The European Commission has launched an investigation into the construction of up to seven artificial intelligence gigafactories. Their combined value is expected to exceed 30 billion euros. Brussels and the Member States will provide up to €10 billion in public funding, whilst private consortia are expected to contribute a further €20 billion. The deadline for submitting bids is 12 November 2026.
This is not about yet another network of laboratories or a grant scheme for start-ups. The gigafactories are intended to serve as industrial infrastructure for training, fine-tuning and operating the most advanced AI models. Each facility should have at least 100,000 state-of-the-art processors. This is around four times more than in the largest AI centres currently operating in the EU. Launching these seven facilities would more than double Europe’s computing resources dedicated to artificial intelligence.
Europe has supercomputers, but lacks scale
The EU is already developing 19 AI factories connected to the EuroHPC infrastructure. These provide access to supercomputers for companies, universities and public institutions. Gigafactories are intended to operate at a higher level: to handle models with trillions of parameters and projects requiring tens of thousands of accelerators running simultaneously.
The scale of the project reflects the scale of Europe’s lag. According to data cited by*Le Monde*, around 60 per cent of advanced AI models are developed in the United States, 30 per cent in China, and less than 5 per cent in Europe. The five largest cloud service providers operating on the European market are American companies. European businesses are therefore purchasing a key resource for AI development outside their own market.
Brussels is attempting to change this structure by shifting some of the computing power, data and operational control to Europe. There is considerable interest. In the first round, the Commission received 76 expressions of interest covering 60 potential locations across 16 Member States. The applicants announced plans to purchase a total of at least 3 million specialised processors.
Sovereignty over Nvidia’s chips
The most significant limitation of the project lies within the servers themselves. Europe does not currently manufacture the most powerful accelerators needed to build such facilities. The first gigafactories will rely primarily on chips from US companies, including Nvidia and AMD.
The EU may gain greater control over data localisation, access to infrastructure and service continuity. However, it will not achieve full technological independence. Processors, some software and key elements of the supply chain will still remain outside Europe.
Gigafactories are therefore an investment in operational autonomy, not in complete sovereignty. They address the problem of a lack of computing power, but do not replace the European semiconductor industry, its own cloud platforms or competitive base models.
Energy will determine the location
The second barrier is access to electricity. The International Energy Agency forecasts that by 2030, energy consumption by European data centres will increase by over 45 TWh, or around 70 per cent compared with 2024. At the same time, overloaded grids and long waiting lists for grid connections may slow down the pace of new investment.
The choice of hosts for gigafactories will therefore depend not only on IT expertise. Available power capacity, energy costs, grid stability, cooling capabilities and the speed at which permits can be obtained will all be key factors. A country without adequate energy infrastructure will not win this race on the basis of subsidies alone.
Poland remains in the running. The government has guaranteed over 400 million zlotys to purchase part of the computing power for the future facility. The money does not directly finance the construction, but secures the demand required of programme participants.
The timetable envisages the selection of projects in 2027 and the commissioning of the first facilities in mid-2028. However, the funding is not yet finalised: €4 billion of the Commission’s declared €5 billion contribution depends on the next EU multiannual budget.
Gigafactories could remove a real constraint on the European AI market. However, they do not guarantee the emergence of European leaders. If the new capacity is not utilised for European models and commercial deployments, a significant portion of the €30 billion will primarily go towards the purchase of American processors and an increase in energy consumption in Europe.
