24/7, programmable movement of the money they already issue, inside existing deposit insurance, regulation and settlement. Stablecoins remain the more mature standalone instrument and will keep their role. For most banks the winning position will not be a single token, but the ability to move clients' money across both rails without the client noticing, with the core ledger, compliance and keys under control.

References