The EU AI Continent Action Plan, seen by Capacity, outlines a sweeping strategy to scale computing power, boost AI talent, and unlock access to high-quality data. It includes plans to build AI ‘Gigafactories’ capable of supporting frontier models, including those edging towards so-called Artificial General Intelligence (AGI).
The plan, first reported by Politico, lays out five core pillars: next-gen infrastructure, open data access, targeted sector adoption, talent development, and regulatory streamlining — all designed to forge what Brussels is calling a “distinctively European” brand of trustworthy, sovereign AI.
The 21-page blueprint includes plans to invest a total of €2 billion to build 13 AI factories (or specialised AI data centres) across 17 Member States.
The already selected sites for AI factories include SLAIF in Maribor, Slovenia; BSC AIF at the Barcelona Supercomputing Centre in Spain; and PIAST AI in Poznan, Poland. A call for further sites will launch in May.
Member states will also be able to apply for ‘AI Factory Antennas’ to gain remote access to centralised AI supercomputing resources, enabling them to develop their national AI ecosystems without investing in costly dedicated infrastructure of their own.
The European Commission also plans to fund the procurement of nine new AI supercomputers across the continent, while existing units will be upgraded in a move it claims will “more than triple” the current AI computing capacity of the EuroHPC initiative.
But it’s the AI Gigafactories that potentially mark the EU’s most dramatic leap in ambition.
These mammoth facilities, described in the plan as “inspired by the ambition of CERN”, will each host over 100,000 advanced AI processors, dwarfing most current European systems.
The proposed facilities are intended to support the development of 'frontier' AI models with trillions of parameters, potentially paving the way for AGI, where AI can think for itself — a concept straight out of science fiction but a prospect the Commission now openly references in its planning.
Each AI Gigafactory is expected to demand massive capital investment, with the EU eyeing up to five sites initially.
A new €20 billion InvestAI Facility, launched in partnership with the European Investment Bank, will underwrite the effort through public-private partnerships. A call for expressions of interest opened today (April 9), with a full call for proposals due in Q4.
Any of the new mega-scale data centre projects outlined in the blueprint would be required to be sustainable, with the document stipulating energy efficiency, water optimisation, and AI-powered automation as design requirements.
The Commission is also hoping this infrastructure push will catalyse European chip development, including AI-optimised processors, under a future revision of the continent's Chips Act hinted at taking place in 2026.
Beyond compute: Europe’s plan to fuel AI with data, skills & rules
Beyond raw compute, the Action Plan also sets its sights on data, talent, and regulation, all listed by the Commission as key enablers needed to unlock Europe’s full AI potential.
A new Data Union Strategy, due later this year, will aim to simplify cross-border data sharing and bolster access to sector-specific datasets, particularly for SMEs and researchers.
As part of the proposed strategy, ‘Data Labs’ will be embedded within AI Factories to help pool, clean and enrich data, making it available to developers across Europe via secure, interoperable platforms.
On the talent front, a pan-EU AI Skills Academy will launch in Q2, offering PhD fellowships, a pilot generative AI degree, and returnship schemes designed to boost female participation in the field.
Meanwhile, the bloc will expand joint European AI degrees and apprenticeships and explore legal migration pathways for international AI experts.
The blueprint also outlines the Apply AI Strategy, set for release in Q3, which will focus on driving adoption in strategic sectors, including telecoms, aerospace, energy, healthcare and defence.
The new Apply AI Strategy will cover use-case-based funding, testing facilities, and targeted R&D programmes, with €700 million earmarked for projects under the EU’s GenAI4EU initiative.
Finally, the Commission wants to avoid burdening Europe’s AI hopefuls with red tape, a concern raised by several big-name brands following the implementation of the EU AI Act.
A new AI Act Service Desk, due to go live in July, will offer guidance and self-assessment tools to help smaller firms navigate the bloc’s landmark AI regulation.
A public consultation launching later this month will gather feedback on pain points and support needs as the AI Act rolls out through to 2027.
Legal clarity and cross-border impact
The EU’s focus on compute and innovation may grab the headlines, but its efforts to simplify regulation could have ripple effects well beyond the bloc’s borders.
John Buyers, global head of AI at law firm Osborne Clarke, sees strong parallels between the EU’s approach and recent moves by the UK government, particularly around infrastructure, data access, and easing regulatory burdens.
“As we expected, key themes of the EU's plans echo those of the UK government's AI Opportunities Action Plan,” Buyers said. “There’s a real emphasis on easing the burden of regulation and removing barriers to innovation, likely in response to concerns raised by the US as well.”
He warns, however, that ambiguity around the EU AI Act remains a source of anxiety for developers, particularly those working on general-purpose AI. The Act’s provisions, already partially in force, will cover cutting-edge generative models, with penalties for non-compliance reaching up to €15 million or 3% of global turnover.
“Despite these potentially large fines, AI companies still don't yet know in detail how [the AI Act] will apply to them,” he said. “This uncertainty makes it difficult to plan or draft the contracts that underpin innovative uses of AI.”
Buyers also welcomed the Action Plan’s emphasis on data access, noting it aligns with the UK’s own AI Opportunities Action Plan and pending data legislation. But he flagged unresolved legal issues around data use, particularly in the UK, where proposed copyright law reforms are proving contentious.
“If the EU is serious about eliminating legal uncertainties around the AI Act, that could be a real boost not just for European developers, but also for those in the UK and US. After all, the AI Act applies to all AI used in the EU, regardless of where it's built.”
RELATED STORIES
Europe risks losing AI race without urgent action on digital infrastructure
EU nations unite to launch coalition to secure semiconductor autonomy
