CERN for AI: The EU’s seat at the table

CERN for AI: The EU’s seat at the table

Daan Juijn, Alex Petropoulos, Bálint Pataki, Max Reddel | September 2024

CERN for AI report cover

“We must now focus our efforts on becoming a global leader in AI innovation. I will propose to set up a European AI Research Council where we can pool all of our resources, similar to the approach taken with CERN.” Ursula von der Leyen, June 2024.

A CERN for AI could boost Europe’s economic performance, improve security against external threats, and develop truly trustworthy AI. Europe is lagging behind the US and China in advanced AI and, more generally, tech innovation, mainly because of lower capital deployment and a fragmented ecosystem.

A CERN for AI could give Europe the computational infrastructure to build its own frontier AI models, and to spur a thriving ecosystem of high-tech startups and scale-ups, underpinned by talent that would be incentivised to work in and for Europe. Such an ecosystem would benefit not only the private, but also the public sector. A large-scale pan-European effort would further promote the Union’s strategic autonomy and enable the development of more trusted, AI-assisted responses to external threats in domains such as cyberwarfare. 

Finally, and perhaps most significantly, making frontier AI safe and reliable remains an unsolved scientific problem. The EU cannot gamble on foreign, profit-driven companies to solve this problem, nor can it bank on regulation alone. History has shown that ambitious, European research efforts—like the original CERN—can rapidly expand the scientific frontier. Trustworthy AI can be invented in Europe. 

The idea, then, is compelling, but designing and creating an institution like this will require deep planning, strategic allocation of resources, and serious ambition in Brussels and beyond. To succeed, a CERN for AI should have:

  • Multiple paths to trustworthy AI: Make solving the scientific problem of trustworthy AI its core mission, and tackle it through multiple, targeted, research bets;
  • Competitive compute: Be allocated a budget of €35 billion over three years to ensure access to competitive computational infrastructure.
  • World-class leadership: Appoint leadership that can quickly attract top talent and hit the ground running;
  • Agile and democratic governance: Balance decisiveness and oversight in the governance structure;
  • Multi-level security: Strive for openness and transparency where responsible, and security where necessary;
  • Private sector involvement: Enable the private sector to build upon public, foundational research, and accept their co-funding after rigorous screening;
  • Talent and compute hubs: Create a single, dedicated talent hub, accompanied by 1-5 separate compute hubs;
  • International partnerships: Remain open to partnerships with like-minded non-EU countries; and
  • Benefit sharing: Ensure a benefit-sharing structure among participating governments and businesses.

The EU has a unique opportunity to succeed, However, it needs to act quickly: the steps taken in the following months can make or break a CERN for AI. The EU can deliver a Union-wide initiative to pool resources, talent, and ambition into a single, focused effort to develop world-class, trustworthy AI models. 

To do so, the new College of Commissioners should swiftly create an action plan to bring a CERN for AI from dream to reality.

Read the full report here (PDF)

Scroll to Top