How CBDC Pilots Are Reshaping Cross-Border Payment Infrastructure
Wholesale CBDC pilots from Project Agorá to mBridge are testing new settlement models for cross-border payments. Here is what payment strategists and providers should track.
Key takeaways
- FedNow is an instant payment rail, not a CBDC, while the digital euro is the advanced-economy retail CBDC closest to a first issuance
- Wholesale multi-CBDC platforms such as Dunbar, mBridge and Agorá target the bank settlement layer, not consumer wallets
- The cost gap is the driver: the World Bank puts the global average cost of sending $200 at 6.36% in Q3 2025, with banks at 14.99%
- Every pilot still leaves KYC, AML and sanctions screening outside the new settlement layer, so integration work remains
Why CBDC pilots matter to payment providers
Cross-border payments have a well-documented cost problem. The World Bank's Remittance Prices Worldwide database puts the global average cost of sending US$200 at 6.36% in Q3 2025, with the corridor-weighted average at 5.04% and banks the most expensive provider type at 14.99% (World Bank, Remittance Prices Worldwide Q3 2025).
Those figures are not abstract. They track a public benchmark: UN Sustainable Development Goal 10.c calls for reducing the transaction cost of migrant remittances to less than 3% by 2030, and for eliminating corridors where costs exceed 5% (UN SDG 10.c; UNECE indicator 10.c.1). That gap between the target and the current average is the commercial case behind most sovereign payment projects.
Central bank digital currency pilots are one attempt to close it. But "CBDC cross border payments" is a loose label covering three very different designs, and conflating them leads to bad strategic planning. The distinction that matters most for anyone building or buying payment infrastructure:
- Retail CBDC — a central bank liability held by households and firms, as in the digital euro project.
- Wholesale or multi-CBDC platforms — settlement between financial institutions using tokenised central bank money, as in Project Dunbar, Project mBridge and Project Agorá.
- Instant payment system interconnection — no CBDC at all, as in Project Nexus, which links domestic fast payment rails.
A related trap: FedNow, the Federal Reserve's instant payment service launched in July 2023, is not a CBDC. The Federal Reserve states that the service is not related to a digital currency and is a payment service offered to banks and credit unions to transfer funds for their customers (Federal Reserve, FedNow FAQ). Mechanically, FedNow settles interbank obligations by adjusting balances in participating institutions' master accounts at the Reserve Banks — central bank money at the settlement layer — while the end user experiences a transfer between commercial bank deposits (Richmond Fed; Boston Fed). Either way, there is no general-purpose retail CBDC in the United States, and the Federal Reserve has said it would only issue one under an authorising law.
Below we compare the main pilots and assess what they change for correspondent banking, FX settlement and fintech integration. These initiatives are on our research watchlist, not recommendations.
| Project | Participants | Status | Design focus |
|---|---|---|---|
| Digital euro | ECB / Eurosystem, EU co-legislators | Preparation phase concluded Oct 2025; pilot planned from H2 2027 | Retail CBDC with holding limits, distributed by banks and PSPs |
| Project Dunbar | BIS Innovation Hub, Australia, Malaysia, Singapore, South Africa | Prototypes delivered Mar 2022 | Multi-CBDC shared platform for direct FI-to-FI settlement |
| Project mBridge | BIS IH, China, Hong Kong, Thailand, UAE (Saudi Arabia exited) | MVP Jun 2024; BIS handed over Oct 2024 | Wholesale multi-CBDC ledger with PvP FX |
| Project Agorá | BIS, IIF, seven central banks, 40+ institutions | Prototype report May 2026; real-value testing Jul 2026 | Tokenised reserves plus tokenised commercial deposits |
| Project Nexus | BIS IH, India, Indonesia, Malaysia, Philippines, Singapore, Thailand | Handed to Nexus Global Payments in 2025 | Interconnection of domestic instant payment systems |
What each pilot actually tests
Project Dunbar: the design blueprint
Dunbar, run by the BIS Innovation Hub's Singapore Centre with the Reserve Bank of Australia, Bank Negara Malaysia, the Monetary Authority of Singapore and the South African Reserve Bank, delivered two prototypes in March 2022 on the Corda and Quorum distributed ledgers (BIS).
Its lasting contribution is problem framing rather than code. The project isolated three questions: which entities may hold and transact in foreign CBDCs; how to simplify cross-border flows while respecting differing national rules; and what governance lets central banks share critical national infrastructure.
