Europe opens bidding for seven AI gigafactories, and the numbers explain the problem

The European Commission has opened the call for what it calls AI gigafactories: up to seven large sites across the EU built to train and run big AI models. Up to 10 billion euros in EU and national funding is meant to attract at least 20 billion euros of private investment, for a headline figure of around 30 billion euros. Eighteen member states are taking part, Germany and France among them. Applications close on 12 November 2026, award decisions are expected in early 2027, and construction should start the same year.

Interest is not the bottleneck. According to the Commission, 76 parties came forward proposing sites at 60 locations across 16 member states. The Commission has also signed letters of intent with AMD, Nvidia and Qualcomm to secure hardware, which is the part that money alone does not solve. Sites can be cross-border, spread across several member states as distributed compute rather than one enormous building. The stated goal is to give startups, small and mid-sized firms, universities and public bodies access to training infrastructure they currently cannot get in Europe.

Then there is the comparison nobody in Brussels wants to make out loud. Major US tech companies plan to spend more than 600 billion dollars on data centres this year alone, and that figure keeps being revised upward. Europe’s entire package, public and private together, is roughly twenty times smaller, and it is spread over years rather than one budget cycle.

What is behind this. Compute is the raw material of modern AI, and Europe currently rents almost all of it from American clouds. That is the sovereignty argument: if the infrastructure sits elsewhere, so does the pricing power, the data, and the ability to say no. The gigafactory plan is the EU’s attempt to own some of the base layer rather than only regulating what runs on top of it. Whether 30 billion euros buys meaningful independence is a fair question, and the honest answer is probably not on its own. What it can buy is access for the people who currently have none, which is a smaller but more achievable goal than catching up with hyperscalers.

What this means for you: nothing this year, and probably nothing next year either, since the first concrete is not poured until 2027. If you run a small European company or work at a university, this is the news worth bookmarking, because subsidised access to training infrastructure is exactly the thing that is currently out of reach at that scale. If you follow AI as a citizen rather than a builder, the useful takeaway is the ratio. Europe writes the rules for AI and rents the hardware. This is an attempt to change the second half, and the size of the gap tells you how long that will take.

Sources

• European Commission: EU launches AI Gigafactories call
• European Commission: AI Gigafactories overview
• The Decoder: EU pools up to 30 billion euros for AI gigafactories