Wholesale CBDC, explained: central bank money for the settlement layer

Wholesale CBDC is central bank money in tokenized form, issued for financial institutions to settle transactions among themselves on programmable ledgers. It is the counterpart of the reserves banks already hold at the central bank, moved onto infrastructure where it can settle against tokenized assets directly. Retail CBDC, the version that generates the political headlines, would put central bank money in the public’s hands; wholesale CBDC stays inside the institutional circle that already uses central bank money every day.

That distinction does most of the explanatory work. The controversies attached to retail CBDC, surveillance, deposit flight, the US statutory ban through 2030, largely do not apply to the wholesale form, which is why central banks that will not touch retail issuance still run wholesale pilots.

Reserves are already digital. So what changes?

Banks settle with each other today in central bank money held in RTGS systems, and those balances are electronic. What an RTGS balance cannot do is leave its system. It settles the instruction types the system supports, during the hours the system runs, and it cannot be locked against a tokenized bond on another platform so that both legs move as one.

Tokenized central bank money could. The point of wholesale CBDC is to let the safest settlement asset that exists reach the new places settlement is happening: atomic delivery-versus-payment against tokenized securities, payment-versus-payment across currencies, around-the-clock finality, and conditions attached in code. The BIS’s unified-ledger blueprint puts it at the core of the future stack for exactly this reason: everything else can tokenize, but the tier-one settlement asset has to come from the central bank.

The projects, and where each actually stands

Project Helvetia (Switzerland) is the furthest along anywhere. The SNB has issued wholesale CBDC on the SIX Digital Exchange since December 2023 and real tokenized-bond issues have settled in it. In June 2025 the SNB extended the pilot to at least mid-2027 and expanded it to test a second approach, synchronizing tokenized-asset settlement with traditional central bank money over an RTGS link. The SNB states plainly that the extension is not a commitment to introduce wholesale CBDC permanently.

Pontes and Appia (euro area) are the ECB’s dual track, committed in July 2025. Pontes is the pragmatic bridge: link market DLT platforms to the existing TARGET services so DLT-based transactions settle in central bank money, with a pilot planned from the third quarter of 2026. Appia is the long-term program for an integrated European platform, with the ECB pointing to the late 2020s.

Project mBridge (BIS Innovation Hub, then handed over) built a shared multi-CBDC platform on which central banks in China, Hong Kong, Thailand, the UAE, and Saudi Arabia settle cross-border in their own digital currencies. It reached minimum viable product in mid-2024, at which point the BIS handed it to the partner central banks, its General Manager noting it was “not mature enough to start operating.”

Project Agorá (BIS and the Institute of International Finance) is the broadest test of the two-tier vision: eight central banks and more than 40 private financial institutions putting tokenized central bank reserves and tokenized commercial bank deposits on one platform. It published prototype results in May 2026 demonstrating atomic multi-currency settlement and moved toward real-value testing, with the BIS explicit that Agorá “is not about building a finished product.”

The United States sits out this list entirely: the CBDC prohibition enacted in July 2026 bars Federal Reserve issuance through 2030, wholesale included on its face, which leaves tokenized deposits as the American route to programmable settlement money.

What it would change for settlement

For cross-border settlement, wholesale CBDC attacks the problem at the settlement-asset level. Today a cross-border payment settles across chains of commercial-bank nostro accounts because there is no shared risk-free asset both ends can hold. A platform where multiple central banks’ money coexists, the mBridge and Agorá design, would let banks in different jurisdictions settle with each other in central bank money directly, collapsing the correspondent chain for the flows it covers. Combined with atomic settlement, it would also remove the principal risk that CLS was built to contain in FX.

The reach question remains, as it does for every settlement network: each pilot covers certain currencies, certain members, certain asset types. A wholesale CBDC platform is another venue with a perimeter, not a universal solvent.

Realistic timelines

Read the official language and the pattern is consistent: dated pilots, undated production. Pontes has a pilot date (Q3 2026); Helvetia has an extension horizon (mid-2027); Agorá is moving to real-value testing; mBridge has no published launch date. No central bank has committed to permanent wholesale CBDC issuance. A reasonable planning assumption is that tokenized central bank money starts settling meaningful volumes in a first jurisdiction in the late 2020s, and that coverage stays partial well beyond that.

Where Frame fits

Frame is the settlement layer for global finance: one integration to orchestrate payments at scale across fiat rails, stablecoins, and tokenized deposits, with compliance enforced on every transaction and settlement in seconds instead of days.

Wholesale CBDC, when and where it reaches production, slots into that picture as another rail. Frame is rail-neutral by design: an institution’s payments route across whichever rail fits the corridor, the counterparty, and the governing policy, and the Rules Engine enforces those policies inside settlement on every rail alike. Institutions do not need to predict which central bank projects ship, or when; a settlement layer that spans the rails absorbs each one as it arrives, while the payments keep moving on the rails that exist today.

See how a rail-neutral settlement layer works: the Frame Blueprint.