Saudi Arabia left the mBridge CBDC platform in May 2025 and the world found out sixteen months later. The same week, diesel hit a record $6.51 and two of the largest suppliers reached for the same lever: a ban. Permissioned systems fail quietly, and the instrument is always permission.
On 13 May 2025, Saudi Arabia's central bank completed its proof of concept on mBridge and left the platform. The world learned about it on 20 September 2026, when the Financial Times reported it. Sixteen months passed between the decision and the disclosure, and nobody announced anything — because in a permissioned system there is nothing to announce. A participant either holds a seat at the table or it does not, and the seating chart is not public.
That gap is the story. Not the exit itself.
mBridge was built in 2021 by the BIS Innovation Hub together with the central banks of China, Hong Kong, Thailand and the UAE. It is not a stablecoin. Each participating central bank runs its own CBDC node on a shared ledger, and the platform settles cross-border payments and foreign exchange between them. The pitch is efficiency: fewer correspondent banks, less settlement time, less dollar.
The exits have been steady. The BIS itself left in October 2024, after the project reached minimum viable product. Then-General Manager Agustín Carstens said the departure was not politically motivated, which is the kind of thing one says when it might have been. Saudi Arabia joined as a full participant in June 2024 and was gone by May 2025. Its statement to the FT is a masterpiece of process language: "As planned, SAMA successfully completed its mBridge [proof of concept] on 13 May 2025. Following the completion of the PoC, SAMA is no longer a participating member of mBridge."
The platform has not collapsed. Macao joined in 2026 and activated the system for participating banks on 2 June. Four core central banks remain. An anonymous source told the FT that Riyadh continues to engage with mBridge discreetly despite the formal exit — a claim I cannot verify, and one worth holding loosely.
The mechanism is worth naming, because it repeats.
A multi-CBDC coalition requires that each member run infrastructure another member's rules can reach. Settlement is final only inside the club's rules. That is fine while interests align and expensive the moment they do not — and a country that has spent years building alternatives to dollar rails is not going to advertise that it stepped back from one. Saying so publicly would be a statement about the project's viability, and the statement would be its own news.
So the exit is quiet, the discovery is late, and by the time it is public the decision is sixteen months old. This is how permissioned systems age. Not by failing an audit, but by losing participants who never say why.
On 21 September the AAA national average for diesel in the United States reached $6.51 a gallon — an all-time record, roughly 76 percent above the pre-war average of about $3.70, and past the previous record of $5.82 set in June 2022. Gasoline sits at $4.48, up from $3.18 a year earlier.
The cause traces to the US-Iran war that began in late February 2026 and the Strait of Hormuz disruption that followed. But the detail that matters is what came next. Russia, which restricted diesel exports on 9 July after Ukrainian drone strikes damaged refining capacity, is considering extending that ban beyond 30 September. The United States — one of the world's largest diesel exporters — faces congressional pressure to impose an embargo of its own. Interior Secretary Doug Burgum has said a US export ban could ultimately raise prices rather than lower them, which is an unusually candid admission from a government about its own instrument.
Two of the largest suppliers, both reaching for the same lever. And the lever is a ban.
That is what state control of a strategic good looks like in practice. The tool is not production. The tool is permission.
The strongest case against this argument is that mBridge is not a failure, and it would be dishonest to write it as one. Cross-border settlement between central banks is genuinely slow and genuinely expensive. Correspondent banking is genuinely concentrated. A shared ledger for FX is a real engineering answer to a real problem, and the members that stay get settlement finality they did not have before. Macao joined, which means someone examined the thing and wanted in.
Export bans can be rational, too. If domestic refining capacity is damaged by drone strikes, keeping fuel at home is a defensible policy rather than a plot.
The argument is not that the people doing this are villains. The argument is about what kind of instrument they are holding.
I could not verify the FT's account of continued discreet engagement. It rests on an anonymous source, and I have not confirmed it and I am not treating it as fact. The sixteen-month gap, by contrast, is on the record: a May 2025 completion and a September 2026 disclosure, both public.
Not that bitcoin "wins." That framing is lazy, and it is not what the evidence shows this week.
What the evidence shows is a difference in what can be pressured. mBridge's members can be leaned on because there is a member list, a governance body and a seating chart. A diesel exporter can be ordered to stop exporting because there is a ministry to issue the order. Both are permissioned systems in the precise sense: participation is granted, and what is granted can be withdrawn, or used as leverage.
Bitcoin's difference is not moral. It is structural. There is no administrator to threaten, no member list to leak, no proof of concept to "complete as planned" on a date that suits everyone.
On 21 September the price closed the week at $85,472 — above its 50-week average of $78,628 for the first time in 45 weeks. I measured both numbers myself from Kraken's weekly closes rather than taking them from a newsletter. The Fear and Greed index sits at 78, extreme greed.
The price is the least interesting part. What it is pricing is the absence of a lever.