Europe launches bidding for seven AI 'gigafactories' in a €30 billion effort to catch up.

      The European Commission has initiated the bidding process to construct up to seven AI “gigafactories,” which are extensive computing centers aimed at empowering the region to train advanced AI models independently from the United States. The call for tenders, announced today, represents a roughly €30 billion initiative to catch up with the American and Chinese AI advancements.

      The framework is designed as a collaboration between the public and private sectors. Brussels and the member countries are expected to contribute approximately €10 billion collectively, while the remaining €20 billion is anticipated from private investors, on the premise that government funding can attract the necessary capital for these ventures.

      The scale of the project is intentionally vast. Each gigafactory is projected to include at least 100,000 state-of-the-art AI chips, making it about four times more potent than the largest data centers currently operational in the EU, and collectively, they would enhance the bloc’s AI computing capabilities more than twofold.

      The choice of the term “gigafactory” is a significant escalation. Europe already invests in a network of 19 smaller “AI factories” linked to its supercomputers, but the new facilities are designed to be substantially larger, aimed at training the advanced models that only a few US and Chinese laboratories can currently afford.

      The demand is evident. A previous call for expressions of interest yielded 76 responses from various consortia across Europe, with ten countries, including Germany, France, Italy, Spain, and Poland, showing interest in hosting, and France already indicating plans to proceed independently.

      The motivation stems from competitive concern. Europe has observed the US and China heavily invest in AI infrastructure while its own companies rely on computing from American cloud services. Henna Virkkunen, the Commission’s chief for tech sovereignty, emphasized that such substantial computational power is "a strategic necessity."

      However, the ambition is currently ahead of financial commitments. Only about €1 billion of the EU's contribution is formally allocated, with the remainder contingent on the next long-term budget, the Multiannual Financial Framework, which is still under discussion.

      A senior official acknowledged this uncertainty clearly, stating, “We cannot pre-empt the decisions about the next MFF,” and characterized the figure as a “best estimate” rather than secured funds.

      The timeline poses another challenge. Construction is slated to start in early 2027, with the first gigafactories expected to be operational by mid-2028, contingent on budget approvals, site selections, and supply chain arrangements.

      Past experiences suggest a need for caution. The gigafactory proposal was initially suggested in early 2025 but has since faced multiple delays that have frustrated potential collaborators essential to the project’s realization.

      Additionally, there is a paradox in the sovereignty narrative. The chips for these European gigafactories will primarily be sourced from Nvidia, AMD, and Qualcomm—American companies—which implies that the bloc would be purchasing independence from the very companies from which it seeks to gain autonomy.

      Energy costs pose another limitation. Electricity prices in Europe are two to three times higher than in the US or China, and the energy demands of data centers are already a constraining factor that subsidies cannot swiftly address.

      Critics have raised doubts about the entire model, arguing that leasing sovereign-branded capacity constructed on foreign-owned hardware perpetuates a misconception of independence rather than actually achieving it, highlighting that Europe’s primary deficiency lies in the absence of chips and models, not in facilities.

      However, proponents contend that infrastructure is a crucial initial step rather than an end goal. While compute resources alone will not create a European OpenAI, without them, the continent’s startups lack the capability to train at scale, which is the gap the Commission is striving, albeit at a high cost, to bridge.

      The Commission argues that a beginning is necessary. Public funders would receive a proportional allotment of the compute for research and public initiatives, with the hope that establishing the infrastructure will attract talent, startups, and eventually chip manufacturing to Europe.

      For now, the bidding process is open, and the financial commitment remains largely theoretical. Companies are invited to participate in shaping Europe’s AI future, under the realization that the continent has announced an ambitious €30 billion plan while securing only around a thirtieth of that amount so far.