Every form of money is a claim on somebody. Central bank money is the only one where that somebody issues the currency, and on a trading desk the difference is not academic: it shows up in your credit lines, your capital, and the length of the chain your payment has to survive.
Start with the definition, because most published writing on this subject blurs four different things into one.
Central bank money is a claim on the central bank itself. Reserves, balances held in a real time gross settlement system. The issuer of the claim is the issuer of the currency, and cannot run short of it.
Commercial bank money is a claim on a commercial bank. Your deposit, your nostro balance, the cash leg of most of what settles in wholesale markets today. It behaves like central bank money because it is redeemable at par into central bank money on demand, and that redeemability holds almost all of the time.
Tokenised deposits are commercial bank money with a new transfer mechanism. The ledger is different. The claim is identical: it is still an obligation of the issuing bank, carrying that bank's credit and liquidity risk. Tokenisation changes how the claim moves. It does not change who owes you.
Stablecoins are a claim on a private issuer against a pool of reserves. Their quality is a function of what is in the pool, what redemption rights exist in law, and whether those rights are enforceable at speed under stress. Some are conservatively constructed. None of them is a claim on a central bank, whatever they hold.
The distinction that matters is not technological. It is the answer to one question: when you are holding this at four in the afternoon, who are you exposed to?
Not safer in the sense of better engineered, but safer in the specific sense that the settlement asset itself carries no credit risk. The receiving party does not have to form a view on the paying party's bank. Finality is defined in law rather than in a contract between intermediaries.
And this is not a novel opinion: Principle 9 of the Principles for Financial Market Infrastructures directs financial market infrastructures to conduct money settlements in central bank money where practical and available. The international standard has said so for well over a decade. The interesting question has never been whether central bank money is the right settlement asset. It is who can reach it, and what it can be coordinated with when they do.
Four things change when the cash leg settles in central bank money rather than in a commercial bank's books.
Intraday credit. If you deliver an asset at eleven and the cash arrives at four, somebody has extended credit across that window. That credit is priced, or collateralised, or it is a limit that quietly constrains what the desk can do after lunch. It is rarely invoiced and it is never free.
Capital. An exposure to a settlement bank is an exposure and it is counted. Unsettled transactions and free deliveries attract capital treatment that escalates the longer they stay outstanding.
Chain length. Every intermediary in a payment path is a place the payment can stop, a party that must be reconciled with, a cut off time in another jurisdiction, and a fee. Shortening the chain removes all four at once, which is why the chain is the variable worth attacking rather than the speed of any single hop.
Timing across currencies. The classic settlement loss in this industry happened because two legs of the same trade settled in two jurisdictions whose banking days did not overlap. Payment versus payment arrangements exist precisely because that window was expensive enough to build an institution around.
This is not abstract for us. On desks in London, Madrid, New York and Geneva, the cost of settling in commercial bank money never arrived with a label on it. It arrived as a cut off that made a perfectly good trade impossible after a certain hour. It arrived as a funding buffer that had to sit somewhere doing nothing. It arrived as a limit against a name we were otherwise happy to face. It arrived as an operations team spending an afternoon establishing where a payment had reached. Nobody ever sent an invoice for settlement risk. It was collected as friction, and friction does not appear as a line item, which is exactly why it survived thirty years of technology spending.
Now the honest boundary, because this is where most writing on the subject becomes promotional.
Central bank money settlement is not new and it is not scarce. Real time gross settlement systems have settled enormous wholesale volumes in central bank money for decades. What is genuinely in motion is two narrower things: who is eligible to hold an account, and whether settlement in that account can be conditioned on something happening on another ledger.
On access, the United Kingdom has moved furthest among the large jurisdictions. The Bank of England has admitted non bank payment service providers to settlement accounts since 2017, introduced omnibus accounts for recognised payment system operators in 2021, consolidated its access policy in 2025, and launched its renewed real time gross settlement service, RT2, in April 2025. Eligibility is still a decision the central bank makes, institution by institution, against published criteria. It is not a technology gate.
On conditionality, the Bank of England's Synchronisation Lab began in 2026 with eighteen selected organisations, testing arrangements in which funds in RT2 settle if and only if an asset settles on an external ledger, orchestrated by a new kind of participant called a synchronisation operator. In the euro area, the Eurosystem is running a two track programme: Pontes, a solution linking distributed ledger platforms to TARGET Services so that transactions recorded on those platforms settle in central bank money, with an initial launch planned for the third quarter of 2026, and Appia, a longer term track.
Internationally, the Bank for International Settlements published the prototype report for Project Agorá in May 2026, a public and private collaboration spanning seven currency areas and more than forty regulated institutions, demonstrating tokenised commercial bank deposits settling atomically against tokenised central bank reserves on a shared platform. In early August 2026 the project published the results of real value testing: five central banks, twenty eight commercial banks, thirty transactions across six currencies, settling in an average of approximately eighty seconds despite no direct integration with existing settlement systems.
Keep that last number in proportion. The reported pilot moved on the order of one million dollars. Cross border flows run to hundreds of trillions a year. The direction of travel is now established by the institutions that operate the currencies themselves, which is what makes it credible. The scale is not established at all, and the BIS is explicit that the work remains experimental with no production timetable.
Central bank money removes the credit risk of the settlement asset. It removes nothing else. It does not remove market risk, it does not remove the need to assess the counterparty you traded with, and it does not make a badly designed venue safe.
Gross settlement in central bank money can consume more liquidity than netting through a correspondent you trust. The chain we criticise is also a liquidity saving machine, and a design that ignores that will be rejected by the treasurers who actually pay for intraday funding. The answer is netting before settlement, not pretending the trade off does not exist.
And access is granted, not taken. IDBX does not hold a central bank account, has no agreement with any central bank, and is not authorised. Our architecture is designed for central bank money settlement on the cash leg, currency by currency, as access is granted by the relevant central bank. We will describe the design in as much detail as anyone wants. We will not put a date on somebody else's decision.
The safest money in the room has always existed. What has changed is that the institutions issuing it have started building the doors.