Swift carries the messaging that moves global value. More than 11,000 financial institutions across more than 200 countries operate on the network. In 2023, Swift and the Chainlink network ran public experiments on whether Swift messages can orchestrate tokenized asset transfers across multiple blockchains using the Chainlink CCIP Cross-Chain Interoperability Protocol. The 2023 experiments were designed with more than a dozen institutions but simulated entirely by Swift and Chainlink. In January 2026, Swift completed a separate tokenized bond milestone with three institutions, and Chainlink’s account of it does not say CCIP carried it. This article covers how the tested design works, what was actually tested and what a production stack would look like underneath.
This article is the infrastructure article. It assumes the reader is a bank architect, a payment systems strategist, a tokenized asset platform engineer or an institutional buyer trying to understand what is real versus what is press-release framing. It covers the partnership scope, the timeline, the named participants, the technical stack, the architectural rationale, the open questions, plus the operator role beneath the headlines.
Matrixed.Link is a Chainlink node operator. The view in this article is the operator view.
What the Swift and Chainlink CCIP Partnership Actually Is
Swift and Chainlink CCIP is not a payments network replacement. In the design the two organizations tested, banks keep sending Swift messages (see blockchain for banks for the wider picture), and CCIP carries out the resulting token movements on the blockchains where the assets live.
The Swift plus Chainlink CCIP work is an interoperability layer. Swift remains the messaging network that 11,000+ financial institutions rely on to instruct value movement. CCIP is the cross-chain protocol that lets those instructions reach blockchain networks where tokenized assets actually live.
The mechanical pattern the experiments tested is this. A Swift-connected bank uses its existing Swift infrastructure to send a standardized message instructing the movement of a tokenized asset. The message gets bridged to an on-chain instruction via the CCIP infrastructure. The on-chain instruction settles across one or more blockchains where the tokenized asset is held. The result returns through CCIP back to Swift, providing the originating bank with end-to-end confirmation.
Three properties matter for an institutional buyer.
First, the bank keeps its existing Swift integration. No new connectivity standard. No new messaging format that the bank’s existing software cannot read. Swift’s results report says the approach “minimises the need for substantial additional investments”.
Second, the cross-chain step is verified onchain rather than by a single bridge operator. The 2023 experiments used the v1-era CCIP design, and the results report’s own diagram shows an Executing DON. CCIP 2.0 came in September 2026 and verifies each message with a default Committee Verifier network of 16 independent node operators, and the destination chain checks the signed attestations onchain before it releases tokens or delivers data. The CCIP article covers the difference.
Third, the architecture is vendor-neutral at the chain layer. The design goal is that a bank can reach different chains without negotiating bilateral integrations with each network. The 2023 experiments covered Ethereum, Avalanche and a private Quorum network.
The next sections cover how the work developed.
The Timeline: 2022 to 2026
The Swift plus CCIP work has been public across several phases.
2022. At SmartCon 2022, Swift and Chainlink announced work on using CCIP with Swift messaging, per Chainlink’s history of the partnership.
2023 (June). Swift published the blockchain interoperability experiment announcement. As Cointelegraph reported it, the first use case was the transfer of tokenized assets between two wallets on the same public blockchain network, the Ethereum Sepolia testnet. The question was specific: can a Swift messaging instruction trigger tokenized asset settlement across multiple blockchains without requiring banks to operate distinct connectivity to each chain.
2023 (August). Swift released the press communication on the results of the experiments. Sergey Nazarov, co-founder of Chainlink, commented: “When combining Swift and CCIP, we were able to show that this new level of interoperability across various blockchains is now possible with minimal resources from even the largest banks and market infrastructures.” (The Fintech Times). The participant list included ANZ, BNP Paribas, BNY Mellon, Citi, Clearstream, Euroclear, Lloyds Banking Group, SIX Digital Exchange and DTCC. Swift’s results report adds an important qualification: “Although the solution was designed jointly with participants, the technical development work was simulated entirely by Swift and Chainlink.”
