Who owns the rails when machines move the money

In my last post I wrote about the big banks shutting their doors to outside AI agents. JPMorgan, Citi, Bank of America, Morgan Stanley: agents may look at your balance, but they may not move your money.

Here's the problem with closing a door: water doesn't stop, it finds another way. And there is already another way. On blockchain-based payment rails, AI agents pay for things today with no bank account, no card and no human pressing a button.

So the real question isn't whether banks let agents in. It's this: when machines start moving money at scale, whose rails will they run on? State money, bank money or private money? That race is on right now, and most people haven't noticed it started.

An AI agent doesn't fit into today's banking. It can't wait for a banking day. It can't pass a login built for a human. It needs money that moves the way data moves: instantly, around the clock, through code.

That is exactly what stablecoins offer. They are digital dollars or euros on a blockchain, and with the US GENIUS Act they got real legitimacy for the first time.

The machine economy is already running on them:

• Coinbase's x402 protocol revives an old web code, "402 Payment Required", so an agent can pay for a service inside the request itself. It has reportedly handled around 188–205 million transactions with roughly 69,000 active agents. No account, no card, no human approval.
• The stablecoin market passed $320 billion in value by mid-2026.
• Much of the use is invisible: the shopper sees a card, while stablecoins settle the money in the background, every day of the week.

Compare that with the roughly $19 trillion sitting in US bank deposits, and stablecoins are still small. But the direction is clear. What banks are blocking at the front door, crypto rails already do at the side door.

Strip away the jargon and three kinds of digital money are competing to become the rails of the agent economy.

Agents will drift to whatever is fastest, cheapest and least gatekept. Right now that is private money. The question is whether states and banks can offer something good enough before habits set.

The digital euro. In July the European Parliament voted 416 to 169 to open negotiations on it, aiming for a deal by the end of 2026. If the timeline holds, a pilot would start in mid-2027 and the first digital euros could arrive in 2029. The stated goal is to cut Europe's dependence on Visa, Mastercard, Apple Pay and Google Pay. Most businesses would have to accept it, and individuals would face a cap on how much they can hold, so deposits don't drain out of banks.

One point that often gets lost: the digital euro is not a cryptocurrency. It's central bank money, centrally run, closer to digital cash than to anything on a public blockchain.

Tokenized deposits. Banks aren't waiting. Their answer is to put their own money on a blockchain: programmable like a stablecoin, but still a bank deposit. HSBC and Ant Digital recently ran a test where an AI agent found a digital service, used it and paid for it on its own with tokenized bank deposits.

A bank-issued stablecoin. A consortium of major European banks, Qivalis, including BNP Paribas, ING, UniCredit and CaixaBank, plans to launch a euro stablecoin under the EU's MiCA rules in the second half of 2026. That is years before the digital euro. Europe's real payments race may be decided before the official currency even arrives.

Notice the pattern. Banks block outside agents in October and launch their own programmable money the same year. They aren't rejecting the machine economy. They want to own its rails.

Each major economy is betting on a different kind of money, and each bet mirrors its politics.

The pattern is striking. The US bets on private dollar stablecoins, which spread the dollar's reach worldwide, and has banned a state digital dollar. China goes the other way: state money, tight control, and private alternatives pushed out. Europe tries to do both, driven by one goal above all: less dependence on American payment companies.

The cross-border layer is where it gets geopolitical. China's mBridge platform, built with other central banks, settles payments directly between central bank currencies without the correspondent banking system, and the digital yuan makes up about 95% of its volume. The BRICS countries are discussing linking their digital currencies and payment systems to trade with less reliance on the dollar.

And for AI agents, the underdogs may be India and Brazil. Public, cheap, instant rails like Pix and UPI are far easier for software to use than a bank login. As columnist Andy Mukherjee noted, agents may well prefer exactly these state-backed networks.

If money moves onto blockchains, one more question becomes a banking question, not a hobbyist one: will these rails survive quantum computers?

Most blockchains are secured by elliptic-curve cryptography, which a powerful enough quantum computer is expected to break. Google's research this year cut the estimated hardware needed to do it roughly twentyfold. And the attack can start before Q-Day: an adversary can harvest public keys from a permanent ledger today and use them later.

None of the 20 largest blockchains has fully deployed quantum-resistant cryptography. Together they hold about $2.5 trillion. Here's where some of the most discussed ones stand:

Two things stand out. First, Stellar and XRP Ledger are built so that keys can be swapped without moving funds, an advantage for the coming migration. Second, Bitcoin, holding around $1.6 trillion, is the most exposed. Its problem is less the code than the decision-making: getting a decentralised community to agree on a migration in time.

Regulators are already setting deadlines. The US NSA requires quantum-resistant algorithms in all new national security systems from 1 January 2027. Financial rails won't be far behind.

This is a technical comparison, not investment advice. Quantum readiness says nothing about a coin's price.

In my architecture years I learned one rule that applies here perfectly: whoever owns the integration layer owns the system. Not the flashiest app, not the smartest feature. The pipes.

Money rails are pipes. Whoever owns them decides:

• who may pay: humans only, or software too
• what can be undone: card payments can be disputed, most blockchain payments can't
• who sees what: a bank, a central bank, a public ledger or nobody
• who earns from it: card fees, deposit margins or network fees

That's why this isn't a crypto-versus-banks story. States want monetary sovereignty and less dependence on US payment giants. Banks want to keep deposits and stay the gatekeeper. Private networks want to be the open road the machines prefer. All three are right about their own interests, and none of them is neutral.

And the agents themselves add a twist. Whichever rail wins, it will have to answer the same questions I raised about delegation in my last post: who is acting, what may it decide, who carries the risk when it goes wrong? A rail that answers those well is worth more than one that is merely fast.

My guess is that we won't get one winner. We'll get layers: state money as the anchor, bank money for everyday life, private rails at the machine-to-machine edge, with constant fights at the borders between them. The rails that will last are the ones that are programmable, quantum-ready and trusted by regulators all at once. Few are all three today.

So when the next headline says a bank blocked AI, or a coin jumped on a quantum roadmap, look past it. Ask who is building the pipes, and who will control what flows through them. That's where the real decisions are being made.

Note: this post reflects the situation as of 23 October 2026. Timelines for the digital euro, bank stablecoins and quantum migrations are still moving.

Henri Isola

The Code Breaker · Holistic & Business Coach

Twenty-five years as a Solution Architect and Senior Technical Expert in enterprise data centers and cloud migrations — IBM, HP, Sanoma, CGI. Now reading human systems instead. Start the series from the beginning: PHP2Shell and the Bank's Forgotten Side Door.