The Wrong Race
Everyone asks whether Europe has lost the race to build artificial intelligence. That is the wrong race to be watching.
On the last day of his first official visit to China, in February, Friedrich Merz watched two humanoid machines stage a martial arts demonstration at a robotics company in Hangzhou. Around him stood executives from some thirty German companies, the head of Siemens among them. The chancellor of Europe’s largest industrial economy had come to see what a company called Unitree could now build, and he left impressed, which was the honest reaction and also the revealing one.
The scene is usually read as a verdict. Europe has fallen behind, and even Germany knows it. That reading is not wrong, but it answers a question that has been put badly. Everyone asks whether Europe has lost the artificial intelligence race, and measured one way, the answer is plain. The European Union’s flagship plan aims to mobilize some 200 billion euros for artificial intelligence by the end of the decade. Five American companies expect to spend around 700 billion dollars on AI infrastructure in 2026 alone. No roadmap closes a gap of that size. If the race is to build the most powerful models and the chips beneath them, Europe is not behind so much as in a different category.
But that is one race, and it is not self-evident that it matters most.
···
The economic value of a technology rarely comes from inventing it. It comes from spreading it through the economy, from the unglamorous business of putting it to work in factories, hospitals, offices, and supply chains. A model that no one uses changes nothing, while an ordinary tool deployed everywhere can reshape an economy. The prize in artificial intelligence will go disproportionately to whoever applies it widely, rather than to whoever announces it first.
On that measure, Europe’s position looks different and considerably less hopeless. Western European factories use more industrial robots per worker than North American ones, and Germany is the third most automated economy on earth, behind only South Korea and Singapore. This is not a continent that fears machines that replace human labor. It has spent four decades installing them, for the most part by negotiating their arrival with its workforce rather than imposing it. The picture of a timid, technophobic Europe does not survive contact with the data.
The same shift is reaching the office. The share of European companies using artificial intelligence climbed from 13.5 percent in 2024 to 20 percent a year later, and in Denmark, Finland, and Sweden, it now runs above a third. Most of that is the ordinary cognitive work of a service economy: sorting text, handling customers, drafting, and searching. Where Europe leads, it does so on the strength of trained workforces, dense digital infrastructure, and a short distance between a firm and the skills it needs, not on lighter regulation.
Why this matters more for Europe than for anyone is demographic. A shrinking workforce can sustain its pensions and living standards only by producing more from each remaining worker, which is what applied artificial intelligence delivers. Europe needs that productivity more than the younger United States, and far more than India, which has workers to spare and the opposite problem. The continent with the weakest demographics has the strongest reason to be the world’s most aggressive adopter.
···
So Europe’s problem is one of attention. It is pouring scarce political energy into the race it cannot win while neglecting the one it might. In June, the European Commission unveiled a sweeping package to build sovereign European clouds, chips, and models, and to reserve sensitive public contracts for European suppliers. The instinct behind it is to catch up in inventing, to wall off a market and raise champions inside it. Almost none of it addresses the duller and decisive task of closing the gap between firms that already use these tools and those that do not, or between large companies and small.
The competition there is real. China is aging too and has made deployment a national project, and 87 percent of the humanoid robots delivered last year were built there. Japan, older still, turned to automation early and deliberately, in large part to offset a workforce it knew would shrink. The track Europe could plausibly run is not empty, and it is the one whose prize Europe actually needs.
Merz was right to be impressed in Hangzhou. The machines were genuinely remarkable, and they were Chinese. But the factories most ready to put such machines to work are still in Germany, and the economies that most need what those machines produce are still in Europe. The danger is not that Europe cannot compete, but that a continent which needs the productivity more than anyone has decided to prove it can build the machines instead of using them.
THE VERDICT
Europe is making a category error it can still correct and a sequencing error it might not. Two hundred billion euros aimed at chips and models will not close a gap measured in the hundreds of billions, and chasing it mainly advertises the chase. Diffusion offers no equivalent spectacle. There is no flagship to fund, no national champion to photograph, no single contract that proves a government is serious about it. Adoption happens unevenly, firm by firm, mostly invisible to anyone outside the firm itself, and that invisibility is exactly why Brussels keeps choosing the other race. A political system built to reward leaders who can point at something prefers the contest that produces a building over the one that produces a percentage point of productivity nobody can put a ribbon on. The continent with the most urgent demographic case for adoption is behaving like one that still has time to invent its way out, and it can keep affording that mistake for exactly as long as nobody asks what the 200 billion was actually supposed to fix.
This piece is adapted from “The Wrong Race,” published in Rajasthan Patrika on 16 June 2026 as part of the monthly “Letter from Europe” column. Rajasthan Patrika, one of India’s largest Hindi-language daily newspapers, carries the column on a monthly basis.