2024 (Sibos). Per Chainlink, Sergey Nazarov introduced a pre-production solution that enables banks to connect to blockchains using the Swift messaging standards.
2025 (September). Chainlink published a history of the Swift partnership on 29 September 2025, updated on 4 August 2026.
2026 (January). As part of Swift’s work with Chainlink and UBS Asset Management, Swift completed an interoperability milestone for tokenized bonds with BNP Paribas Securities Services, Intesa Sanpaolo and Société Générale. Chainlink’s account does not say that CCIP carried this exchange, so this article does not say it did.
2026 (June). Chainlink and partners launched Project Pangea. Chainlink says it brings together more than 50 banks representing over $10 trillion in assets under management, with CCIP providing connectivity and CRE serving as the orchestration layer between Swift and blockchains. Chainlink says the project was launched to develop a novel solution for international FX markets. No go-live date has been published.
2026 (September). Chainlink’s CCIP 2.0 post says the Swift work and DTCC’s corporate actions work are already “leveraging CCIP”, and its history of the Swift partnership calls phase 2 of the corporate actions work “a production-grade system”. Those are Chainlink’s statements. Swift’s 31 August 2023 press release, cited above, covers the 2023 experiments only.
What the 2023 experiments showed: in a simulated environment, a tokenized asset can move between wallets on one public chain, between two public chains and from a public chain to a private one. What remains open: the legal, regulatory and operational framework for production volume.
The Participants and What Was Tested
The 2023 experiments named specific institutional participants. Per Swift’s results report, the participant list included:
- ANZ
- BNP Paribas
- BNY Mellon
- Citi
- Clearstream
- Euroclear
- Lloyds Banking Group
- SIX Digital Exchange
- DTCC
This is a list of established banks and market infrastructures, not startups. Their role was to take part in the experiments and co-design the solution. The technical development was simulated entirely by Swift and Chainlink, so the experiments show that the pattern is feasible. They do not show that these institutions run it in production.
What the 2023 experiments covered. Swift’s results report describes three transfers of a tokenized asset:
- Between two wallets on the same public network (Ethereum)
- Between two wallets on different public networks (Ethereum and Avalanche)
- From a public network (Ethereum) to a private network (Quorum)
The chains involved:
- Ethereum (the Sepolia testnet: Swift’s press release names “the Ethereum Sepolia network”; the results report says only Ethereum)
- Avalanche
- A private Quorum network
The later bond milestone. In January 2026, tokenized bonds were the asset class of Swift’s interoperability milestone with BNP Paribas Securities Services, Intesa Sanpaolo and Société Générale. The sources we checked do not name the chains involved or say that CCIP carried it.
The Technical Stack: Swift Messaging Meets On-Chain Settlement
The design has three sides.
The Swift side. The participating bank uses its existing Swift connectivity, a settlement instruction in the MT 543 format (ISO 15022), plus its established compliance and reconciliation procedures. From the bank’s existing core systems perspective, the transfer looks like a standard Swift instruction. The bank’s internal controls, audit, plus regulatory reporting all flow through familiar paths. This is the entire point: institutional adoption costs drop when the integration meets banks where they already operate.
The CCIP side. The 2023 experiments used the v1-era CCIP design. Where this article mentions CCIP 2.0, it describes CCIP as it is since September 2026, not the design that was tested. Chainlink CCIP provides three capabilities. Token transfers across supported chains. Arbitrary data transmission to a contract on another chain. Programmable token transfers, which send tokens together with instructions for the receiving contract. CCIP 2.0 adds optional compliance hooks on token pools through the Chainlink Automated Compliance Engine. CCIP 2.0 allows one token transfer per message. The CCIP architecture is documented in the Chainlink CCIP overview.
The connection layer. This is where, in the tested design, the Swift instruction becomes an on-chain transaction through CCIP and where the settlement result is reported back to the bank. The appendix of the 2023 results report maps the instruction to a Swift MT 543 settlement message and maps the status codes that come back to MT 548. It notes that ISO 20022 is also possible. As one example of a field, the quantity of tokens goes in the :36D::SETT field of the MT 543 and the wallet addresses go in :97D::BCAW. The report also says a few new field options should be created.
