Swift Never Moved Money. Its New Blockchain Doesn't Either.

Stablecoins were supposed to be the thing that finally routed around Swift.

The pitch was easy. A dollar token settles in seconds, any hour, any day, straight from sender to receiver, with no chain of correspondent banks passing a message down the line and taking a cut and a day each. By this summer the stablecoin float had passed three hundred billion dollars, and billions of it kept moving every weekend while the banking system was closed.

If you wanted one point for the threat, that was it: value moving around the clock on rails that had never once asked Swift for permission.

On the 9th of July 2026, Swift answered. Its blockchain-based ledger was declared ready for initial use, with seventeen major banks (ANZ, BNP Paribas, BNY, Citi, HSBC, UBS, Standard Chartered, Wells Fargo and nine more, across six continents) preparing to pilot 24/7 tokenised cross-border payments. Swift built it in nine months from a Consensys-supported prototype (for a fifty-year-old messaging co-op, that is a sprint).

On paper, the incumbent shipped the thing that was meant to bury it. Look closer and it is a better story than that.

The tokenised deposits move around the clock. Final interbank settlement still happens the old way, through RTGS systems, correspondent accounts, or another arrangement the banks agree between themselves, potentially only when the relevant rail reopens.

The blockchain moves where the promise is. Final settlement follows on the rails Swift has run for half a century. The old plumbing underneath is not a crutch Swift will rip out later. Swift built it this way on purpose.

Which puts a bigger question on the table than "can a bank consortium ship a blockchain."

For fifty years Swift moved messages, never money. Can that franchise cross into a world where value itself has gone programmable, or has Swift just built the one thing that finally makes it optional?

Start with the tokenised deposit, because everything turns on it.

It is not a stablecoin. A conventional stablecoin is a transferable token issued against a pool of reserve assets, and most behave like bearer instruments anyone can hold. A tokenised deposit is the balance already sitting in your bank account, issued as a token on a ledger the bank controls. HSBC's tokenised dollar is a claim on HSBC. Citi's is a claim on Citi. They are not the same instrument, and a euro of one is not automatically a euro of the other, which is the entire reason this is hard.

Swift's ledger runs on Hyperledger Besu, an Ethereum client wearing a bank's clothes: EVM-compatible, so it speaks the same language as the public networks, but permissioned, so only vetted institutions validate and nobody joins without an invitation. Enterprise Onchain readers have watched this exact move before, when the central banks started running Ethereum and declined to call it Ethereum. It sits underneath Swift's existing ISO 20022 messaging, not on top of it. The message layer still says what should happen. The ledger now records and sequences it.

The ledger does not hold the money. It holds the commitments.

A banker will object that a tokenised deposit is itself commercial-bank money, and the banker is right. The ledger does move a claim on a bank, around the clock. What it does not move is the asset that finally settles the obligation between the banks.

When a bank in Singapore pays a bank in Brazil on a Saturday, Swift's ledger records and validates the interbank commitment, and each bank updates the tokenised-deposit position on its own environment, so the receiving customer can be paid. The interbank settlement, the part where central-bank money changes hands, waits for the RTGS systems in each country to open.

Follow one payment through a weekend, assuming the receiving bank advances its customer's funds before interbank settlement. Friday night, the commitment moves and the receiver's customer has funds. All weekend, one bank is out the money and the other owes it. Monday, the old rails settle up. The token was fast. The settlement was normal.

Which surfaces the oldest question in correspondent banking, in a new outfit.

Somebody must fund the interval between the customer payment and final settlement. Swift has not published whether each transaction is prefunded, collateralised, or supported by bilateral credit. If the receiving bank pays its customer before it has final funds, it is carrying settlement exposure to the sending bank across the gap. If the position is prefunded, that credit risk falls, but the liquidity cost of parking the money in advance does not.

There is nothing blockchain-native about either choice. It is the same trade between liquidity cost and credit risk that correspondent banking has always run on. What stablecoins collapse is the delay, by making transfer and settlement of the token the same event; the trade itself resurfaces in issuer reserves and redemption queues. Last week's piece followed the float through a stablecoin. Here it is again, wearing a bank charter.

As far as the public record shows, the ledger coordinates the promise and leaves the funding to the banks. That is the point: you can take the delay out of the customer's experience without taking the choice between liquidity cost and credit risk out of the system. The settlement exposure did not vanish. It changed venue.

Money that moves on a Saturday is a new product, one that stablecoins forced into existence. The plumbing under it is the same plumbing with a faster front end. Both are true, and the second is exactly why Swift, and not a crypto rail, is the one handing this to seventeen regulated banks.

Think about air traffic control. It owns no aircraft and carries no passengers. It sequences the traffic, standardises the instructions, and gives every airline a common operating picture. Its power comes from coordination rather than flight, from being the neutral party each airline has to trust precisely because they do not trust each other.

