Every country trying to escape American AI dominance is buying it wholesale. Nations are pouring billions into sovereign AI — domestic models, local data centers, infrastructure meant to keep sensitive data within their own borders — and nearly all of it runs on chips from one company: Nvidia Corp. (NASDAQ:NVDA).
According to Counterpoint Research, 92% of sovereign large language models use the company’s chips. The firm studied more than 80 countries outside the U.S. and China. Roughly 55 had developed domestic large language models.
But domestic does not necessarily mean independent.
"We have observed in our Sovereign AI LLM research that hosting is increasingly Sovereign while the AI chip source is not," Counterpoint Research Director Marc Einstein said.
That distinction gets to the heart of the investment story.
Nvidia Keeps Inking Multi-Billion-Dollar Partnerships
Countries can control where models run and where data is stored. Yet the computing infrastructure underneath remains overwhelmingly concentrated around Nvidia.
Counterpoint attributes that dominance to Nvidia’s advanced silicon and mature CUDA software ecosystem.
"As the Sovereign AI infrastructure expands further, Nvidia continues to strike extensive multi-billion dollar partnerships across geographies," Einstein said.
That creates a difficult cycle for competitors to break. Developers use Nvidia because its ecosystem is mature, while every new deployment makes that ecosystem harder to leave.
Nvidia’s growing importance across AI infrastructure is already pushing demand beyond GPUs into memory, networking, cooling and power equipment.
AMD Is Small, But Strategically Positioned
Counterpoint estimates Advanced Micro Devices Inc. (NASDAQ:AMD) holds just a 4% share of sovereign AI infrastructure.
Still, that small footprint could matter.
Europe’s flagship LUMI supercomputer in Finland runs on AMD architecture and supports several European AI initiatives. AMD has also expanded through partnerships in Australia and South Korea.
Einstein said AMD is building momentum through Moreh, Motif and Upstage.
For AMD, the opportunity is less about overtaking Nvidia than benefiting from governments seeking greater supplier diversification.
Why Investors Should Care
The first AI investment cycle was driven by hyperscalers building enormous data centers.
The next one may be driven by nations.
Governments are becoming AI infrastructure buyers and, for now, most of them are buying Nvidia.
Image: Shutterstock