As of mid-2026, the project is facing serious headwinds:

• Delays and funding impasse. The formal bidding process has been pushed to July 2026, and reports indicate only two of the five planned centers are likely to receive funding before the EU’s next budget cycle begins in 2028 . The European Commission and European Investment Bank did not sign a memorandum of understanding to unlock financing until December 2025, months behind schedule . The EIB’s slow, multi-phase advisory process has been a sticking point for private consortia .

• Vague subsidy conditions. The financing structure requires roughly 65% private capital, but investors complain that the terms, eligibility rules, and subsidy frameworks remain too unclear to commit . At least two consortia are reportedly reconsidering their participation due to the uncertainty .

• No clear demand signal. This may be the plan’s most fundamental vulnerability. Critics argue the Commission has taken a purely supply-side approach, assuming that a lack of compute is Europe’s primary AI problem. Analysis from the think tank Interface EU concludes that while gigafactories may support training medium-sized research models, they are insufficient for driving large-scale commercial AI innovation . Legislators and experts have openly questioned what specific use cases the infrastructure would serve beyond research .

• Total Nvidia dependency. Each gigafactory requires approximately 100,000 cutting-edge GPUs, but Europe has no domestic supplier of advanced AI chips. The plan is entirely dependent on continued access to Nvidia hardware, which is subject to global supply bottlenecks and export restrictions set by the previous U.S. administration . Data center experts had flagged this risk early on .

• Regulatory and energy drags. High energy costs, protracted permitting timelines, and the broader EU regulatory landscape—including the phased rollout of the AI Act—add further friction to an already slow timeline .

The result is a credibility gap. A flagship infrastructure plan announced with fanfare has drifted into a holding pattern, with more questions than answers about whether and when it can actually deliver.

France’s ‘Choose France’ Summit: A Counter-Model in Action

On June 1, 2026, at the Palace of Versailles, President Macron announced that 71 foreign investment projects worth a combined €93 billion ($108 billion) had been pledged at the ninth edition of the Choose France summit . The total surpassed the combined €87 billion raised over the previous eight editions of the event .

Nearly half of that sum—€45 billion—comes from a single deal: Japanese tech investment giant SoftBank’s commitment to develop 3.1 gigawatts of AI data center capacity in the Hauts-de-France region by 2031, with a pathway to scale the investment to €75 billion . Canadian asset manager Brookfield added another €10 billion for AI-related infrastructure, while a Franco-Emirati consortium committed up to €50 billion for a 1 GW AI campus .

The implications for Europe’s AI infrastructure race are stark:

• Private capital, no complex co-financing. The €93 billion is overwhelmingly private corporate investment, requiring none of the multi-layered EU co-financing, joint venture structuring, or EIB advisory phases that have slowed the gigafactory plan. SoftBank is not waiting for a subsidy framework to be clarified .

• Speed and political accountability. Macron explicitly staked his personal credibility on delivering these projects—a sharp contrast to a Commission initiative diffused across multiple institutions . The pledges are tied to concrete buildout plans and an expected 15,600 jobs .

• Nuclear energy as a strategic asset. France’s low-carbon nuclear grid became a central bargaining chip. Macron leaned heavily on the country’s energy independence to attract hyperscalers and tech investors facing power constraints elsewhere in Europe . The EU plan, by contrast, has no coordinated energy advantage and depends on member states solving power and permitting bottlenecks individually.

Two Europes, One Race

The EU’s top-down approach relies on stitching together public funding, multi-member-state coordination, and a hope that if Europe builds compute, an AI industry will follow. France, instead, is selling what global capital wants: shovel-ready sites, reliable clean energy, and a single government that can deliver on its promises.

Whether either model produces a genuinely competitive European AI ecosystem remains an open question. Critics of the French approach note that investment pledges are not completed data centers, and the EU’s more deliberative model may ultimately produce infrastructure with broader public access rather than concentrated corporate control. But on the metrics that matter most to investors—speed, clarity, and credible energy answers—the gap is widening. The race for AI infrastructure in Europe is no longer a single bloc-wide project. It is a contest between competing visions of how the continent should build its digital future, and France just placed a very large bet.

