Three settlement assets, no shared rulebook: the IMF warns tokenization could fragment global finance as easily as it unifies it.

Naming all three as rivals, not a public-versus-private choice, is the actual news.

Rate each claim by what the IMF actually delivered on 1-2 July, not by the word "fragment" in the headlines it produced.

Four shipped claims and one open question. The open question decides whether this note ages into policy or into a citation.

Lay the IMF's taxonomy over the reserve rules already in force and the fragmentation risk stops being hypothetical.

• One ledger, three assets. Tokenization is a technology, not a settlement asset. Deposits, stablecoins, and wholesale CBDC are competing to do the same job, not variants of the same answer.
• The split in the diagram is already live. Circle running separate US and EU reserve pools is the exact incompatible-platforms scenario the IMF warns about, not a hypothetical.

The three-way race finally has a referee, not just competitors making their own case. BIS's June chapter judged stablecoins against a singleness, elasticity, and integrity test. The Fed's CBDC ban committed the US to private stablecoins. The Bank of England built rules for its own tokenized-deposit and stablecoin regime. Adrian's note is the first to place tokenized deposits, stablecoins, and wholesale CBDC on the same page as rival architectures for the same settlement job, not one institution defending its own preferred model.

The fragmentation risk it names is already the reserve rulebook in force. GENIUS requires 100% T-bills and cash with no deposit floor. MiCA requires significant e-money token issuers to hold at least 60% of reserves in EU bank deposits. Circle now runs the two reserve pools separately to satisfy both. That is not a future risk scenario. It is the operating reality the IMF's warning describes in the present tense.

Naming the taxonomy is also naming the lobbying map. Every incumbent building tokenized deposits, every stablecoin issuer running GENIUS or MiCA reserves, and every central bank staffing a wholesale CBDC pilot now has an IMF citation to argue its own architecture is the one regulators should converge on, not fragment around.

• No quantitative model. The note describes a fragmentation scenario. It does not estimate a probability, a trigger threshold, or a stress-test result.
• No enforcement mechanism. IMF Notes are not binding. The call for common settlement standards and cooperative oversight has no assigned standard-setter and no date, unlike BIS's Project Agora, already under construction with 8 central banks.
• The taxonomy doesn't call the winner. Tokenized deposits, stablecoins, and wholesale CBDC remain live candidates in every jurisdiction tracee tracks. The note describes the race; it does not settle it.
• The GENIUS/MiCA example is a citation, not a discovery. tracee flagged the same reserve split on 24 June. The IMF uses it as illustrative evidence of a risk it is already naming, not new information.
• The urgency is partly rhetorical. Central banks have debated tokenization's fragmentation risk since Project Agora launched in 2022. "Strengthens or fragments" sharpens the headline on the same multi-year argument.

The note lands nine days after the BIS's own Annual Economic Report chapter graded stablecoins against a singleness, elasticity, integrity test, covered in tracee's 25 June briefing, and a day after Crédit Agricole's CACEIS settled its first tokenized fund trade in a bank-issued euro stablecoin, covered 3 July. All three sit inside the same five-week run that produced the US ban on a Fed CBDC through 2030, covered in the 26 June briefing, the Bank of England's systemic stablecoin rulebook, covered 23 June, and MiCA's transitional deadline, covered in the 24 June briefing. The IMF's three-way taxonomy is the first to put all three national bets, US stablecoins, UK tokenized deposits, EU-leaning wholesale CBDC, on a single page as competing answers to the same question.

GENIUS's 100% T-bill and cash requirement and MiCA's 60% EU bank deposit floor for significant e-money tokens are not new rules; both were final by 22 June 2026. What the IMF adds is the explicit claim that this divergence is itself the systemic risk, not merely a compliance cost Circle absorbs by running two reserve pools.

Adrian's note adds no new data. It is the first time a multilateral institution has placed tokenized bank deposits, stablecoins, and tokenized central bank reserves on equal footing as rival settlement assets, rather than judging one in isolation, and admitted the world has already chosen fragmentation over convergence: a US reserve rule built for T-bills, an EU reserve rule built for bank deposits, and an issuer, Circle, running both books at once because no shared standard exists to pick one. The "strengthens or fragments" framing sharpens an argument central banks have had since Project Agora launched, but the evidence behind it, GENIUS, MiCA, and Circle's dual reserve pools, is real and already in production.

Watch three things:

• OCC's GENIUS Act final rules, 18 July 2026. Sets whether the US reserve regime narrows or widens the gap with MiCA's bank-deposit floor.
• ECB's Pontes decision, Q3 2026. The clearest signal on whether the EU doubles down on wholesale CBDC as its answer to the IMF's taxonomy.
• DTCC's Canton go-live, October 2026. The first large-scale test of whether settlement can interoperate across more than one of the IMF's three asset types, or confirms the fragmentation it warns about.