Nvidia’s CEO Jensen Huang spent last week in Europe making a simple pitch: every country needs its own AI. And Europe’s leaders, tired of tech dependency on the U.S., are buying in.

‎

The idea of “sovereign AI”—where each nation builds its own language models and data infrastructure—has long been championed by Huang. Now, the EU is treating it as a strategic priority.

‎

In London, Paris and Berlin, Huang unveiled partnerships with telecoms, startups and national governments. He warned that Europe is moving too slowly and risks missing the AI wave. “We are going to invest billions in here,” he said in Paris. “But Europe needs to move into AI quickly.”

‎

UK Prime Minister Keir Starmer pledged £1 billion to boost computing power. French President Emmanuel Macron called AI infrastructure “our fight for sovereignty.” And Germany’s Chancellor Friedrich Merz endorsed Nvidia’s plan to build a cloud platform with Deutsche Telekom, calling it vital to his country’s digital future.

‎

For Nvidia, this is about locking in demand. Its AI chips are already essential for data centres worldwide. If Europe builds “independent” infrastructure, Huang wants it built on Nvidia hardware.

‎

Europe trails far behind the U.S. and China. Most of its cloud services still rely on Amazon, Microsoft, and Google. Local champions are few and far between. One exception is French startup Mistral, whose 31-year-old CEO Arthur Mensch sat beside Huang at VivaTech and called European AI “a gigantic dream.”‎

‎

But Mistral’s dream needs muscle. It will partner with Nvidia to build a French data centre using 18,000 of Nvidia’s latest GPUs—one of the largest of its kind. It’s the first phase of a broader rollout set to expand in 2026.

‎

The European Commission is also on board. It has announced four planned AI “gigafactories” costing €20 billion. According to an EU official, Nvidia has agreed to allocate some chip production to Europe for the project.

‎

In theory, everyone stands to gain. Sovereign AI could jumpstart homegrown cloud providers, AI developers like Mistral, and even European chip manufacturing. In practice, Nvidia wins either way. Even as Europe aims for “independence,” the hardware still comes from Santa Clara.

‎

There’s a catch. Energy. Data centres already consume around 3% of the EU’s electricity. That figure is set to rise sharply as AI adoption explodes. And Europe’s energy costs are far higher than in the U.S.

‎

Mistral may be a homegrown hero, but it has raised just over $1 billion—chump change next to U.S. hyperscalers spending $10–15 billion per quarter. “Who in Europe can afford that exactly?” asked Capgemini’s innovation chief Pascal Brier, a partner to both Nvidia and Mistral.‎

‎

Even if Europe builds the infrastructure, it may never close the spending gap.

‎

Mistral has launched its own AI models, which are already used by businesses. But companies rarely commit to a single provider. “Most of the time it’s not Mistral or the rest,” Brier said. “It’s Mistral and the rest”—with OpenAI, Anthropic and Meta often in the mix.

‎

So Europe’s AI revolution may be sovereign in theory—but multinational in practice. And Nvidia will likely supply the chips for both.

‎

‎