EU expands supercomputer access to ethical and responsible AI startups

The European Union has announced that it will allow ‘ethical and responsible’ AI startups to use its high-performance computing (HPC) supercomputers to train their models. The new initiative was announced yesterday by European Commission president Ursula von der Leyen during the annual ‘State of the Union‘ address and was later detailed by Thierry Breton, the EU’s internal market commissioner. The EU has eight supercomputers located around the Union, often in research centers. The goal is to expand access to the EU’s HPC and promote the development of responsible AI. AI startups can apply to use the computing resource if they commit to the EU AI Pact, a set of principles and guidelines.

Why does it matter?

This initiative is consistent with the EU’s broader efforts to regulate AI and promote ethical and responsible AI development. It is an integral part of a global 3-pillar AI framework: guardrails, governance, and guiding innovation. The EU AI Act, which adopts a risk-based regulation, is in the final stages of negotiation between the parliament and member state governments, and discussions are underway with the US, the UN, and other partners to help mend regulatory gaps and produce a global AI governance framework.

Access to HPC for ‘responsible’ AI startups is a new incentive offered within the EU AI package. The plan will build on an existing program that already allows industry players to access the resource.
The EU’s supercomputers are among the most powerful in the world, and giving access to startups for AI training would constitute a significant contribution to the innovation ecosystem, with the aim of giving a competitive edge to select startups. It would expand access to these powerful computing resources and help level the playing field for smaller companies.