EU launches EUR10B tender for up to seven AI Gigafactories

The tender is framed by Brussels as a cornerstone of European technological sovereignty, yet the hardware plan tells a different story. The European Commission has secured letters of intent with Nvidia, AMD and Qualcomm to guarantee chip access for the gigafactories, following the EU-US trade deal ^[1]. Tech Times summed up the contradiction bluntly: the gigafactories are pitched as sovereign infrastructure, but the chips inside them are still American, and much of the cash behind them remains notional ^[2]. Each facility is required to house at least 100,000 cutting-edge AI chips, layered with cloud software, high-speed connectivity and energy-efficient data centres ^[3]- but nothing in the tender changes who manufactures the silicon. On Reddit, a France-focused thread pushed the critique further, arguing that real sovereignty would require EU-based semiconductor fabrication rather than just EU-based data centres, pointing to ASML as the closest thing Europe has to that capability. The gigafactories, on this reading, secure where the compute physically sits, not who controls the chips that power it.

Financially, the headline EUR10 billion in public funding is designed to be a multiplier: EU and national money is meant to unlock at least EUR20 billion in private capital, pushing total investment above EUR30 billion, split across two funding lots worth up to EUR400 million and EUR800 million per project respectively ^[4]. But the public half of that equation is less solid than the topline figure suggests. A senior Commission official cautioned that only about EUR1 billion of the EUR5 billion EU-budget share is currently confirmed, with the remainder contingent on the outcome of the next Multiannual Financial Framework negotiations - a process the official said could not be pre-empted ^[5]. In other words, the EUR10 billion figure driving headlines is closer to a ceiling than a guarantee, with the bulk of the EU's own contribution still unlocked.

Even if the money materializes, critics question whether the demand does. German Greens MEP Sergey Lagodinsky said flatly that proponents could not explain a concrete business case for the compute: "Nobody could explain to me what is the business case they are planning with these gigafactories" ^[6]. CEPS researcher Nicoleta Kyosovska went further, warning the facilities risk becoming "cathedrals in the desert" - prestige projects disconnected from actual EU industrial demand ^[7]. Reporting tied to that critique named Mistral AI as essentially the only EU-based company with a plausible use case for gigafactory-scale compute, noting it is already building its own data centers independently rather than waiting on the tender ^[6]. The skepticism echoed in public reaction: one heavily upvoted Reddit thread reframed the initiative as spending disconnected from ordinary demand, while a separate European thread split between calling the plan too small and too late against the US and China, and calling it reckless given the bloc's energy and climate constraints.

The scale math explains the urgency. The EU's existing network of 19 AI Factories falls well short of frontier-training capacity, and the new gigafactories are meant to more than double the bloc's AI computing power to close the gap with the US and China ^[8]. Demand for the program is real: the original plan for five gigafactories was expanded to up to seven after 76 expressions of interest arrived from 16 member states across roughly 60 candidate sites ^[9]. But Europe is racing that scale-up against a structural cost disadvantage - electricity prices in the EU run roughly two to three times higher than in the US and China, raising the operating-cost floor for every facility before it trains a single model ^[8]. The clock is also tight: bids close November 12, 2026, award decisions are expected by early 2027, construction can begin that year, and winning facilities must be operational within 18 months of signing their contracts ^[10].