France Courts AI With Power and Scale

France’s AI Pitch Moves From Models to Megawatts

France is making an increasingly direct bid to become a home for the physical backbone of artificial intelligence in Europe, pairing a diplomatic charm offensive with some of the continent’s largest proposed data-center investments.

President Emmanuel Macron has invited OpenAI’s chief executive, Sam Altman, to attend the Group of 7 summit France is hosting in Évian-les-Bains on June 15 to 17, according to OpenAI. The invitation comes as Paris presses a broader case to global technology companies: that France can offer not only political access and market ambition, but also one of the ingredients now most prized in AI — large amounts of electricity.

That strategy was underscored days earlier at the government’s “Choose France” summit, where SoftBank said it planned an initial €45 billion investment to build 3.1 gigawatts of AI data-center capacity in the northern Hauts-de-France region by 2031. The longer-term plan, according to French officials, could rise to €75 billion and as much as 5 gigawatts nationwide. The Élysée Palace has said the aim is to develop and operate at least 3 gigawatts and as much as 5 gigawatts of AI-dedicated data-center capacity in France.

Taken together, the moves point to a shift in Europe’s AI ambitions. The contest is no longer only about which region produces the best models or attracts the most start-ups. It is also about who can secure land, transmission lines, permits, cooling systems and low-carbon power for the warehouses of chips that increasingly determine who can build and run cutting-edge AI systems.

A Different Kind of AI Competition

For Mr. Macron, the appeal to Mr. Altman is consistent with a campaign he has been waging for more than a year: persuading investors and executives that France should be seen as a practical base for AI expansion, not merely a regulator of technologies invented elsewhere.

France used the AI Action Summit in Paris in February 2025 to argue that Europe should take a more investment-friendly stance toward artificial intelligence. OpenAI, which opened a Paris office in 2024, has also singled out France in its international expansion. At the European level, Ursula von der Leyen, the president of the European Commission, said in February 2025 that Europe would mobilize €200 billion for AI investment.

But if Europe has often talked about digital sovereignty in terms of software, chips and rules, the latest French push is more elemental. It is grounded in the proposition that AI’s future will depend as much on power grids as on code.

France believes it has a case to make. Its electricity system, backed heavily by nuclear power, gives it a comparatively abundant source of low-carbon energy at a moment when AI companies are scouring the world for places that can support enormous and continuous electricity demand. In that sense, France is selling not only policy stability, but electrons.

The Power Question Comes to the Fore

SoftBank’s proposed buildout in Hauts-de-France illustrates the scale of that demand. A 3.1-gigawatt project would be among the largest announced AI data-center bets in Europe, and one that would test the capacity of local and national infrastructure alike.

Across the continent, the AI boom is forcing a difficult reckoning. Data centers designed for training and running advanced models consume far more power than traditional computing facilities, and their needs can arrive in concentrated bursts that strain grids, substations and transmission networks. Securing enough electricity is quickly becoming as important as securing capital.

That presents Europe with both an opening and a risk. On one hand, countries with available land, relatively clean power and industrial partners can try to capture more of the AI supply chain. France has promoted that possibility by pairing foreign capital with domestic companies such as EDF, the utility giant, and Schneider Electric, a major maker of electrical equipment and industrial systems.

On the other hand, the surge in announced projects raises difficult questions about whether grids can be upgraded quickly enough, whether permitting can keep pace and whether all the promised capacity will ultimately be built. Multi-gigawatt AI campuses require not just financing, but years of coordination across utilities, local authorities, chip suppliers and cloud customers.

What Europe Stands to Gain — or Lose

The deeper issue for Europe is economic as much as technical.

For years, European policymakers have worried that the region risks becoming a market for AI services developed elsewhere while the highest margins accrue to American technology companies and the dominant suppliers of advanced chips. Hosting more data centers could help Europe claim a larger share of the value created by AI, through construction, operations, power sales, engineering work and, potentially, a stronger local ecosystem of suppliers.

Yet hosting infrastructure does not automatically translate into strategic autonomy. A central uncertainty is whether Europe will own and control enough of the higher-value layers — cloud services, model development, semiconductors and platforms — or whether it will mainly provide sites and energy to foreign firms.

That question hangs over the French push. If projects like SoftBank’s move ahead, they could create jobs and cement France’s position as a hub for AI compute. But they could also reinforce an arrangement in which Europe supplies the physical inputs while non-European companies capture much of the commercial upside.

Why the Timing Matters

Mr. Macron’s invitation to Mr. Altman lands at a moment when governments are treating AI infrastructure less as a narrow tech issue than as a matter of industrial policy and geopolitical positioning.

The public details of Mr. Altman’s role at the G7 remain unclear. Still, the symbolism is notable. Inviting the head of OpenAI to a summit of major industrial democracies suggests how closely AI is now tied to national competitiveness, energy planning and diplomacy.

For France, the message is that Europe should not be resigned to playing a secondary role in the AI economy. It can try to host a larger share of the stack — from power generation and equipment to data centers and, eventually, model deployment.

Whether that ambition can be realized will depend on more than summit invitations and headline investment figures. It will require turning political momentum into substations, permits, transformers, transmission capacity and customers willing to commit to vast computing loads.

In Europe’s AI future, the bottleneck may be less imagination than infrastructure. France is betting that if it can solve for the latter, the rest of the industry will follow.