September 25, 2026:


IBM took a significant step toward making blockchain-based money movement routine for regulated banks on September 24, announcing in beta the IBM Digital Asset Haven ISO 20022 Messaging Adapter — a piece of middleware that lets financial institutions instruct tokenized deposit transactions on Swift’s permissioned blockchain ledger using the same payment message format they already use for conventional cross-border transfers. The adapter eliminates the single largest obstacle to bank participation in the Swift blockchain pilot: the need to build entirely new operational systems to communicate with a distributed ledger. In a separate announcement, IBM simultaneously released a beta on-premises deployment of Digital Asset Haven, allowing regulated institutions to run the full platform inside their own data centers on IBM Z and IBM LinuxONE hardware, with no dependency on public cloud infrastructure.
ISO 20022 became the mandatory standard for all cross-border payment instructions on Swift’s network on November 22, 2025, when the coexistence period with legacy MT-format messages ended. Every one of the 12,500 financial institutions connected to Swift’s network across more than 200 markets now sends and receives payment instructions in this XML-based format.
IBM’s new adapter sits between a bank’s existing payment application and the Digital Asset Haven platform’s blockchain layer. When a bank’s payment system transmits a standard ISO 20022 credit transfer instruction, the adapter translates it into the corresponding on-chain transaction instruction for Swift’s shared ledger — constructing and signing the smart contract call that moves the tokenized deposit representation. Banks in the pilot can move tokenized deposits around the clock, seven days a week, ahead of final settlement, which continues to flow through existing channels such as real-time gross settlement systems and correspondent banking arrangements.
The practical consequence for a bank’s operations team is meaningful: compliance workflows, KYC checks, and sanctions screening remain anchored to the existing banking system. No staff retraining on blockchain-native tooling is required. The ISO 20022 format carries enough structured data to accommodate the additional fields that tokenized deposit instructions require, which is one reason the migration to ISO 20022 — mandated years in advance precisely because of its extensibility — now doubles as the infrastructure layer for on-chain money movement.
Swift’s shared ledger is built on Hyperledger Besu, an open-source, Ethereum Virtual Machine-compatible permissioned blockchain client maintained by the Linux Foundation’s Hyperledger project. Unlike public Ethereum, which uses proof-of-stake consensus with thousands of anonymous validators, Hyperledger Besu in a consortium configuration runs proof-of-authority consensus — meaning a known, pre-approved set of validator nodes, in this case the pilot banks — confirms transactions deterministically, without mining and without the gas fee volatility of public chains.
What this architecture means for the IBM integration: when a bank’s payment application sends an ISO 20022 message through IBM’s adapter, that message is ultimately converted into a Solidity smart contract call on the Besu ledger built by Consensys. The ledger’s logic — which governs which banks can participate, how tokenized deposits are moved, and when interbank liabilities are recorded — runs as Ethereum-compatible smart contracts that the pilot institutions agreed to in the network’s design phase.
This detail matters beyond workflow convenience. Swift’s ledger is not simply a new database with a faster API. It is an EVM-compatible programmable layer inside the regulated global banking perimeter — meaning the tokenized deposits moving on it can, in principle, execute conditional settlement logic: payment on delivery against a documented shipment, real-time margin calls tied to live position data, cross-border escrow that releases automatically on a specified date. Swift has explicitly named programmable money and agentic commerce as intended future applications of the ledger. That capability was not available in any form on the legacy Swift MT or ISO 20022 messaging infrastructure, which carries instructions to human-operated settlement systems rather than executing logic autonomously.
IBM Digital Asset Haven provides the wallet infrastructure, transaction signing capability, policy-based governance tooling, and the Hyperledger Besu connectivity that allow a bank to participate in the pilot without running its own blockchain node. Key management is handled by IBM Crypto Express Hardware Security Modules embedded in the LinuxONE hardware — certified to FIPS 140-2 Level 4, the highest commercial certification for cryptographic devices. Multi-party computation enables institutional approval rules (such as requiring two officers to authorize transactions above a specified threshold) to be enforced before any signing operation reaches the ledger.
Swift first announced the shared ledger at its Sibos 2025 conference in Frankfurt, developed the initial prototype in partnership with Consensys, and moved from concept to activation in less than nine months — a timeline that reflects the unusual urgency major financial institutions brought to the project’s design phase, with more than 40 institutions involved.
As of September 25, 2026, seventeen first-mover institutions are piloting tokenized deposit transactions on the network. By the time IBM announced its adapter, the ledger had already recorded live cross-border transactions. Standard Chartered and HSBC completed the first live interbank tokenized deposit transaction on the ledger in August 2026. First Abu Dhabi Bank and Citi completed live dollar transactions at scale on the ledger in September 2026, demonstrating that tokenized commercial bank money can support cross-border payments from the Middle East and Africa. IBM said financial institutions already enrolled in the Digital Asset Haven program have successfully tested tokenized deposits on the Swift infrastructure, though the company has not named the specific participating banks.
The seventeen pilot institutions include Citi, HSBC, UBS, BNP Paribas, Standard Chartered, Wells Fargo, BNY, DBS, and MUFG Bank, among other global lenders.
The second beta announcement — the on-premises deployment of Digital Asset Haven — addresses a separate concern from the Swift adapter’s workflow problem: where the digital asset infrastructure itself physically operates.
