BIS exited mBridge after $55.5B in digital-yuan settlements, then launched Agorá with G7 banks and SWIFT. The two architectures cannot converge, and the bilateral instant-payment patchwork is filling the gap.
Alpha Score of 50 reflects weak overall profile with moderate momentum, poor value, moderate quality, moderate sentiment.
The Bank for International Settlements did not graduate Project mBridge. It walked away from a live, renminbi-denominated wholesale settlement rail that had processed $55.5 billion across more than 4,000 cross-border transactions, roughly 95 percent of it in digital yuan. The exit, announced on October 31, 2024, was framed as a natural handover to the central banks that remain: the People’s Bank of China, the Hong Kong Monetary Authority, the Bank of Thailand, the Central Bank of the UAE, and Saudi Arabia’s SAMA, which joined as a full participant that same year. No Western central bank is on the list.
The simple read treats the BIS departure and the subsequent launch of Project Agorá – a tokenized correspondent banking experiment with seven G7-aligned central banks and over forty private institutions including JPMorgan (JPM), Citi (C), HSBC (HSBC), and SWIFT – as two parallel CBDC experiments. The better read is that they are competing answers to the same question, sponsored by competing blocs, with zero overlapping membership. The era of multilateral CBDC interoperability, the central idea behind the BIS Innovation Hub’s payments work, is dead.
mBridge began as Inthanon-LionRock between HKMA and the Bank of Thailand in 2019, expanded to include the PBOC and the UAE central bank, and reached minimum viable product status in 2024 with SAMA’s accession. The technical premise was the only meaningfully cross-bloc multi-CBDC platform in operation: settlement of central bank digital currencies directly, peer-to-peer, without going through correspondent banks. That bypass is what made the project both technically interesting and politically combustible. Correspondent banking is the practical layer through which Western sanctions are imposed. A rail that does not need it cannot be sanctioned through the conventional pipes.
By late 2025, mBridge was in practice a renminbi-denominated wholesale settlement rail for trade between China and the Gulf, running outside the dollar correspondent system. The technology worked. BIS General Manager Agustín Carstens framed the exit in operational terms: the project had matured, the partners could run it themselves, the BIS had completed its hub function. The framing is too clean to take at face value.
Three months before the announcement, Vladimir Putin had publicly floated a “BRICS Bridge” based on mBridge’s architecture as a route around dollar sanctions. At the Santander International Banking Conference in Madrid where Carstens declared the graduation, he was asked directly whether mBridge could be used by sanctioned states. He answered that mBridge is not the BRICS bridge and the BIS does not operate with any countries subject to sanctions.
Key insight: The BIS exit removed the last Western institutional presence from a live renminbi settlement rail, four months after a sanctioned head of state advertised the technology as a sanctions workaround.
Why did a project that had reached MVP, with willing participants, suddenly need the BIS to leave? The graduation framing does not explain it. The geopolitical pressure does. BIS departed a project it had spent four years building, four months after a sanctioned head of state advertised the technology as a sanctions workaround. The exit was political. The post-exit participant list confirms it.
mBridge’s architecture replaces correspondent banking with a shared ledger on which participating central banks issue and settle wholesale CBDCs directly. No intermediary bank holds nostro-vostro accounts. No SWIFT message carries the payment instruction. The settlement asset is the digital yuan for the vast majority of volume, with the other currencies available but far less used. The corridor links the world’s largest non-Western trade relationship – China to the Gulf – and it is growing.
Because settlement occurs directly between central bank digital currencies on a shared ledger, there is no correspondent bank to freeze, no SWIFT message to block. The only way to sanction the rail is to sanction the participating central banks themselves, an escalation that carries systemic risk. The BIS exit removed the last Western institutional presence from the project, leaving it entirely in the hands of central banks that have no incentive to build sanctions compliance into the rail.
Project Agorá is not the BIS pivot from mBridge. Architecturally they are not in the same category. Agorá preserves correspondent banking. mBridge disintermediates it. The Agorá design integrates tokenized commercial bank deposits with tokenized wholesale central bank money on a unified ledger, then routes those tokenized claims through the existing institutional structure. The central banks involved – the Federal Reserve Bank of New York, Banque de France for the Eurosystem, the Bank of England, Bank of Japan, Bank of Korea, Banco de México, and the Swiss National Bank – are the central banks with the most institutional capital invested in correspondent banking. The forty-plus private participants are the incumbent banks and infrastructures that operate it. SWIFT is in the working group.
