The ECB launched Pontes on 21 September 2026 — a permissioned ledger for tokenised securities settled in central bank money. The 126-page service description names the whitelist, the issuer node that mints, and the certificates that replace proof-of-work. The vocabulary is decentralisation; the architecture is a gate.
On 21 September 2026, according to its own press release, the European Central Bank switched on Pontes. The name is Latin for bridges, and the bridge is real: it connects private distributed-ledger platforms to TARGET, the Eurosystem's settlement system, so that tokenised securities can be paid for with central bank money. The same morning the ECB announced it will invest part of its own funds in tokenised bonds, settled through that same platform.
The press release reads like a technical footnote. That footnote is wrong. This is the moment the largest issuer of a fiat currency built itself a private blockchain, restricted to banks, and called it innovation.
Start with what Pontes actually does, because the detail is where the argument lives.
A tokenised bond is a bond whose ownership is recorded on some ledger rather than in a central securities depository's database. The interesting problem has never been the asset — it has been the cash leg. Pay for a digital bond with a commercial bank token and you have reintroduced the credit risk of that bank into a settlement that was supposed to be final. Pay with a commercial stablecoin and you have done the same thing with a worse balance sheet and no lender of last resort.
So the Eurosystem did the honest engineering: it put its own money on the ledger. The cash leg settles in T2, the real-time gross settlement system, and settles only when the transaction completes there. They call this settlement finality in central bank money. Delivery versus payment is handled with a hash-link protocol, so both legs either complete or neither does.
The service description — 126 pages, published by the ECB — is more explicit than the press release. It names the architecture: an Extended Interoperability Interface as the single access point, the Eurosystem DLT platform handling the cash leg, and a T2 interface as the single entry to TARGET services through ESMIG. The ledger itself consists of central bank nodes, one per national central bank, an operator node that manages network parameters and the participant lists, and an ECB issuer node that mints and redeems the cash tokens.
Read those four components and notice what they are. This is not a peer-to-peer network. It is a client-server system wearing a blockchain's vocabulary, hosted by a service provider, with four kinds of privileged nodes and a central issuer whose token supply mirrors an account in the old system, cleared to zero daily.
The definition of the cash token in the ECB's own glossary is worth quoting in full, because it settles several arguments at once.
A cash token is a proxy of euro-denominated central bank money, representing a claim against the token issuer (that is, the ECB) to transfer the respective amount of central bank money in T2 to the account of the market participant that owns such tokens — without legal tender status.
A proxy. A claim against the issuer. Explicitly not legal tender. That is not my characterisation; it is the Eurosystem describing its own instrument. The digital euro for retail is a separate programme with open political questions. Pontes is wholesale, and it demonstrates the technique in the place where nobody has to vote on it first.
Then follow the funding mechanics. To obtain cash tokens, a participant's T2 account is debited and the corresponding wallet is credited — the glossary uses the word minting. To get out, the wallet is debited, which the document calls burning, and the T2 account is credited. Funding and defunding are requests to the Eurosystem, executed by its issuer node.
A participant may hold balances at the central bank and request their conversion into tokens. That is a central bank digital currency account. The only things standing between this and the thing people spent years being told was a conspiracy theory are the words wholesale and pilot.
Who may participate? The service description answers without ambiguity: any entity with access to T2. Who may operate a platform? Central securities depositories authorised under the CSDR, including T2S depositories; operators of a DLT settlement or trading and settlement system authorised under the DLT Pilot Regime; operators of a payment system established in the EU or EEA; central counterparties authorised under EMIR; credit institutions.
The document then goes further and dedicates a section to whitelists — in the plural, with participant blocking and unblocking as neighbouring subsections. There is also a configuration called instruct on behalf, which lets one actor act for another.
A whitelist is not a technical constraint. It is the administrative act that replaces proof-of-work with permission, and the Eurosystem writes the process down. Where Bitcoin settles by expending energy on a puzzle that anyone can attempt, Pontes settles by appearing on a list that a node operator maintains. The gate is the architecture.
Now the security argument, and I want to be precise rather than loud, because the loud version of this is wrong and easy to refute.