The on-premises configuration runs entirely inside a client’s own data center on IBM Z and IBM LinuxONE without requiring any connection to public cloud infrastructure. Clients can deploy it on compatible IBM equipment already in their environment or add new capacity. The architecture keeps both the solution layer and the key management layer entirely within the client’s own environment — Crypto Express HSMs are embedded directly in the LinuxONE hardware, and confidential computing with secured environment partitioning isolates production, test, and development workloads from each other.
IBM’s Offline Signing Orchestrator supports formal key-ceremony processes for generating root certificate authority keys, producing auditable documentation that clients can present to regulators — a requirement in several jurisdictions that govern how private keys for digital asset operations must be generated and controlled. IBM states that configurations using IBM LinuxONE Rockhopper 5 with the required resiliency stack — including GDPS, IBM DS8000 series storage with HyperSwap, and Red Hat OpenShift in a Single System Image — are designed to achieve 99.999999% availability. That figure is based on IBM’s internal measurements and projections and requires enabling all specified resiliency components; application-induced outages are not included.
The on-premises option carries the same APIs and workflows as Digital Asset Haven’s existing SaaS and hybrid versions. Clients can switch between deployment models without rewriting their integrations.
IBM’s adapter introduces a dependency that a native blockchain integration would not: banks must trust IBM’s middleware to correctly translate their ISO 20022 payment intent into on-chain instructions. A bank that sends an ISO 20022 message through the adapter never directly crafts or inspects the smart contract call that reaches the Besu ledger — IBM’s layer does that translation. For institutions with the deepest regulatory aversion to third-party software dependencies in payment-critical systems, that intermediary role may warrant scrutiny before committing to the adapter in production.
The adapter and on-premises option are both in beta. IBM has not set a timeline for moving either to general availability. Transaction volumes on the Swift ledger are not publicly disclosed by IBM or by Swift. IBM cited a J.P. Morgan Payments finding that 93% of financial institutions are currently modernizing their payment infrastructure — a figure that suggests demand for what the adapter enables, though how many of those institutions will specifically adopt the IBM path to the Swift ledger versus alternatives remains to be seen.
For most corporate treasurers and enterprise payments professionals, the practical question is whether joining the Swift blockchain pilot — through IBM’s adapter or any other middleware — is worth prioritizing ahead of general availability.
The strategic case is straightforward: the alternative to settling through Swift’s blockchain ledger is the incumbent correspondent banking system, which operates within limited hours and carries settlement risk that grows during weekends and across time zones. Tokenized deposits, because they remain on the issuing bank’s balance sheet as regulated deposit liabilities, carry FDIC insurance coverage up to $250,000 per depositor — unlike most stablecoins, which are liabilities of non-bank entities and are not insured.
The practical case for the IBM route specifically rests on the workflow-reconstruction argument: institutions whose payment operations already run on ISO 20022 infrastructure — which, since November 2025, means every Swift-connected institution — can connect to the ledger without building parallel operational processes. The cost is the IBM middleware dependency, the beta status of the adapter, and the absence of public transaction volume data to anchor a business case.
Whether programmable money — smart contracts that execute settlement logic automatically rather than waiting for human approval at each step — becomes the dominant reason to join the pilot may determine which institutions treat this as infrastructure, and which treat it as an experiment worth watching from a distance.
A tokenized deposit is a commercial bank deposit represented as a digital token on a blockchain. The token is a direct, one-to-one claim on funds held at a regulated bank, which remain on the bank’s balance sheet as a deposit liability — subject to the same capital requirements, supervisory oversight, and deposit insurance that govern all conventional deposits. A stablecoin, by contrast, is a liability of a non-bank issuer (or, under the US GENIUS Act framework, of a bank acting as a payment stablecoin issuer) that is backed by reserves rather than balance-sheet deposits. The European Banking Authority confirmed in December 2024 that tokenization does not alter the fundamental nature of a deposit under existing regulatory definitions.
A permissioned blockchain restricts which nodes can join the network and validate transactions. Swift’s ledger runs on Hyperledger Besu — an Ethereum-compatible client — using proof-of-authority consensus, in which a known set of validator nodes (the participating pilot banks) confirms transactions. This means no public mining, no anonymous participants, and deterministic transaction finality without the gas fee volatility of public Ethereum. Banks retain authority over their own assets and private keys; Swift operates the coordination layer that records interbank payment commitments, not a custodial settlement system. The permissioned design allows existing compliance, KYC, and sanctions-screening frameworks to operate normally, which is why it gained support from more than 40 financial institutions in its design phase.
Swift’s Hyperledger Besu ledger executes Ethereum-compatible smart contracts, which means the payment logic that governs tokenized deposit movement on the ledger can be conditional and programmable — not simply instructional. Conventional ISO 20022 payment messages carry instructions to human-operated settlement systems; they do not execute logic autonomously. On the Besu ledger, a smart contract can release a payment automatically when a defined condition is met — delivery of goods, expiration of a time window, a change in a reference rate. Swift has named programmable money and agentic commerce as intended future applications. IBM’s adapter is the first enterprise middleware layer that routes bank payment messages directly to this programmable infrastructure inside the regulated global banking perimeter.
IBM Digital Asset Haven was launched in October 2025 as a SaaS and hybrid cloud product. The new on-premises beta runs entirely inside a client’s own data center on IBM Z or IBM LinuxONE hardware — no public cloud connection required. Both the platform layer and the cryptographic key management layer stay within the client’s environment, with IBM Crypto Express HSMs handling key protection at FIPS 140-2 Level 4 certification. This matters for central banks, national institutions, and commercial lenders operating in jurisdictions with strict data-residency or operational-sovereignty rules that make third-party cloud deployment a regulatory or reputational obstacle.