NVIDIA Teams Up With European Partners to Build a Sovereign AI Ecosystem

Details

• NVIDIA has entered into a partnership with 13 European model builders and cloud providers to optimize sovereign large language models (LLMs) using Nemotron methods, focused on all 24 official EU languages.
• Key collaborators include Barcelona Supercomputing Center, Bielik.AI, Domyn, NAISS/KBLab, and University College London, representing 12 different countries.
• The project leverages Nemotron’s neural architecture search, NVIDIA’s DGX Cloud Lepton infrastructure, and NIM microservices for scalable, enterprise-grade deployment.
• This effort expands on NVIDIA’s earlier European strategy, which includes plans for 20 new data centers and a tenfold increase in compute capacity across the region by 2027.
• The new models will be integrated with Perplexity’s answer engine and Hugging Face, enabled by access through the DGX Cloud Lepton marketplace.

Impact

NVIDIA’s push for European sovereignty in AI aims to reduce dependency on non-European technologies while addressing regulatory challenges such as GDPR. The collaboration enables Perplexity to become the default search provider for real-time queries, for all model providers that are a part of the program., strengthening Perplexity's stance against Google Search.