It is not that Pontes will necessarily be hacked. It is that its trust model is categorically weaker than Bitcoin's, and the difference is structural rather than a matter of engineering effort.
Bitcoin has no privileged nodes. No operator node managing parameters, no issuer node that can mint, no list a node maintains. Validation is open: anybody can run a node, nobody needs permission, and the rule is whatever the majority of unconnected participants accept. To compromise it you must take over a majority of that open set, which is expensive, visible and temporary.
Pontes has four kinds. Central bank nodes validate, sign and store. The operator node manages network parameters and the ECB, NCB and market participant lists. The ECB issuer node creates and destroys the tokens. Compromise that issuer node and you do not bend a rule — you mint claims against the central bank. There is no consensus to defeat because there is nothing to reach consensus with: the privileged node is the authority, not a participant in a vote.
The identities are certificates. Access runs over X.509 certificates with a certificate authority, certificate signing requests, a documented lifecycle and the ECB's own guidance on key hygiene. The document suggests storing the private key in a Hardware Security Module, restricting file permissions, and optionally passphrasing the key file. Sensible recommendations for an enterprise system — and a description of exactly the attack surface Bitcoin does not have: a certificate authority whose compromise lets an attacker impersonate any participant, and a private key that lives in a file or a module on somebody's premises, where it can be copied, phished or subpoenaed.
The infrastructure is rented. Hosting and resilience are handled by a service provider across regions, with containerisation and automatic recovery even through a full availability-zone failure. Good engineering for availability — and a small number of data centres operated by named suppliers, with a supply chain, a staff and a physical address. Bitcoin's equivalent has no address and no staff.
And the dependencies are visible in the document itself. Settlement finality is defined by the TARGET Guideline, the DLT Pilot Regime has a sunset, access runs through ESMIG, and the legal framework is doing the work that a proof does in Bitcoin. The ECB even notes that the pilot operates outside the technical perimeter of TARGET Services and requires no changes to their existing legal, regulatory, functional or operational framework. Read that twice: the new architecture is explicitly bolted on top of the old one rather than replacing it.
One candid admission is worth keeping. The service description concedes that delivery-versus-payment is exposed only in machine-to-machine mode and not available to national central bank users, so an actor who loses that connection cannot continue the flow through the interactive channel and cannot delegate to its central bank to act on its behalf. A central bank network where a broken connection means nobody can step in is not a description of resilient money. It is a description of middleware.
None of this means Pontes will fail. Large, well-funded infrastructure can run for decades without being broken. It means the safety comes from the institutions running it, not from the design — and that is a promise, not a proof. Bitcoin's guarantee costs energy and can be verified by anyone with a laptop. The guarantee here costs trust and can only be checked by an auditor.
Here is where I have to be careful, because the loud version of this argument is dishonest and I would rather be right than loud.
Pontes is not a fraud. Nobody is stealing. The bonds are real, the settlement is real, the legal certainty is real, and in several narrow ways this genuinely improves on the status quo — atomic delivery-versus-payment between two separate ledgers is a real problem that existing infrastructure handles badly, and reducing the reconciliation staff at a custodian is a real saving.
And when the ECB writes that tokenisation is not about speculation or private money but about modernising Europe's market infrastructure, it is describing its own position accurately. Brussels wants a capital markets union, it wants European settlement to be less dependent on American plumbing, and it wants the euro to sit at the centre of whatever comes next. Those are coherent institutional goals, not a conspiracy.
The scam is structural, not moral. This is the substitution.
They take the one word that made this technology dangerous to them and reattach it to something that is not. Distributed means no single operator. Permissionless means no whitelist. Immutable means no authority that can rewrite. Sound money means nobody who can print. Pontes keeps the vocabulary, removes the properties, and files the result under the same label. Then every regulator and journalist repeats the label, and a reader who hears "the ECB is adopting blockchain" reasonably concludes the ECB is surrendering some control. The reality is the opposite: extending its control into a domain that was invented specifically to escape it.
