The European Union is stepping up its backing of homegrown AI startups by allowing them access to the bloc's supercomputers for model training. This initiative, launched in September and recently put into action, aims to aid AI startups in effectively utilizing the immense computing power of supercomputers. Initial collaboration with Mistral AI from France highlighted the need for tailored assistance, stressing the importance of transferring skills, knowledge, and experience to optimize high-performance computing.

The strategy involves establishing 'centers of excellence' to assist in crafting AI algorithms that work seamlessly with EU supercomputers. Recognizing the AI community's familiarity with hardware from U.S. hyperscalers, the EU intends to supplement supercomputing access with comprehensive support. The goal is to equip startups with the expertise needed to make the most of the EU's infrastructure in developing AI algorithms that are both secure and competitive.

Furthermore, the EU intends to expand its supercomputing network and integrate quantum computing to enhance capabilities. Projects using this infrastructure include modeling Earth's ecosystems for climate studies and building a digital twin of the human body for medical advancements. The EU has also introduced a 'Large AI grand challenge,' providing significant supercomputing access and financial incentives to chosen European AI startups.

Why does this matter?

This strategic move aligns with the EU's focus on fostering innovation and competitiveness in the realm of AI. Although the program is in its early stages, the EU aims to use its investments in supercomputing to strengthen local AI ecosystems and narrow the gap with dominant US players.