Huang is missing in Europe. The AI bubble offers no consolation for Europe. Draghi’s forgotten lesson

The temptation to interpret America’s difficulties as confirmation of one’s own caution obscures the real issue: insufficient venture capital, weak infrastructure and an ecosystem that rewards start-ups but rarely allows them to become giants
17 NOV 26
Translated by AI
Image of Huang is missing in Europe. The AI bubble offers no consolation for Europe. Draghi’s forgotten lesson

Photo: LaPresse

There is a European way of looking at the crisis in the American AI sector: to view it as reassuring. Nvidia is in the eye of the storm, the markets fear a bubble, and data centres are facing protests. So, people are saying, we were right to be cautious. A convenient consolation for lagging behind. You don’t need to defend Nvidia or exonerate Huang to realise this. Simply re-read the competitiveness report presented by Mario Draghi in September 2024. The analysis should wipe the smile off the faces of those waiting for the bubble to burst. In 2023, as Draghi recalled, venture capital investment in artificial intelligence stood at $8 billion in the EU, compared with $68 billion in the United States. There is no shortage of ideas: all too often, what is lacking is the opportunity to turn them into businesses capable of competing. Nvidia shows what becomes possible under different conditions. Three engineers set up a company in 1993 to improve video game graphics. They got their products wrong, changed direction, found investors and, in 2006, took a gamble on CUDA, a software platform that expanded the uses of their chips, years before the AI boom. This is not the fairy tale of an infallible entrepreneur. It is a system that allows for mistakes, second chances and patient investment. Europe needs more people like Huang: to have more companies capable of challenging his, not new untouchable leaders.
Draghi highlights what makes that path difficult: insufficient venture capital, fragmented markets, a laborious transition from research to products, and differing regulations that weigh particularly heavily on smaller players. We have researchers and savings; we struggle to invest them in ambitious ventures. It is easier to reward a start-up than to let it grow into a giant.
And then there are the data centres. AI does not work on the basis of declarations of principle: it requires processors, computing power, networks and energy. Draghi calls for supercomputers to be made available to businesses and points out that, at the time the report was published, industrial electricity costs were two or three times higher than in the United States. Data centres must cover their own costs, comply with environmental constraints and not add to other people’s bills. However, managing their growth is different from celebrating every project that is blocked. If we give up on building infrastructure, we will end up buying even more computing power from those who build it elsewhere. If Nvidia is too dominant, the answer is more competition: other chips, interoperable software, new suppliers, and greater freedom to switch platforms. Not the relief of a shrinking market. Draghi highlights Europe’s dependence on hardware and models developed outside the EU. The remedy cannot be to invest less.
Could there be a bubble? Certainly. Not all data centres will pay for themselves, not all promises will be kept. That is only right: venture capital must be able to take losses, managers must be held accountable for their mistakes, and overvalued companies must see their share prices fall. But a financial correction does not prove that the technology is useless. The risk for Europe is not that an American bubble will burst. It is that, whilst waiting for it to burst, we convince ourselves that the delay was prudence, the absence of technological leaders was wisdom, and the difficulty in authorising infrastructure was foresight. Draghi has explained to us why this approach does not work. The correction on Wall Street may be healthy; the satisfaction of those who did not invest is merely an excuse. Before saying ‘we were right to hold back’, let us ask ourselves why revolutions are so often built and financed by others.