On January 16, 2026, the Council of the EU formally adopted an amendment expanding the mandate of the European High Performance Computing Joint Undertaking (EuroHPC JU), clearing the legal path for building AI gigafactories across the Union. The instrument — Council Regulation (EU) 2026/150 — amends Regulation (EU) 2021/1173, the 2021 regulation that established the Joint Undertaking, and entered into force on January 20.
For the EU, this is the load-bearing step in its compute-sovereignty strategy. As frontier training costs keep climbing and training-grade compute concentrates in a handful of US providers, the Union is betting on legislation, public money, and joint procurement to keep the capacity to train next-generation large models inside Europe.
What an AI Gigafactory Actually Is
EuroHPC JU already procures and operates the EU’s supercomputers, and a 2024 amendment brought “AI Factories” into its scope. This amendment pushes the ambition to hyperscale: the European Commission defines AI gigafactories as large-scale, AI-optimized supercomputing facilities built to support the training and deployment of state-of-the-art large AI models. The regulation text published on EUR-Lex is more concrete still — the facilities are meant to host the development of models with trillions of parameters.
The same amendment also gives EuroHPC JU a dedicated quantum computing pillar. HPC, AI, and quantum now sit under one joint undertaking, which the Commission frames as strengthening Europe’s “strategic autonomy and global competitiveness” across high-performance computing, AI, and quantum technologies.
From Proposal to Force of Law in Roughly Six Months
The Commission proposed the amendment in July 2025 under the consultation procedure: Parliament delivered an opinion, and the Council had the final word. Adoption on January 16, 2026, followed by entry into force on January 20, makes for a roughly six-month run — fast by EU legislative standards.
The regulation hard-codes its own deadlines: EuroHPC JU’s governing board must adopt a work programme within six months of entry into force. Kilian Gross, head of AI policy at the Commission’s DG CNECT, has said the official call for establishing AI gigafactories is planned for the first quarter of 2026. In practice, siting, operators, and procurement structure should turn from principle into concrete tenders quickly.
The Money: InvestAI and Public Procurement Rules
Gigafactories are not financed through classic grants. They will be procured under Public Procurement of Innovative Solutions (PPI): winning consortia finance construction and operation, with the EU capping its own contribution at 17 percent of each facility’s capital expenditure and member-state-backed consortia carrying the rest. The design stretches a limited EU budget across much larger private and national funding, and shifts delivery risk onto the operators.
The money sits inside the Commission’s InvestAI framework: of the €200 billion overall AI investment push, €20 billion is earmarked for AI gigafactories. Against single US campuses that run to tens of billions of dollars, that is not a moonshot. The point is different — it turns public training compute into a standing institution rather than a one-off stimulus package.
What It Means for Developers and Europe’s AI Ecosystem
Three observations. First, the timeline is faster than expected: a Q1 call plus a work programme due within six months means 2026 should produce concrete facility selections and operator consortia, not another white paper. Second, for European startups and research organizations, this opens a new lane to training-grade compute that does not depend on the quotas and pricing of US cloud vendors — a sovereign option for scheduling and budgeting model training. Third, folding quantum into the same undertaking signals that the EU treats post-Moore computing as one integrated layout, not two parallel systems.
At a moment when model cadence is accelerating and terms and open-source strategies keep diverging (see our 2026 opening outlook), the EU’s answer is not to chase any single model but to fix the infrastructure underneath. That route pays off slowly — but once built, it reshapes the cost structure of training for an entire generation of models.
Sources
- Artificial intelligence: Council paves the way for the creation of AI gigafactories — Council of the EU
- EuroHPC Regulation amended to strengthen Europe’s AI and quantum capabilities — European Commission
- Council Regulation (EU) 2026/150 — EUR-Lex
AI-assisted summary compiled from the sources above, reviewed by a human before publishing.