Since 2024, the Commission has accelerated the creation of 19 AI Factories across 16 Member States, backed by roughly €10 billion for the 2021-2027 period. According to the AI Continent Action Plan, AI Factories integrate “AI-optimised supercomputers” and will function as “dynamic ecosystems that foster innovation”. Formally, however, they are oriented toward open scientific research rather than commercial use. Announced in 2025 under the InvestAI framework and tied to a €20 billion public-private funding scheme, the five planned AI Gigafactories across the EU aim at training industry-scale “AI frontier models”, which require massive computing and energy power. At least 65% of AI Gigafactories’ computing capacity will be reserved for commercial use, marking a point of divergence from AI Factories and signaling the EU’s ambition to compete with US hyperscalers. These public-private partnerships promise to deliver digital infrastructures that promote public values yet present significant challenges. To what extent will these facilities mitigate or reinforce dependencies? How can the EU operationalise open-source or other public-interest solutions that can be competitive enough to serve as alternatives to hyperscalers? Can they comply with the EU’s climate pledges considering their energy and water needs?