CBDC Cross-Border Settlement: How Multi-CBDC Platforms Work

Why cross-border payments still need fixing

Send money from one country to another and, in most cases, it does not travel directly. It hops through a chain of correspondent banks, each holding pre-funded accounts with the next, each applying its own cut-off times, compliance checks and fees. The payment is only as fast as the slowest link, and the bank in the middle is often closed when the sender is awake.

The costs are structural rather than accidental. Banks lock up liquidity in nostro accounts so that payments can settle on time. Foreign exchange trades settle in two separate legs, which leaves a gap in which one side can pay and the other fail to deliver. That risk has a name, Herstatt risk, after the German bank whose closure in 1974 left counterparties exposed after they had already paid. CLS Bank, launched in 2002, removed most of that risk for the major currencies through payment-versus-payment settlement, but only for a limited set of currencies and only within its own operating window.

The G20 made the problem official in 2020 with a cross-border payments roadmap that set targets for cost, speed, access and transparency, to be met by 2027. The Financial Stability Board's progress reports have since described improvements on several measures as slow, which is a large part of why central banks are testing a different settlement design instead of polishing the old one.

Wholesale versus retail CBDCs

A central bank digital currency (CBDC) is a digital form of central bank money. Two kinds matter here, and they solve different problems.

	Wholesale CBDC	Retail CBDC
Who uses it	Banks and regulated financial institutions	Households and businesses
Main purpose	Interbank and securities settlement, including cross-border	Everyday payments, cash-like use
Privacy question	Limited, participants are known institutions	Central, a major political issue
Cross-border relevance	High: most live pilots are wholesale	Possible, but harder (legal, privacy and compliance differences)

Almost every serious cross-border experiment is wholesale. Banks already settle with central bank money, so putting that money on a shared platform changes the plumbing without asking ordinary customers to adopt anything new.

Three ways countries can connect their systems

A 2021 report to the G20 by the BIS Committee on Payments and Market Infrastructures, the BIS Innovation Hub, the IMF and the World Bank grouped multi-CBDC arrangements into three designs. The grouping is still the most useful way to compare projects.

• Compatible systems. Each country runs its own CBDC, but they share standards such as message formats and compliance rules. Integration is light, so progress is fast, but the payment still crosses a border between two separate systems.
• Interlinked systems. Technical links or a common clearing mechanism connect separate CBDC systems. This removes much of the friction without forcing any country to give up control of its own ledger.
• A single multi-CBDC system. Several central banks put their currencies on one shared platform. Settlement is simplest and fastest here, which is also why governance is hardest: someone has to decide who can join, who pays for the platform and what happens during a dispute.

What the main pilots actually showed

Project mBridge. Began as a collaboration between the Hong Kong Monetary Authority, the Bank of Thailand, the People's Bank of China and the Central Bank of the UAE, supported by the BIS Innovation Hub. A 2022 pilot involved around twenty commercial banks and processed real-value transactions of about 22 million US dollars. Saudi Arabia's central bank joined as a full participant in 2024, and later that year the BIS withdrew from the project, leaving it to the participating central banks. The technology worked; the lesson was that the group running a platform matters as much as the platform.

Project Dunbar. A BIS Innovation Hub project with the central banks of Australia, Malaysia, Singapore and South Africa. It tested a shared platform on which banks could hold and exchange several wholesale CBDCs, and its report in 2022 set out how that could reduce reliance on correspondent chains, while flagging that governance and legal design would be the harder tasks.

Project Jura. Run by the BIS Innovation Hub with the Banque de France and the Swiss National Bank. It settled euro and Swiss franc wholesale CBDCs against each other on a shared platform operated by a third party, and explored how to keep that arrangement compatible with each country's rules.

Project Agorá. Announced in 2024 by the BIS with seven central banks, including the Federal Reserve Bank of New York, and more than forty private financial firms. It differs from earlier pilots by combining tokenised commercial bank deposits with wholesale central bank money on one programmable platform, which reflects a growing view that deposits, not only CBDCs, will carry the value.

