Nvidia powers 52% sovereign projects

Nvidia sits at the center of a market that is supposed to reduce dependence on outside technology. Benzinga reported on May 22 that Nvidia GPUs are used in 52% of tracked sovereign AI infrastructure projects worldwide, citing research from the Center for a New American Security, or CNAS. (cdn4.benzinga.com) That number matters because “sovereign AI” usually refers to a country’s effort to keep AI infrastructure, data, and strategic control closer to home. (interactives.cnas.org) CNAS describes sovereign AI as a stack that spans infrastructure, models, and data, with infrastructure projects making up 59% of all projects in its index. (nvidia.com) ### What does the 52% figure actually measure? CNAS tracks government-led sovereign AI projects across infrastructure, models and data. In the infrastructure slice, it includes AI data centers, supercomputers, GPU clusters and compute-access programs. Benzinga said Nvidia supplies GPUs for 52% of those tracked infrastructure projects. (cdn4.benzinga.com) That does not mean Nvidia controls 52% of all sovereign AI spending or 52% of every national AI program end to end. It means that in the infrastructure projects CNAS tracks, Nvidia hardware appears more often than any other vendor. ### Why are “sovereign” projects still buying Nvidia? (interactives.cnas.org) Nvidia has spent the past two years pitching “sovereign AI” directly to governments. On its public-sector site, the company says nations need domestic AI capabilities for economic growth, national security, cultural preservation and innovation, and it offers products, technical briefs and case studies aimed at public-sector buyers. (interactives.cnas.org) CNAS has also written that governments are increasingly treating AI as infrastructure rather than only as a policy problem. That helps explain why national programs often start with compute procurement first, even when the political goal is autonomy. (interactives.cnas.org) ### Where is the spending concentrated? CNAS data summarized in secondary coverage shows more than 130 government-led sovereign AI projects worldwide. The same reporting says infrastructure is the largest category and that more than 80% of disclosed investment is concentrated in the Middle East and East Asia, with the United Arab Emirates and Japan accounting for more than two-thirds of disclosed funding. (nvidia.com) That regional concentration helps explain why sovereign AI is showing up first as a capital-intensive infrastructure story. Large state-backed programs in a handful of markets can shape vendor share quickly because GPU clusters and AI data centers require large, early hardware commitments. (cnas.org) This is an inference based on the concentration and infrastructure data reported by CNAS and summarized in secondary coverage. ### Does this weaken the idea of sovereignty? The tension is built into the category. CNAS defines sovereign AI around domestic control of infrastructure, data and capability, while Nvidia is an American supplier whose chips remain embedded across many of those efforts. (finance.biggo.com) That does not make the projects meaningless. It means sovereignty in AI often starts with where systems are deployed, who governs access, where data stays and who can operate the stack, rather than with a fully domestic hardware supply chain. Nvidia’s own materials frame sovereign AI in those terms, emphasizing domestic infrastructure, local data and national capacity-building. (interactives.cnas.org) ### Why should enterprise and media buyers care? Regional control is becoming part of the buying conversation beyond governments. CNAS says infrastructure dominates the sovereign AI projects it tracks, and Nvidia is marketing country-specific deployments, including an expansion in South Korea involving more than a quarter-million GPUs across sovereign clouds and AI factories. (cnas.org) For companies with compliance, residency or public-sector customers, that means sovereignty questions may increasingly focus on deployment geography and operating control even when the hardware vendor is unchanged. The next places to watch are the CNAS Sovereign AI Index for new project disclosures and Nvidia’s government-focused sovereign AI program for announced country deployments. (nvidia.com) (interactives.cnas.org)