The oracle layer underneath. Every cross-chain operation depends on data that does not live on any single ledger. FX rates for cross-currency tokenized transfers. Settlement asset price feeds. Reserve attestations for tokenized commercial bank deposits or stablecoins used as settlement assets. Compliance data feeds (sanctions lists, KYC attestations) where required. Oracle networks deliver this data on-chain with cryptographic verification. The Chainlink Data Streams plus Chainlink Data Provider primers cover this layer in detail.
Risk controls. This describes CCIP 2.0, which came after the 2023 experiments, not the design they tested. CCIP 2.0 verifies every message with a default Committee Verifier network of 16 independent node operators, and a lane, token pool or receiver can require additional independent verifiers on top of it. The destination chain checks every required attestation onchain before it releases tokens or delivers data. An onchain RMN contract can mark a chain or lane as cursed, which makes sends and executions involving it revert, and token pools enforce per-chain rate limits.
Why CCIP, Not a Bilateral Bridge or a Stablecoin Rail
A bank architect evaluating cross-chain settlement infrastructure has three structural options. Each has different risk plus integration characteristics.
Option A: bilateral bridges. Banks build (or contract for) direct integrations between specific blockchains they need to transact across. Bilateral bridges do not scale. N participating chains means roughly N-squared integrations to maintain. For an institution that needs to transact across five or ten chains, the operational burden compounds.
Option B: a single stablecoin rail. Use a major stablecoin (USDC, USDT or similar) as the settlement asset across all chains. The bank holds the stablecoin on whichever chain is needed. This pattern works for some use cases but has institutional limitations: the bank is exposed to the stablecoin issuer’s reserve management plus regulatory status. The settlement asset is not central bank money or tokenized commercial bank deposit. For wholesale settlement at bank scale, the stablecoin rail is often a supplementary path, not the primary path.
Option C: a vendor-neutral interoperability protocol. Use a protocol like CCIP that provides token transfer plus messaging across an expanding set of chains with onchain verification of signed attestations and optional additional verifiers. The bank does not pick a single chain or a single settlement asset. The architecture supports multiple settlement assets (tokenized commercial bank deposits, tokenized treasuries, stablecoins, eventually CBDC) on multiple chains.
Option C is the pattern the Swift and Chainlink experiments explored. In our view, cross-chain interoperability at institutional scale needs onchain verification, the option to require additional verifiers and support for many chains. CCIP 2.0 offers those as protocol features rather than bilateral commitments.
The What Is a Blockchain Oracle? explainer covers the underlying cryptographic primitives that make this verification possible at protocol scale.
What This Enables: Tokenized Assets at Bank Scale
The Swift plus CCIP design is not built for crypto-native users. It targets the institutional tokenized asset stack. The asset classes in play as of 2026:
Tokenized bonds. Swift’s January 2026 interoperability milestone with BNP Paribas Securities Services, Intesa Sanpaolo and Société Générale covered tokenized bonds. The sources we checked do not say that CCIP carried it.
Tokenized funds. Funds (money market funds, mutual funds, ETF-equivalents) are one of the asset classes institutions are tokenizing. The Tokenized Money Market Funds article covers the broader TMMF infrastructure landscape.
Tokenized commercial bank deposits. The BIS Project Agorá initiative brings together eight central banks and over 40 financial institutions, and explores tokenized commercial bank deposits combined with tokenized central bank reserves. The Swift plus CCIP architecture is an adjacency to this work. Banks that issue tokenized deposits need cross-chain infrastructure that meets the same institutional grade as the underlying assets.
Cross-border settlement. Settlement across jurisdictions requires FX conversion, multi-party reconciliation, plus regulatory compliance per leg. CCIP plus Swift target the technical workflow. The Central Bank Digital Currency Infrastructure article covers how CBDC-side rails fit into this architecture.