For thirty years the monetary system has run on a settlement layer that is slow, expensive, and closed — TARGET settles in central bank money during Frankfurt business hours, and everything else is interbank promises stacked on top. The cost of that system is the business model of a thousand intermediaries — the European Commission's own estimates of cross-border settlement costs ran into the tens of billions of euros a year before it stopped publishing them — and the opacity of it is what allowed 2008.
Pontes fixes the speed. It fixes the atomicity. It does not touch the gate, and it hardens the position of the money at the centre of it. Every efficiency gain in this design flows to the institution that issues the settlement asset, because the settlement asset is now programmatically required by the infrastructure. That is not a side effect. That is the point of building it.
Consider what is absent. There is no way for a person to hold the cash token. Not a citizen, not a small company, not a foreign institution — the eligibility list runs to central securities depositories and credit institutions, and stops.
For now that is a limitation. Watch what it becomes. The stated purpose includes preserving the anchor role of central bank money, which means the token is designed to remain the unit everything settles in. The mechanism for widening access is the eligibility list, and eligibility lists are administrative. Nothing in the architecture prevents the perimeter from expanding from banks to large corporates to platforms to, eventually, the public — and everything in the architecture is built to make that expansion a configuration change rather than a new system.
That is the path from a settlement pilot to a programmable currency with a permission layer. It does not require a single dramatic decision. It requires only that the list gets longer, in steps small enough that each one is presented as a technical improvement.
And note what the instrument already is, in the ECB's own words: a proxy of central bank money, a claim against the issuer, without legal tender status. A claim held in a wallet the issuer can freeze, on a list the operator maintains, settled by nodes the central bank runs. That is the design of a CBDC. The word wholesale describes who is allowed in today, not what the thing is.
Now put the two facts from the same press release next to each other:
The ECB is building the infrastructure for tokenised assets, and the ECB is buying tokenised assets. One institution is simultaneously the rule-setter, the operator, the settlement provider and a market participant. That is not an accusation from a Bitcoin blog; that is the plain reading of a document the ECB published about itself. They call it building institutional expertise. A more direct name for it is a closed loop.
I am not going to end this by telling you the euro is dying and Bitcoin wins tomorrow. That is the cheap version.
The useful observation is that the ECB has now publicly admitted what the technology is for. The wording is unambiguous. No mention of a gimmick, no dismissal of the technology. The claim is the future of market infrastructure is programmable ledgers settled in a digital native unit — and then it built one where it decides who may connect.
Every institution that spends three years learning this stack is training its staff, its auditors and its regulators on a mental model that Bitcoin had for seventeen years: bearer assets, cryptographic finality, no reconciliation. The people at those institutions are not stupid, and some fraction of them will notice that the properties the ECB deliberately left out are the ones that make settlement cheap in the first place. The whitelist is not free; it costs a compliance department, an onboarding process, and a permanent dependency on whoever owns the list.
That is the door. It is narrow, and Pontes is on the other side of it by design.
Three things, none of them dramatic, all of them testable.
How much volume the ECB actually does. The own-funds programme is described as a small portion, with operational details and timing still to be decided. If tokenised bonds stay a rounding error two years from now, the experiment failed and the plumbing was the reason.
Whether the eligibility list widens or narrows. The DLT Pilot Regime is currently capped and temporary. If it is extended and the caps lifted, the structure is being made permanent. If it is left as a sandbox with a sunset clause, the ECB is hedging.
Whether settlement finality remains in T2. This is the technical heart. The entire claim of central bank money rests on the last word being written in the old system. If that ever moves onto the DLT platform — if the ledger becomes the ledger — the bridge has become the destination, and the architecture stops being an interoperability layer and becomes the monetary system itself.
A digital euro settled on a permissioned ledger, restricted to banks, with a central bank buying the assets it settles, is the safest possible way to give an institution absolute control over money while telling everyone it modernised. It is also, genuinely, better plumbing than what Europe has now. Both are true. The scam is not in the engineering; it is in the word. They kept the vocabulary of freedom and removed every property that made it real, and then they published the whitelist and expected nobody to read it.
Bitcoin does not need Pontes to fail. It needs people to notice that the bridge only runs in one direction.