Nvidia Chief Executive Officer Jensen Huang used the VivaTech 2025 trade fair in Paris to unveil an aggressive European expansion that includes at least 20 so-called “AI factories” and new research hubs across the continent. The chipmaker said it is broadening ties with industrial groups such as Schneider Electric and Siemens to advance simulation-first, AI-driven manufacturing, underscoring Europe’s importance as it contends with sharply lower sales to China because of U.S. export restrictions. Central to the push is a partnership with French startup Mistral AI to create “Mistral Compute,” a sovereign cloud platform offering private AI stacks that bundle Nvidia GPUs, orchestration software, application programming interfaces and related services. The companies said the rollout will deploy tens of thousands of Nvidia processors, beginning with a 40-megawatt data center about 20 miles south of Paris that will house roughly 18,000 of the latest chips. Mistral also introduced “Magistral,” marketed as Europe’s first domain-specific, multilingual reasoning model and available in both open-source and enterprise versions with compliance features aimed at government and corporate users. The Paris-based firm, which the Wall Street Journal reports is on course to surpass US$100 million in annual revenue, positions the product as an alternative to U.S. AI offerings. French President Emmanuel Macron hailed the Nvidia-Mistral collaboration as “historic,” saying it will create hundreds of jobs and bolster Europe’s quest for technological sovereignty. Huang, for his part, called artificial intelligence “the greatest equalizer” and urged European governments and companies to build and run AI infrastructure locally rather than outsource it.
IA : à Vivatech, Emmanuel Macron célèbre le partenariat « historique » entre Mistral IA et Nvidia https://t.co/s9AJSmi9n6 #artificialintelligence, #datascience, #datascience #ds, #machinelearning, inoreader
$AMD introduces "Helios" AI rack featuring Instinct MI400 GPUs and EPYC 'Venice' CPUs at Advancing AI 2025. A Helios rack with 72 MI400 GPUs delivers approx. 31 exaflops of tensor performance, 31 TB of HBM4 memory, and 1.4 PBps of bandwidth. This positions AMD as a direct https://t.co/QRFwcjD3hx
Pour Jensen Huang, patron de Nvidia, assure que la France dispose "d'équipes incroyables de chercheurs en IA" https://t.co/fJKSrsrjCP