How to accelerate investment in sovereign AI computing infrastructure, models and services in the EU?

The Draghi report pointed out that the EU needs to substantially accelerate investment in AI computing infrastructure and promote the use of AI services. The Commission launched several taxpayer-subsidized AI infrastructure investment plans but also adopted the world’s first AI regulation that imposes stringent rules on the deployment of AI models and services.

But EU AI investment and roll-out is mainly driven by commercial firms. Nvidia, the leading producer of AI chips, is shipping more AI hardware to the EU. Mistral, the leading EU AI start-up, has partnered with several US Big Tech firms to overcome financing and infrastructure bottlenecks, including with Nvidia, and with ASML in the Netherlands. Accenture operates at the interface between model developers and firms that deploy these models in their businesses. It recently reported a 65% annual growth rate in AI consultancy work. That suggests a very fast uptake of AI by companies.

At this event we explored current bottlenecks in EU AI model roll-out and uptake by firms. How can hardware and financing bottlenecks be addressed? How to balance between model training and inference hardware? Is there a need for government intervention? What can we learn from the experience of Mistral and EU firms that integrate AI services in their business? Is EU AI regulation an obstacle?