What a bank’s deployment pipeline would look like in 2026:
- Swift connectivity (already in place at the participating bank)
- CCIP integration via a participating institutional infrastructure provider
- Custody arrangement for tokenized assets (see Digital Asset Custody)
- Oracle layer integration for FX, market data, plus compliance data
- Internal compliance plus risk framework adapted for cross-chain transactions
- Regulatory engagement on settlement finality under applicable jurisdictional rules
Open Questions and Unsolved Problems
The 2023 experiments addressed the engineering questions in a simulation. The forward questions through 2026 are these.
Production volume scaling. The 2023 experiments were simulated by Swift and Chainlink, so they show feasibility, not volume. Production volume across thousands of banks plus billions of dollars of daily tokenized asset flow is a different operational profile. The deployment cadence depends on bank-side readiness plus regulatory approval per jurisdiction.
Settlement finality under FMI rules. Financial market infrastructure (FMI) regulations define when a transaction is legally final. Probabilistic finality on a public chain is not automatically equivalent to FMI-grade settlement finality. CCIP’s default is to wait for full finality, and CCIP 2.0 lets senders and token issuers opt in to faster-than-finality transfers. Whether a given finality rule satisfies FMI rules is a legal question that varies by jurisdiction.
Operational governance. Who runs the verifiers that attest to each cross-chain message. Who attests to the on-chain state. Who indemnifies counterparties if a CCIP transfer fails. Who operates the connection between Swift and CCIP. In the 2023 experiments, Swift and Chainlink carried out the technical work themselves. A production deployment has to settle who carries the operational responsibility.
Cross-jurisdictional compliance. Different chains have different KYC plus AML enforcement profiles. A transfer that originates in a jurisdiction with strict transaction-level compliance plus terminates in a jurisdiction with lighter requirements creates compliance complexity. The infrastructure can carry compliance metadata. The legal frameworks that govern the metadata vary per jurisdiction.
Settlement asset choice. A tokenized cross-chain transfer settles in some asset. The choices include tokenized commercial bank deposits, regulated stablecoins, tokenized treasuries, plus eventually wholesale CBDC. Each settlement asset has different regulatory plus operational properties. Banks deploying a Swift plus CCIP stack would make the settlement asset choice per use case.
These are open questions about deployment as much as engineering. The experiments show the pattern is feasible. The institutional layers governing how it would be used in production are still catching up.
Where oracle node operators fit in this stack
In this design, two groups of independent node operators sit on the blockchain side. CCIP itself is secured by a default Committee Verifier network of 16 independent node operators that attest to each cross-chain message. Underneath it, Chainlink oracle networks run by node operators deliver the data no single ledger holds: FX rates for cross-currency transfers, prices for the settlement assets and reserve attestations for tokenized deposits or stablecoins.
That puts operators in the operational governance questions above. When a bank or its infrastructure provider assesses this layer, the useful checks are the same as for any critical vendor: an on-chain performance history anyone can verify, an audited security management system and a record of how the operator handled network upgrades and incidents. The What Is a Chainlink Node Operator? primer covers the operator role in detail.
For the broader infrastructure context, the Web3 Infrastructure for Enterprises and Blockchain for Banks articles cover the surrounding stack that interfaces with Swift and CCIP institutional integration.
Sources & References
- Chainlink, The Swift and Chainlink Partnership
- Swift and Chainlink, blockchain interoperability experiment results report (2023)
- The Fintech Times, Swift demonstrates interoperability of tokenised value transfer across blockchains
- Cointelegraph, Top banks work to make blockchains interoperable
- Chainlink, CCIP Architecture Overview
- Chainlink Blog, Chainlink Introduces CCIP 2.0 (28 September 2026)
- BIS, Project Agorá
Work with Matrixed.Link
Matrixed.Link operates Chainlink oracle infrastructure, validator nodes, full-stack blockchain infrastructure for protocols and institutions that demand institutional-grade reliability. ISO/IEC 27001:2022 certified. AAA-rated provider on StakingRewards. Continuous operations since the Chainlink Oracle Olympics.
Long-term partnerships with Chainlink, Lido, Enjin, Stake.link, bitsCrunch.
Contact Matrixed.Link to discuss your infrastructure needs.
