Beyond the Hegemon

The morning after a reception in Rome, I am still turning over a conversation I had deliberately started. An African ambassador, a glass in his hand, and a question I had chosen to make blunt: do you agree that your country will never spend five hundred billion dollars on artificial intelligence, or anything close to it? He agreed at once. He was not offended; nobody from a small country ever is. The person who took offence that evening, when the phrase "middle power" came up, was the German. He is not, he pointed out tactfully, the same size as Ghana.

He is right about the size. He is wrong about the category, and the distance between those two facts is this episode.

A middle power is any country that is not the United States and not China. That includes Germany. It includes the European Union taken as a whole. It includes Canada, a G7 country whose prime minister stood up in Davos in January and used the phrase about himself. Mark Carney called the state of the post-war order "a rupture, not a transition," and described great powers "using economic integration as weapons, tariffs as leverage, financial infrastructure as coercion, and supply chains as vulnerabilities to be exploited." The discomfort the word causes in Berlin is not an argument against it. It is the symptom the word was coined to describe.

The commitments being made to AI infrastructure in the United States, in a single year, are on a scale that the whole of Europe does not approach. No country in the middle closes that gap by spending, and every national strategy premised on catching up is a strategy premised on losing slowly. Maria Ressa put it to the ministers in Geneva in July: "most of your countries cannot yet test these systems, audit them, or govern them on your own terms. That is not a failure of will. It is a structural fact this report documents."

What changes it is the plural. There is an "s" on the end of middle powers, and that letter is the whole argument. Ghana with Canada with the Holy See with Hungary is not a sentiment. It is the only combination in which the numbers work. A Ghanaian speaker on the Vatican panel gave it the older name, Ubuntu: I am because we are.

And there is an instrument already in the room. Public authorities in Europe buy roughly fourteen per cent of GDP every year, more than two trillion euros. Nobody in the middle can match a hegemon's capital spending, but everybody in the middle has a procurement budget, and coordinated demand is the one weapon that does not require matching anyone's spending. Europe has even built the vehicle for pooling money across governments. What it has not yet done is buy as a bloc.

Now the objection, and it is the one I would make myself. Independence is easy to call for and brutally hard to build. Coalitions of the not-quite-powerful fail in two ways. The first is that they become a vassal to a new hegemon: a bloc organised around one dominant member's champions, which replaces a dependency on Seattle with a dependency on Paris. The second is that they become a talking shop, producing communiqués and frameworks while the capacity gap widens underneath them. I have sat in rooms that were producing the second outcome in real time. Some of them were in Rome.

The difference between those outcomes and a real one is not sincerity. It is whether the coalition builds shared infrastructure or shared positions. Positions are free, which is why they get produced. Infrastructure costs money and sovereignty in roughly equal measure, which is why it does not: shared compute, data held under each member's own law, models they can retrain, procurement that commits them to each other. A coalition that owns nothing together will not hold together on the day a hegemon makes one of its members an offer.

The encyclical supplies the test. Justice, it says, "requires questioning the global distribution of power that decides who in fact can train these models and who is merely subjected to them." A coalition is worth the name only if it moves its members from the second group into the first.

When it was my turn to close the Vatican panel, I said that the race for the big engines had already been won, and that this was the beginning of the argument, not the end of it. The middle countries can still build something different: a third rail, which does not yet exist. There is a wish for it in every capital I have visited this year, and not yet any momentum. It will not come from the countries that can spend; they do not need it. It will come from the people who can convene. That is the assignment.

Previously: Data Embassies. Next Tuesday: The Automation Trap.