Nvidia: ‘Sovereign AI’ will change digital work

Nvidia executives think sovereign AI has the potential to change digital work as generative AI (genAI) aligns with national priorities and local regulations.

“AI is rapidly becoming critical infrastructure, just like electricity and the internet,” Kari Briski, vice president of generative AI software for enterprise at Nvidia, said during a briefing ahead of this week’s GTC Paris developer event.

Countries need localized models reflecting unique cultural identities and that can understand the local language, work culture, regulations and strengths, Briski said.

While genAI agents can help customize language models, those models also need a localized distribution path for regional use, she said. With that in mind, Nvidia is partnering with a number of countries to shape their sovereign AI plans and working with local companies that have unique knowledge of their regions.

“These agents are our digital workforce,” Briski said. “And for those concerned about AI taking our jobs, these agents will not. But those who know how to use AI might.”

The company made a variety of announcements ahead of GTC Paris cloaked in the theme of sovereign AI. Many focused on industrial AI in Europe, which houses some of the world’s largest automotive, chemical and aerospace firms.

“For these industries, [large language models] will only go so far,” said Rev Lebaredian, vice president of Omniverse and simulation technology at Nvidia. “They need physical AI — AIs that can understand and interact with the physical world, fine-tuned for industrial applications.” 

Nvidia announced it is building the world’s first industrial AI cloud in Germany, where many industrial giants are located. The industrial AI cloud is built on the company’s RTX PRO and B200 GPUs.

It will “run every industrial manufacturing workload from every industrial software maker from accelerated design and engineering to factory digital twins to robotics training and simulation and AV development,” Lebaredian said.

The company also announced a partnership in France with Mistral to build out AI hardware and software infrastructure in France. Mistral is building a data center with 18,000 Grace Blackwell systems that’s expected to be operational in 2026.

While Nvidia’s rhetoric could help it sell GPUs, sovereign AI is a real consideration among countries hoping to gain a competitive advantage, and it will trickle down to office workers, analysts said. Specifically, sovereign AI is linked to reskilling local workforces in industries that are priorities for governments.

Country-specific models will be a backbone of knowledge on how companies can do business within borders, said Jim McGregor, principal analyst at Tirias Research. “Finance, health care and government are typical applications, but this could extend to companies with strict security requirements,” he said.

Nations are actively trying to gain an information advantage with sovereign AI strategies, said Jack Gold, principal analyst at J. Gold Associates. “Countries are seeing sovereign AI as infrastructure, just like weapons and power plants. Countries don’t want to be beholden to others — it’s strategic,” he said.

Sovereign AI comes in the form of localized models and AI infrastructures, and is designed to create jobs and upskill workers in AI to get used to the emerging work environments, Gold said.

Germany is prioritizing investments in AI technology to regain its industrial advantage, which will boost the local workforce, Gold said.

Specifically, Nvidia said the Mercedes-Benz CLA sedan with its Nvidia Drive autonomous driving AI software is now available in Europe. It also announced that Volvo Cars and JLR are also starting production for their next generation fleets built on the Nvidia Drive AI software stack.

The company also announced that Siemens is integrating Nvidia’s CUDA-X RTX and Omniverse libraries in its simulation software used to engineer products that include machines, engines and other equipment.