That is the position Swift is defending. When a dozen banks each issue their own tokenised dollar, on their own ledgers, in their own jurisdictions, Swift's job is to be the one neutral place where those promises get recognised, sequenced and reconciled. Interoperability across institutions, in Swift's own framing, is the whole business: your token and my token, made mutually legible by a party we both already plug into.

This is why the moat might cross over after all. Swift's asset was never settlement. It was 11,500 institutions in more than 200 countries agreeing to act on the same messages, coordinated by a member-owned operator that no single bank controls. A permissioned shared ledger needs exactly that, a trusted convener with a full address book, and Swift is the rare party that arrives holding both.

The tell is in who is missing.

When Swift unveiled this at Sibos last September, the coalition ran past thirty banks, and later past forty, and it included JPMorgan, Bank of America and Deutsche Bank. The live pilot has seventeen. Those three are not in it.

The announcement does not say why the cohort narrowed from more than forty design contributors to seventeen pilot banks. One reading is readiness: connecting a live tokenised-deposit product is a harder ask than contributing to a design exercise. Another is strategic. JPMorgan already runs Kinexys, its own tokenised-deposit rail, moving more than seven billion dollars a day between its clients, and it may prefer its own network to become the standard rather than join someone else's first production cohort. The evidence is mixed, though. Citi and HSBC both run proprietary tokenised-deposit platforms of their own, and both joined Swift anyway. Having your own rail does not automatically mean you have no use for a neutral one.

Swift's model assumes money stays plural. Air traffic control only has a job because there are many airlines. If a handful of tokens, JPMorgan's deposit coin, a dominant regulated stablecoin, grow large enough that everyone simply holds them, the specific problem of reconciling many bank-issued monies gets smaller. You do not need a neutral convener to translate your token against mine once the market has quietly settled on one. Coordination itself does not disappear, identity, compliance, sanctions screening, liquidity and reconciliation still have to happen somewhere, but the part Swift is monetising here, the reconciling of plural monies, is the part that shrinks. That pressure comes from inside the tent, from the biggest banks, long before it comes from crypto.

So the risk is no longer only execution. Nine months to a working ledger and a seventeen-bank pilot is the opposite of a stalled consortium, and Swift walked in with a validated participant set and a chain that runs. But the harder questions, production volume, legal finality, liquidity commitments, operating rules and dispute resolution, have barely begun.

If the tokenised-deposit world consolidates, Swift is left coordinating the long tail while the whales settle among themselves and never touch the ledger. If it stays fragmented, Swift's convening role is safe, but fragmentation is also what keeps the settlement exposure and the delayed final leg exactly where they are. Swift's best case for its own network is also the case where the underlying system barely changes. That tension does not resolve. It is the business.

There is a governance question underneath it, narrower than the pilot's silence first suggests. The pilot banks are participants in the ledger rather than validators, since Swift currently acts as the single sequencer. That does not mean the banks lack ownership or governance. Swift is a member-owned cooperative, run through a bank-elected board and national member groups that have written its rulebook for five decades. The ledger's operating rulebook is therefore likely to develop through those existing cooperative structures, modified for the new infrastructure, rather than from a blank page, which is part of why seventeen of them were comfortable connecting.

That governance also sits inside Swift's existing oversight. Swift is overseen by the G10 central banks, with the National Bank of Belgium as lead overseer, the kind of scrutiny applied to systemically important financial infrastructure. What that machinery has not yet settled is one level down, on the ledger itself: what counts as an irrevocable funding commitment, when a payment becomes legally final, who bears the loss when settlement fails, and how new banks or new settlement assets are admitted. Those are the details permissioned networks live or die on, and it is there, not in the corporate governance, that Swift has so far said the least.

My prediction, within eighteen months, Swift will be running live 24/7 customer payments in at least one production corridor while final interbank settlement still clears primarily through RTGS systems, correspondent accounts or other off-ledger arrangements. Swift becomes the orchestrator of tokenised bank money without ever becoming the settlement asset itself. The signal that this is wrong will not be a bank simply claiming legal finality on the ledger. It will be an on-chain settlement asset, tokenised central-bank money or a commonly accepted commercial-bank liability, that extinguishes the interbank obligation with no downstream RTGS or correspondent leg. And watch whether JPMorgan ever joins: the day the biggest banks decide they need the neutral layer is the day the moat is real, and the day they decide they are the standard is the day it is not.

Stablecoins spent five years proving that a dollar-denominated asset could move without asking Swift. Swift spent nine months proving that banks could move the commitment around the clock while leaving the ultimate settlement asset exactly where it was.

The network that was supposed to be disintermediated has not put money on-chain. It has put the coordination of money on-chain.

Somewhere in Singapore, a treasurer is about to send money through Swift on a Saturday. The customer experience will be new. What settles the banks afterwards will not be.

See you next week.

James Smith