Nvidia and Perplexity are teaming up to collaborate with model builders and cloud providers in Europe and the Middle East to develop sovereign large-language models (LLMs). By leveraging Nvidia’s Nemotron techniques, these models will be fine-tuned to enhance accuracy and reduce costs for various enterprise tasks, including agentic AI applications.
Partners in this endeavor include renowned entities such as Barcelona Supercomputing Center, Bielik.AI, Dicta, and many others. These models will be trained and operated on servers within Europe provided by Nvidia Cloud Partners through the DGX Cloud Lepton marketplace. This collaborative effort aims to establish an integrated regional AI ecosystem that respects local languages and cultural nuances.
Nvidia’s involvement in Europe’s sovereign-AI initiatives is further solidified by this partnership, expanding Perplexity’s reach in the region. The move not only broadens model choices for enterprises but also disrupts traditional procurement practices for GPUs, cloud capacity, and AI infrastructure. As Nvidia introduces its AI Factories across Europe, competition for priority access will increase alongside the supply.
Furthermore, the agreement underscores the importance of managing data compliance concerns, especially with the EU’s AI Act looming. Compliance with regulations such as GDPR and the upcoming AI Act is crucial, requiring transparency in how models are trained, fine-tuned, and updated. CIOs must ensure vendors disclose the origins of training data, explain model processes, and prevent data transfer outside the bloc to comply with regulations effectively.
In conclusion, the collaboration between Nvidia, Perplexity, and various entities in Europe and the Middle East represents a significant step towards fostering local AI expertise and infrastructure. By fine-tuning large-language models and addressing data compliance challenges, this partnership sets a precedent for ethical and innovative AI development in the region.