Every few months, Brussel produces another report explaining why Europe is falling behind in artificial intelligence, and every few months it reaches for the same comforting vocabulary: not enough talent, not enough graduates, not enough "ecosystem.". It is an elegant alibi because it avoids the humiliating truth. As a recent Bruegel analysis makes clear, Europe’s problem is not a shortage of brains or ideas. It is a shortage of compute and capital — the two things the continent has been unwilling to finance at the scale required.
Consider Mistral, the company Europe presents as proof that it still belongs in the AI race. Its flagship models were trained on Microsoft Azure. It opened an office in Palo Alto to attract Silicon Valley engineers and venture capital. Its attempt to build European sovereign compute has secured roughly €830 million for infrastructure and GPUs, barely a down payment in an industry now being shaped investment measured in hundreds, of billions. Europe’s AI champion still depends on American cloud power and still has to go to California to find the scale. https://www.bruegel.org/analysis/europe-needs-strategy-close-artificial-intelligence-compute-gap
The broader record is more brutal. Mario Draghi’s 2024 competitiveness report noted that no EU company founded from scratch in the previous fifty years had reached a market capitalisation above €100 billion. Half a century without producing a new corporate giant of that scale. And the failure is no longer confined to start-ups or frontier models. In a previous post, I examined the disturbing weakness of Germany in AI publication intensity: Europe’s industrial powerhouse, the country wrapped in the mythology of engineering supremacy, was already being outperformed by Portugal and several much smaller economies. The industrial verdict is now even more humiliating. In 2025, the United States produced roughly forty large foundation models, China about fifteen, and the entire European Union around three. https://www.atlanticcouncil.org/in-depth-research-reports/issue-brief/what-drives-the-divide-in-transatlantic-ai-strategy/
Then comes the figure that should bury the talent fairy tale: around 12% of EU artificial-intelligence PhDs relocate to the United States within five years. Europe trains the researchers, congratulates itself on the excellence of its universities, denies them the compute, capital and industrial scale needed to compete — and then watches them leave for the very ecosystem Europe claims it wants to rival. This is not simply a continent losing an AI race. It is a continent refusing to enter it on serious terms. Europe did not run out of talent. It starved talent of scale and exported it to its competitors.
PS — In August 2024, I noted that the companies producing the most highly cited AI research and patents were concentrated in the United States and China, with Europe already humiliatingly absent. What then looked like a warning now reads as a continental autopsy report. https://19-pacheco-torgal-19.blogspot.com/2024/08/the-list-of-companies-that-are.html
