Spain invests €719m in AI compute gigafactory for Europe

Spain’s AI gigafactory push is one of the clearest examples yet of a nation treating frontier AI compute as critical infrastructure on par with energy or telecoms. By earmarking roughly €700–719 million for a public‑private consortium and tying the project into the EU’s InvestAI framework and EuroHPC network, Madrid is explicitly betting that domestic access to large clusters of GPUs will determine who can train and deploy the next generation of foundation models. That’s a direct response to a landscape currently dominated by US hyperscalers and, to a lesser extent, China.

Strategically, this is about European sovereignty in the race to AGI. If Europe has to rent all of its frontier compute from a handful of American platforms, it has limited leverage over safety standards, export controls, or which labs get to run the most capable models. A Spanish‑anchored gigafactory gives EU startups, corporates, and public labs a place to train competitive models under European law and values, from open‑weight systems to highly regulated healthcare or defense applications.

The move also raises the floor under smaller players. A subsidized national compute hub can blunt the "winner‑takes‑all" dynamic where only the richest US labs can afford to scale to tens or hundreds of thousands of GPUs. If implemented well, Spain’s project could become a template other mid‑sized economies copy: federated, multi‑site AI supercomputers tied into both industrial policy and safety governance.