As of mid-2026, the project is facing serious headwinds:

• Delays and funding impasse. The formal bidding process has been pushed to July 2026, and reports indicate only two of the five planned centers are likely to receive funding before the EU’s next budget cycle begins in 2028 . The European Commission and European Investment Bank did not sign a memorandum of understanding to unlock financing until December 2025, months behind schedule . The EIB’s slow, multi-phase advisory process has been a sticking point for private consortia .

• Vague subsidy conditions. The financing structure requires roughly 65% private capital, but investors complain that the terms, eligibility rules, and subsidy frameworks remain too unclear to commit . At least two consortia are reportedly reconsidering their participation due to the uncertainty .

• No clear demand signal. This may be the plan’s most fundamental vulnerability. Critics argue the Commission has taken a purely supply-side approach, assuming that a lack of compute is Europe’s primary AI problem. Analysis from the think tank Interface EU concludes that while gigafactories may support training medium-sized research models, they are insufficient for driving large-scale commercial AI innovation . Legislators and experts have openly questioned what specific use cases the infrastructure would serve beyond research .

• Total Nvidia dependency. Each gigafactory requires approximately 100,000 cutting-edge GPUs, but Europe has no domestic supplier of advanced AI chips. The plan is entirely dependent on continued access to Nvidia hardware, which is subject to global supply bottlenecks and export restrictions set by the previous U.S. administration . Data center experts had flagged this risk early on .

• Regulatory and energy drags. High energy costs, protracted permitting timelines, and the broader EU regulatory landscape—including the phased rollout of the AI Act—add further friction to an already slow timeline .

The result is a credibility gap. A flagship infrastructure plan announced with fanfare has drifted into a holding pattern, with more questions than answers about whether and when it can actually deliver.

France’s ‘Choose France’ Summit: A Counter-Model in Action

On June 1, 2026, at the Palace of Versailles, President Macron announced that 71 foreign investment projects worth a combined €93 billion ($108 billion) had been pledged at the ninth edition of the Choose France summit . The total surpassed the combined €87 billion raised over the previous eight editions of the event .

Nearly half of that sum—€45 billion—comes from a single deal: Japanese tech investment giant SoftBank’s commitment to develop 3.1 gigawatts of AI data center capacity in the Hauts-de-France region by 2031, with a pathway to scale the investment to €75 billion . Canadian asset manager Brookfield added another €10 billion for AI-related infrastructure, while a Franco-Emirati consortium committed up to €50 billion for a 1 GW AI campus .

The implications for Europe’s AI infrastructure race are stark:

• Private capital, no complex co-financing. The €93 billion is overwhelmingly private corporate investment, requiring none of the multi-layered EU co-financing, joint venture structuring, or EIB advisory phases that have slowed the gigafactory plan. SoftBank is not waiting for a subsidy framework to be clarified .

• Speed and political accountability. Macron explicitly staked his personal credibility on delivering these projects—a sharp contrast to a Commission initiative diffused across multiple institutions . The pledges are tied to concrete buildout plans and an expected 15,600 jobs .

• Nuclear energy as a strategic asset. France’s low-carbon nuclear grid became a central bargaining chip. Macron leaned heavily on the country’s energy independence to attract hyperscalers and tech investors facing power constraints elsewhere in Europe . The EU plan, by contrast, has no coordinated energy advantage and depends on member states solving power and permitting bottlenecks individually.

Two Europes, One Race

The EU’s top-down approach relies on stitching together public funding, multi-member-state coordination, and a hope that if Europe builds compute, an AI industry will follow. France, instead, is selling what global capital wants: shovel-ready sites, reliable clean energy, and a single government that can deliver on its promises.

Whether either model produces a genuinely competitive European AI ecosystem remains an open question. Critics of the French approach note that investment pledges are not completed data centers, and the EU’s more deliberative model may ultimately produce infrastructure with broader public access rather than concentrated corporate control. But on the metrics that matter most to investors—speed, clarity, and credible energy answers—the gap is widening. The race for AI infrastructure in Europe is no longer a single bloc-wide project. It is a contest between competing visions of how the continent should build its digital future, and France just placed a very large bet.