Project Nexus. Not a CBDC at all. It connects national instant payment systems so that a payment in one country can reach an account in another within about a minute. It is relevant because it tackles the same cost and speed problem with far less technical change, and it is the practical alternative every CBDC project is measured against.

What changes for banks and corporate treasurers

• Less pre-funding. If settlement is near-instant and final, banks need smaller balances sitting in foreign accounts. That frees capital, but it also removes a steady source of fee income for correspondent banks.
• Intraday liquidity shifts. Round-the-clock settlement means liquidity must be managed around the clock too. Collateral and funding processes built for a weekday window will need rethinking.
• Safer foreign exchange. Payment-versus-payment settlement on a shared ledger can remove Herstatt risk for the currencies on the platform, extending what CLS already does for a fixed list of currencies.
• Treasury visibility. Corporates could see settlement status in real time instead of waiting for a confirmation message, which matters for cash forecasting and supplier payments.
• Data and messaging. Most connections will still need to speak ISO 20022, the structured payment message standard that became mandatory for SWIFT cross-border payment instructions in late 2025. Cleaner data helps both screening and reconciliation.

The risks that slow it down

Legal finality. A transfer is only useful if every jurisdiction agrees on the moment it becomes irrevocable. Legal frameworks differ, and a platform that spans several of them has to settle that question before it scales.

Governance and geopolitics. Several systems that bypass the dollar-based network are attractive to countries that want less exposure to sanctions. That same feature raises concern among regulators in other countries, and it helps explain why the question of who sits on a platform is as contested as how it works.

Compliance. Anti-money-laundering controls, sanctions screening and the FATF travel rule, which requires originator and beneficiary information to travel with a transfer, must work on a ledger as well as they do on today's messaging network. A fast payment that cannot be screened is a liability, not a feature.

Cyber and concentration risk. A shared platform becomes a single, high-value target. Operational resilience, including how to keep paying if the platform goes down, is a design requirement.

Fragmentation. If each bloc builds its own platform and none connect, the result could be a more fragmented system than the one it replaces.

What to do now

1. Follow the official project pages of the central banks and the BIS Innovation Hub rather than commentary. Programme status changes quickly, and announcements often reverse earlier expectations.
2. Map your cross-border flows by currency pair and corridor. The first CBDC links will be in specific corridors, not everywhere.
3. Check that your systems can handle structured ISO 20022 data end to end, since every future settlement option depends on it.
4. Review your intraday liquidity and collateral processes for 24/7 operation.
5. Treat tokenised deposits and wholesale CBDCs as complementary, and watch which model your own central bank and main banking partners favour.

Frequently asked questions

How is a wholesale CBDC different from a stablecoin?

A wholesale CBDC is a direct claim on the central bank, so it carries no credit risk from a private issuer. A stablecoin is issued by a private company and backed by reserves held elsewhere, so its safety depends on the issuer and the quality of those reserves.

Will CBDCs replace correspondent banking?

Not entirely, and not soon. Even the most ambitious pilots cover a handful of currencies. Correspondent banks will remain important for the hundreds of corridors that no platform serves, and many will adapt by joining new platforms as participants.

Does instant settlement eliminate foreign exchange risk?

It removes settlement risk, the danger that one leg of a trade is paid and the other is not. It does not remove price risk. The exchange rate can still move before a trade is agreed.

Can individuals or small businesses use these platforms?

Not directly. Today's cross-border pilots are wholesale and used by banks. Individuals and businesses would feel the effect through faster and cheaper payments offered by their own bank or payment provider.

Sources and further reading

• BIS, BIS Innovation Hub, IMF and World Bank, Central bank digital currencies for cross-border payments: Report to the G20 (2021), and the BIS project reports for Dunbar, Jura, mBridge and Agorá, on bis.org.
• Financial Stability Board, G20 Roadmap for Enhancing Cross-border Payments and its annual progress reports, on fsb.org.
• Financial Action Task Force, Recommendation 16 (the travel rule), on fatf-gafi.org.
• CLS Group, material on payment-versus-payment settlement, on cls-group.com.

Programme details are drawn from public announcements and can change; confirm current status with the sources above. This article is general information, not financial or legal advice.