# Seven AI Gigafactories: The EU's Big Bet on Compute Sovereignty > Source: > Published: 2026-07-30 10:26:53+00:00 # Seven AI Gigafactories: The EU's Big Bet on Compute Sovereignty The gigafactory concept was originally coined for battery production at scale. The EU is now rebranding it for AI: massive, purpose-built compute clusters designed to train frontier models. This isn't just about buying GPUs and racking them up—it's about creating a sustainable ecosystem with energy-efficient cooling, high-speed interconnects, and access for researchers and startups across member states. The call is managed through the EuroHPC Joint Undertaking, which has been running supercomputing infrastructure for years. Now they're adding GPU-heavy systems tailored for AI workload training. What I find interesting is the timing. While the US and China dominate large-scale training runs with clusters like xAI's Memphis facility or China's various exascale projects, Europe has been playing catch-up. The seven gigafactories aim to close that gap. The EU has been home to excellent AI research (DeepMind's roots, etc.), but when it comes to building the next generation of foundation models, the compute has been lacking. This move directly addresses that. One key detail: these gigafactories are designed to train next-generation AI, meaning they're expected to house cutting-edge accelerators—likely Nvidia H100s, B200s, and potentially custom silicon. The call is open to consortia that can demonstrate the ability to build and operate these facilities. National supercomputing centers, private cloud providers, and even large enterprise coalitions could bid. From a practical standpoint, this is huge for European AI startups. No more needing to rent US-based cloud clusters at premium prices while dealing with data residency concerns. Real-world, local compute for LLM training, fine-tuning, and scientific AI applications. The EU is also requiring that these gigafactories make a portion of their capacity available for open research, which could boost open-source model development. There are challenges, of course. Energy consumption: each gigafactory will draw massive power, and Europe's energy landscape is fragmented. Finding sites with enough renewable energy to keep the carbon footprint defensible is non-negotiable. Also, talent: operating these clusters requires engineers who understand distributed training, networking, and large-scale GPU orchestration. That skill set is still rare. Still, this is the most tangible AI investment I've seen from the EU in years. It's not a paper or a study—it's a shovel-ready infrastructure play. If they pull it off, Europe could host seven of the world's largest AI training clusters by 2030. As someone who follows AI infrastructure closely, I'm watching the tender results to see which consortia step up. This could reshape where the next wave of foundation models are trained. [Next Sam Altman's White House Talks: A Call to Decelerate AI? →](/en/news/4379/)