The Bank for International Settlements published the results of Project Agorá on May 28, 2026. Seven central banks, more than 40 regulated financial institutions, and two years of work produced a prototype showing that tokenized wholesale cross-border payments can settle atomically — simultaneously, across currencies and jurisdictions, in seconds. The BIS says the project is “advancing to real-value testing.”
No timeline for that testing has been published. The architecture has not run on live central bank reserves. The legal frameworks that would govern operation across seven jurisdictions don’t exist yet. The prototype established that the technology works. What comes after that is a harder problem.
##What the prototype built##
Project Agorá ran on two layers. The first holds tokenized wholesale central bank reserves — a limited wholesale CBDC for each participating jurisdiction. The second holds tokenized commercial bank deposits. Payments cross both layers simultaneously using atomic settlement: all balance updates complete at once or none complete at all, which eliminates the counterparty exposure window that exists whenever two sides of a transaction settle at different times.
AML and sanctions screening run in parallel with settlement rather than sequentially. This matters because compliance checks are a primary source of delay in current correspondent banking; running them concurrently rather than after the payment is confirmed cuts that latency directly.
Cross-border payments reached $195 trillion in 2024, per FXC Intelligence data in the BIS report. That figure is projected to reach $320 trillion by 2032. The correspondent banking system moving most of that volume requires pre-funded liquidity at banks in each currency corridor, operates on business-hour schedules across multiple time zones, and settles in one to five business days depending on the route. Atomic settlement with continuous availability removes all three constraints. In a working system.
##Three problems the prototype didn’t solve##
The first is legal. In correspondent banking, a failed payment has an established liability chain: which institution is responsible, under which jurisdiction’s law, through which dispute process. A tokenized atomic settlement system spanning seven central bank jurisdictions has none of that in place. Contract law, insolvency treatment, and regulatory authority over cross-border token transfers remain unresolved in most participating countries. The BIS report acknowledged the need for “enabling legal frameworks” without specifying what those look like or when they might exist.
The second is liquidity management. Correspondent banking is friction-heavy, but it does one thing well: it lets participating banks pool liquidity using pre-funded nostro accounts they can draw on for same-day settlement in a given corridor. In a tokenized system where a bank’s holdings of a given currency’s tokenized reserves run low during peak volume, the mechanism for replenishing that position in real time is unclear. The prototype didn’t address this.
The third is integration. Project Agorá ran on purpose-built, permissioned infrastructure. The payment infrastructure it would connect to in production — SWIFT messaging, domestic RTGS systems, correspondent networks — is not tokenized and has no near-term path to becoming so. Both systems would need to run in parallel through a transition period whose length, cost, and operational complexity no prototype can accurately model.
##What it means for banks currently in the correspondent business##
The intuitive read is that tokenized wholesale settlement threatens correspondent banking. The more granular read: large correspondent banks with regulatory relationships in multiple jurisdictions are better positioned to participate in a tokenized system than they are to be displaced by one. The institutions at structural risk are smaller correspondents providing FX liquidity in thin currency corridors where a tokenized alternative would make their intermediary function unnecessary.
That transition, if it materializes, is probably measured in years rather than quarters. The BIS is not projecting a deployment timeline. Project Agorá produced a working proof of concept and a body of research. Institutions building tokenized payment infrastructure — on XRPL, on permissioned networks, on CBDC rails still under construction — now have BIS validation that the destination they’re building toward is technically achievable. The question of when, and under whose legal framework, is still open.
##Sources##
– BIS: “Project Agorá: Final Report” — May 2026
– CoinTelegraph: “BIS completes 2-year cross-border payments prototype with 7 central banks” — May 28, 2026
– FXC Intelligence: Cross-border payment volume data, cited in BIS report
