The EU AI gigafactories tender officially opened on July 30, 2026, with the European Commission inviting bids to build up to seven large-scale AI computing facilities across the bloc, backed by up to 10 billion euros in public funding and a targeted 20 billion euros in matching private investment. The pitch, in the words of Commission Executive Vice President Henna Virkkunen, is that “access to the raw scale of computing power within AI Gigafactories is a strategic necessity for Europe as AI development accelerates.”

The ambition traces back to a promise European Commission President Ursula von der Leyen made at the Paris AI Action Summit in February 2025: build Europe’s own CERN for artificial intelligence, a facility so large and so central to the field that it becomes a genuine gravitational centre for European AI research and industry rather than a smaller-scale complement to infrastructure built elsewhere.

Eighteen months later, the EU AI gigafactories initiative has moved from speech to tender document, and the gap between the original vision and what is actually being funded tells its own story about the pace at which Europe can move on the technology it says it urgently needs to catch up on.

What the EU AI Gigafactories Call Actually Funds

The European Commission’s own announcement describes each gigafactory as combining advanced AI processors, software and cloud technology stacks, high-speed connectivity, and energy-efficient data centres, built to train, run, and fine-tune frontier AI models. They will sit alongside Europe’s existing network of 19 smaller AI Factories, and will be required to follow EU standards on data protection, safety, security, and ethics. Ten countries, including Germany, France, Italy, Poland, and Spain, have already expressed interest in hosting one, with both single-country and multi-country consortia eligible to bid.

Interest from industry has been strong enough that the Commission expanded the programme’s scope from an original four or five gigafactories to seven, after 76 potential consortia expressed preliminary interest in submitting a proposal. Successful bidders are expected to begin physical construction in early 2027, with facilities becoming operational by mid-2028.

The Funding Math That Undercuts the Headline Number

Reporting from Euronews lays out a funding picture considerably less generous than the initial 10-billion-euro figure suggests. The Commission has scaled back its financial commitment over time, and under the currently approved budget, Brussels can only actually commit 1 billion euros right now, with the remainder of its share dependent on the next Multiannual Financial Framework, the EU’s long-term budget, which is itself still being negotiated among member states. A senior Commission official told Euronews plainly: “We cannot pre-empt the decisions about the next MFF. We gave you our best estimate of how much money we would have from the next MFF to be able to support phase two.”

The procurement process has been split into two phases specifically because of that funding uncertainty, staggering the buildout over roughly six and a half years rather than delivering it in one push. Euronews also reports that the Commission has faced criticism for repeatedly delaying the initiative in a way that undercuts its own rhetoric about the urgency of the US-China AI race, and that the projects, whatever their public funding, will still depend on private operators making them commercially self-sustaining, since all operating costs fall on the companies involved.

The Chip Dependency the Initiative Cannot Build Its Way Out Of

A pointed criticism of the EU AI gigafactories plan is that it aims to build sovereign European infrastructure while remaining almost entirely dependent on foreign-designed chips to run it. The Commission has signed memoranda of understanding with Nvidia, AMD, and Qualcomm, all non-European chipmakers, and has built anti-lock-in criteria into the tender specifically because none of the seven planned facilities will run on European-designed silicon.

That tension sits at the centre of LiveAIWire’s earlier coverage of the EU’s Cloud and AI Development Act, which found that Europe’s own sovereignty framework requires levels of ownership and control over AI infrastructure that the bloc’s current compute base, chip supply, and capital position cannot yet deliver, regardless of how the legal categories are defined on paper.

The dependency runs deeper than chips alone. LiveAIWire’s coverage of the ongoing chip alliance competition between the US and China has documented how both powers are actively building coalitions of allied states whose AI infrastructure choices lock them into one technology ecosystem or the other, and Europe’s gigafactory bet is, in effect, a wager that building physical capacity on its own soil is enough to preserve genuine independence even while the underlying components remain sourced from American suppliers.

Why Owning the Software Layer Might Matter as Much as Owning the Chips

A separate and less discussed vulnerability sits one layer up from the hardware question entirely. LiveAIWire’s recent reporting on Nscale’s 1.65 billion dollar acquisition of Anyscale documented a broader industry shift toward vertically integrated compute providers who control not just chips and data centres but the orchestration software that decides how efficiently a GPU fleet actually runs. Europe’s gigafactories will own the physical infrastructure the tender describes, but whether the software stack running on top of that infrastructure is also European, rather than licensed from the same small set of American and increasingly consolidated vendors dominating that layer globally, is a question the current tender documents do not resolve.

What This Means for You

If your organisation operates in the EU and currently depends on US-hosted frontier AI models, the EU AI gigafactories tender is the clearest concrete signal yet of how seriously Brussels intends to build an alternative, but the 2028 operational target means that alternative is years away, not months. Public sector bodies and companies with EU compute needs should treat this as a multi-year infrastructure story rather than an imminent shift in where their workloads can run.

For businesses in the ten countries competing to host a facility, the more immediate opportunity lies in the consortium bidding process itself, since national champions in industries from cloud services to energy are already assembling proposals ahead of the submission deadline. France has already signalled it intends to bid alone rather than as part of a multi-country consortium, a sign of how much national prestige, as well as investment, is riding on which countries actually secure a facility.

The honest reading of the EU AI gigafactories initiative is that it represents genuine political will paired with a funding and chip-dependency reality that has not yet caught up to the ambition Brussels announced in Paris eighteen months ago. Whether the finished facilities deliver the technological sovereignty the programme promises, or simply a more expensive, EU-based way of running infrastructure that still depends on American chips and, quite possibly, American software, is the question the next six and a half years of construction will actually answer.

What is not in dispute is the scale of appetite the tender has already revealed. Seventy-six consortia expressing preliminary interest in a programme that has not yet resolved its own long-term funding is a strong signal that European industry believes this infrastructure is worth building regardless of how the sovereignty question ultimately resolves, even if the political and fiscal questions underneath the announcement remain genuinely unsettled.

About the Author

Stuart Kerr is Technology Correspondent at LiveAIWire, covering artificial intelligence, cybersecurity, and the social impact of emerging technology. He publishes daily at LiveAIWire.com.