France is betting that its cheap, low-carbon electricity is the overlooked AI advantage for Europe. And now it is forcing a real decision out in the open: who gets to draw from that resource as European AI demand accelerates. The debate is blunt. Should that power go to homegrown AI firms, or should it flow to the American giants building data centres on French soil?

The stakes are not theoretical. French policymakers are trying to set the “rules of the runway” for AI expansion, at a moment when data-centre capacity and electricity supply can become the gating items for training and deploying models. In that context, France’s position is simple: low-carbon power is a strategic differentiator, and the allocation of that power is now part of national and European competition. The Next Web notes that the head of Europe’s biggest AI lab appeared at last month’s G7 working lunch on artificial and related topics, which underscores how quickly this issue is moving from industry preference to government-level agenda.

To understand why this debate is hot enough to travel to the G7 table, you need to know what AI buyers are really purchasing. Yes, they are buying chips and cloud services. But under the hood, the limiting factor increasingly becomes energy. Data centres are electricity-hungry and require stable, predictable power. If you cannot reliably get the megawatts, you cannot scale compute, and you cannot hit delivery timelines. That means a country with abundant low-carbon electricity can market itself as a natural AI platform, not just a manufacturing base.

France’s framing matters because it reframes “cheap power” as more than a cost benefit. It becomes a policy lever and an industrial strategy. If France can turn low-carbon electricity into AI capacity, it can potentially attract AI workloads that produce jobs, research activity, and infrastructure investment. But it also risks creating an energy funnel that mostly benefits external players if the biggest data-centre customers are US firms. In other words, the allocation choice can determine who captures the value chain: local innovators building models, or foreign operators hosting compute.

That is where the second-order fight starts. Boards and investors typically think in terms of talent, distribution, and funding access. But energy allocation adds a new constraint that can reshape the competitive map. A homegrown AI lab may look like it has the talent and research output, but if power availability is controlled or prioritized differently, its growth can slow or become dependent on partnerships with larger data-centre providers. Meanwhile, an American giant may arrive with capital and operational experience, and secure data centre development, which could cement its position even if local startups have promising technology.

Regulatory background is also doing some quiet work here. Europe has been pushing decarbonization and grid planning in ways that influence which projects can move fastest and what compliance costs will be. When a national government signals that electricity is a strategic advantage, that can trigger a more deliberate approach to permitting, grid connection priorities, and the balance between industrial policy and open investment. The debate described in The Next Web effectively turns power into an economic question that regulators must govern.

For French decision-makers, the dilemma is essentially a trade between speed and domestic benefit. Give access to the companies already ready to build data centres, and you get faster capacity. Give access primarily to homegrown AI firms, and you potentially increase national capture of AI investment, but you may slow buildout if those firms do not have the infrastructure scale on day one. Neither choice is painless because power is physical and capacity additions take time, so getting it wrong can lock in years of competitive outcomes.

This is also why similar executives across Europe should pay attention. The French debate is a template for what will likely come next in other energy-influenced AI markets: governments will start treating electricity not just as a commodity, but as a scarce input for national competitiveness. That means CFOs and CTOs at AI labs should assume that energy supply discussions will become part of governance, not just operations. And it means boardrooms should watch allocation policy as closely as they watch model performance, because the best model in the world cannot run if the plug is not available.