Its answer on access was a "hybrid" model: non-local banks hold CBDCs as a direct claim on the issuing central bank, but a sponsoring bank in the local jurisdiction performs KYC, onboarding and AML/CFT monitoring. The BIS notes this limits the efficiency gains of removing intermediaries, because a control layer survives even when the settlement layer is flattened — and it raises a commercial question about why banks would take on the sponsoring role.
That finding is the most useful takeaway from Dunbar for payment strategists: eliminating correspondent banks from settlement does not eliminate the compliance function they perform.
Project mBridge: real volume, unresolved maturity
mBridge, begun in 2021 by the BIS Innovation Hub with the Bank of Thailand, the Central Bank of the UAE, the Digital Currency Institute of the People's Bank of China and the Hong Kong Monetary Authority, reached minimum viable product stage in June 2024 (BIS).
It is the most advanced multi-CBDC platform in live-value terms. The Atlantic Council's CBDC tracker puts cumulative transaction volume at US$55.49 billion, a roughly 2,500-fold increase on early-2022 pilots, with the e-CNY accounting for over 95% of settlement volume (Atlantic Council CBDC Tracker). Treat that concentration as two signals at once: it demonstrates scale in one currency context while showing how far the platform is from balanced multi-currency usage.
In October 2024 the BIS ended its involvement and handed mBridge to its central bank partners. General manager Agustín Carstens said the BIS had "graduated out" of the project, that it was "not mature enough to start operating", and that it remained years from that point (BIS, speech by Agustín Carstens, 31 October 2024). The handover, days after a summit where alternative payment systems were discussed, made mBridge a proxy in debates about sanctions circumvention — a characterisation the BIS disputed. Membership has also shifted: the Saudi Central Bank, announced as a full participant in 2024, subsequently confirmed its participation ended after its proof-of-concept work concluded (Financial Times).
For practitioners, mBridge is the clearest example of a platform that removes correspondent banks from the payment path while remaining dependent on participating central banks for governance, liquidity and legal finality.
Project Agorá: tokenising the existing system
Agorá takes the opposite philosophical route. Convened by the BIS and the Institute of International Finance in April 2024, it brings together seven central banks — including the Bank of England, the Federal Reserve Bank of New York, the Bank of France for the Eurosystem, the Bank of Japan, the Bank of Korea, the Bank of Mexico and the Swiss National Bank — plus more than 40 regulated financial institutions (BIS).
Its May 2026 report found that tokenised commercial bank deposits can be combined with tokenised central bank reserves on a shared platform to achieve atomic, multi-currency settlement, with legal settlement finality achievable across all seven participating jurisdictions (BIS). Real-value testing in July 2026 involved 28 institutions and central banks completing transactions totalling roughly CHF 800,000 across 17 scenarios, with individual values between CHF 9,000 and CHF 125,000 (IIF, Project Agorá).
Agorá explicitly aims to modernise correspondent banking rather than replace it, preserving the two-tier monetary system. The prototype also shows where friction remains: cross-currency amounts in testing were entered manually by a cross-currency provider, and issuance and redemption of tokenised deposits required manual bank approvals.
Project Nexus: the same goal without CBDC
Nexus is the reminder that distributed ledger technology is optional. Started in 2022 by the BIS Innovation Hub's Singapore Centre, it connects domestic instant payment systems through a single standardised gateway using APIs and ISO 20022 messaging, targeting end-to-end delivery in under 60 seconds in most cases (BIS).
In 2025 the central banks of India, Indonesia, Malaysia, the Philippines, Singapore and Thailand incorporated Nexus Global Payments to take the scheme into live implementation, and the BIS handed the project over. For payment providers weighing CBDC exposure, Nexus is the counterfactual: faster, cheaper cross-border payments may arrive through interconnection of existing rails before any CBDC reaches production scale. Our explainer on ISO 20022 migration covers the messaging layer both approaches depend on.
Where the digital euro fits
The digital euro is the furthest advanced retail CBDC among major advanced economies, and it has a cross-border dimension even though it is designed for euro area use.
The Eurosystem's preparation phase ran from November 2023 to October 2025. In October 2025 the ECB Governing Council decided to continue preparation, with a pilot exercise planned to start in the second half of 2027 and run for 12 months, and readiness for a potential first issuance during 2029 — conditional on adoption of the digital euro Regulation (ECB).
Legislative progress matters for planning: the Council of the EU adopted its position in December 2025 and the European Parliament adopted its position on the Single Currency Package in July 2026, with trilogue negotiations under way (ECB FAQ). The ECB has said it will not decide on issuance until the Regulation is adopted.