Practical rule: Agorá tokenizes incumbency. mBridge disintermediates it. They cannot converge.
The testing phase is expected to deliver findings in the first half of 2026. The dollar comes out entrenched within the Agorá architecture. The tokenized correspondent design preserves USD’s role as the routing currency. SWIFT gets a reprieve. The messaging layer was supposed to be obsolesced by CBDC interoperability. Instead it is a named participant in Agorá and the de facto fallback wherever Agorá-style flows operate. ISO 20022 migration is complete. SWIFT has been repositioned as the messaging spine of the Western tokenized stack rather than its replacement.
For corporate treasury teams, the outcome is a more complex compliance surface. Flows in G7 corridors will move through Agorá-style tokenized correspondent rails. Flows in the China-Gulf corridor will move through mBridge or its successors. Everywhere else, a growing patchwork of bilateral instant-payment links is filling the gap. The unified cross-border CBDC future that BIS papers were drafting in 2022 is not arriving. The compliance burden multiplies because each rail has its own settlement finality, its own data standards, and its own sanctions exposure.
The countries with the most to gain from cross-border CBDC interoperability are not in either project, and they are not waiting for one to win. India’s NPCI International has built UPI links into Singapore via PayNow, the UAE, France, Bhutan, Nepal, Sri Lanka, and Mauritius, and is targeting ten or more corridors by year-end 2026. Cross-border UPI volume grew from 37,060 transactions in FY24 to over 755,000 in FY25, with the first four months of FY26 alone showing 601,000. These are retail rails, not wholesale. Their advantage over Agorá and mBridge is that they exist, they settle, and the host countries have signed on without committing to a CBDC architecture.
ASEAN is building the same approach as a bloc. The 2023 Leaders’ Declaration on Regional Payment Connectivity has linked Thailand’s PromptPay, Indonesia’s QRIS, Singapore’s PayNow, and Malaysia’s DuitNow into an expanding mesh, with full ASEAN interoperability targeted by the end of 2025. Intra-regional local currency settlement has more than doubled, from about 7 percent of regional payments five years ago to over 15 percent today. Brazil’s Drex CBDC pilot and Pix’s continued export ambitions follow the same logic. Build bilaterally. Avoid the bloc choice. Hedge with non-CBDC rails that already work.
The fragmentation is not a tail risk; it is the baseline. The risk event is that the split hardens further. Several developments would make it worse.
The only path back to multilateral CBDC interoperability would be a convergence of the two architectures. That requires one of two things: either mBridge adopts correspondent banking features, or Agorá abandons them. Neither is plausible. mBridge’s entire value proposition is the elimination of correspondent banks. Agorá’s entire political constituency is the incumbent banks and infrastructures that operate them. The BIS Innovation Hub was built on the premise that cross-border payments infrastructure could be designed neutrally and adopted multilaterally. Withdrawing from mBridge while launching Agorá ended that premise. The infrastructure was never neutral, and pretending otherwise was the project’s foundational illusion.
Bottom line for traders: The unified CBDC future is dead. What comes next is a fragmented map of bilateral, bloc-specific, and corporate-built corridors, negotiated one pair of countries at a time. There is no single global rail in it.
The practical consequence for anyone allocating to digital-asset infrastructure or tracking the tokenization theme is that the investable universe splits along geopolitical lines. The Agorá stack benefits incumbent financial infrastructure in G7 markets. The mBridge stack benefits renminbi internationalization and the digital-yuan ecosystem. The bilateral instant-payment patchwork benefits domestic payment champions like NPCI and Pix. There is no single bet on “CBDC interoperability” because there is no single interoperability to bet on. The trade is fragmentation, and it is already live.
Drafted by the AlphaScala research model and grounded in primary market data – live prices, fundamentals, SEC filings, hedge-fund holdings, and insider activity. Each story is checked against AlphaScala publishing rules before release. Educational coverage, not personalized advice.