Design choices relevant to infrastructure planning:
- Holding limits per person were analysed at the request of co-legislators, and the ECB's technical analysis found no financial stability harm under the tested limits, including a crisis scenario.
- No interest would be paid on digital euro holdings, and businesses would not hold digital euro balances.
- Distribution would run through EU-licensed banks and payment service providers. The ECB selected 36 banks and payment firms in July 2026 to join the pilot (ECB).
- The ECB estimates total development costs of around €1.3 billion to first issuance, with operating costs of approximately €320 million per year from 2029, borne by the Eurosystem (ECB; ECB digital euro FAQ, Q27).
For cross-border providers, the digital euro is mainly a euro area domestic story today. Its international relevance depends on interoperability arrangements that are not yet settled.
Risks and costs
- Pilot is not production. mBridge reached MVP stage and recorded live volume, yet the BIS's own assessment was that it was years from operating. Treat announced milestones as engineering progress, not availability.
- Legal finality and settlement risk. Agorá found settlement finality achievable across its seven jurisdictions, but noted that technical, operational and contractual requirements still need definition in each. Outside those jurisdictions, finality is untested.
- Compliance does not disappear. Dunbar's hybrid model keeps sponsoring banks responsible for KYC, onboarding and AML/CFT. Sanctions screening remains a national responsibility, and mBridge's structure has attracted geopolitical scrutiny as a result.
- Disintermediation and funding models. Holding limits and no remuneration are designed to limit deposit outflows in the EU, but they also cap utility for larger flows. Banks still carry integration costs, which the ECB describes as close to the European Commission's initial estimates and similar to those incurred for the Payment Services Directive (ECB).
- Liquidity and FX. Atomic payment-versus-payment settlement reduces principal risk, but FX conversion, prefunding and end-of-day cut-offs persist. In Agorá's prototype, cross-currency amounts were still entered manually.
- Concentration and operational risk. Shared platforms concentrate settlement in single infrastructures. Dunbar's design work targeted isolation of failures so that one participant's problems do not cascade — a design requirement, not a guarantee.
- Cost outcomes are not automatic. Today's cheapest channels are digital: the World Bank records digital remittances at 4.59% and money transfer operators at 4.72%, against banks at 14.99% (World Bank RPW, Q3 2025). Whether a new rail lowers end-user prices depends on competition and pricing, not on the technology.
- Fragmentation. Parallel projects in different jurisdictions may deliver interoperability between platforms late or not at all. Availability of any digital currency product is jurisdiction-specific: in the EU the digital euro pilot is limited to selected participants; in the US there is no retail CBDC in development, only instant payment infrastructure.
Bottom line
CBDC pilots are reshaping the cross-border stack at the settlement layer, not the consumer interface. Dunbar and mBridge show that direct institution-to-institution settlement is technically feasible and that compliance intermediaries persist regardless; Agorá shows the same outcome can be reached by tokenising reserves and deposits inside the existing two-tier system; Nexus shows that interconnection of instant rails may deliver speed and cost improvements with no CBDC at all. The digital euro is the nearest-term retail example, but its first issuance depends on EU legislation and is not expected before 2029. For payment providers, the practical work today is ISO 20022 readiness, hybrid access and sponsoring arrangements, and liquidity design — not CBDC product build-out. For related background, see our explainers on cross-border instant payments, SWIFT gpi and cross-border friction, RTGS vs net settlement and retail CBDC design choices.
Follow the shortlist
Following payments, banking and market-infrastructure change? Our global ETF and stock research watchlist is tracked daily. Premium members get the full list, the reasoning behind every entry and updates when entries change.
See premium accessInformation and education only, not investment advice.
Related
Stablecoin Regulation Compared: EU MiCA vs the US GENIUS Act
How MiCA and the GENIUS Act differ on who may issue payment stablecoins, what must back them, how redemption works and where depeg risk still sits.
FinTechTokenised Deposits vs Stablecoins: How Bank-Issued Money Differs
Tokenised bank deposits and reserve-backed stablecoins both move value on programmable rails, but they differ in legal claim, deposit insurance, balance-sheet treatment and risk.
FinTechInstant Payments for Small Business: FedNow, RTP, Zelle Compared
How FedNow, RTP and Zelle differ for US small businesses on speed, settlement finality, limits and cost, plus which emerging options fit SME cash flow.