Nvidia CEO Jensen Huang says China excels in AI research

Jensen Huang has a talent theory, and it does not flatter Washington.

The Nvidia CEO says China can scale its AI research workforce better than any other country. He framed it as a manufacturing story, applied to the people who build the models, not the chips.

Speaking with Yahoo Finance at GTC Berlin in October 2026, Huang said nearly half of the world’s AI researchers are of Chinese origin. His summary was blunt.

“Nobody manufactures AI researchers like China does.”

What Huang actually said

Huang’s argument went beyond headcount. He pointed to Chinese strength across three fronts: researchers, chip production, and the development of open AI models.

He described China as a formidable competitor to the United States in AI. That framing matters, because it treats the race as a contest between peers rather than a leader and a fast follower.

Huang did not stop at diagnosis. He called on the US to keep working to attract talent from around the world.

He also urged continued dialogue between the US and China on AI safety risks.

The numbers behind the claim

Huang’s estimate lines up with recent research. A September 2026 analysis from the Carnegie Endowment tracked elite AI talent through authors of NeurIPS 2025 papers.

According to the Carnegie study, 57% of those elite researchers earned their undergraduate degrees at Chinese institutions. That figure was 46% in 2022.

The study found a second shift that may be even more consequential. China has overtaken the US as the main place these researchers actually work.

Carnegie’s data shows 41% of the tracked elite researchers employed in China, compared with 34% in the US.

The analysis credits Chinese universities such as Tsinghua and Peking with rising to prominence in training elite researchers. That rise has come as US visa policies for STEM students have tightened.

Why Nvidia has skin in this game

Huang is not speaking as a neutral academic. Nvidia has felt the US-China rivalry directly on its income statement.

He has noted that Nvidia’s share of the AI GPU market in China fell from approximately 95% to zero after US export controls took effect.

The vacuum did not stay empty. Huawei has used the opening to build a stronger domestic position in China’s AI market.

Huang has made this case repeatedly. He has consistently tied the rising share of Chinese-origin researchers to the wider competitive picture between the two countries.

His position is that the US needs to engage with talent globally. Walking away from markets and people, he argues, weakens America’s hand rather than strengthening it.

What this means

Policy is the other lever. Tighter visa rules for STEM students may be intended to protect national interests, but the Carnegie data suggests they coincide with researchers choosing to build careers elsewhere.

Policymakers face a genuine tension. Export controls and visa restrictions aim to slow a rival, yet Huang’s argument is that they may also weaken American companies and accelerate domestic alternatives like Huawei.

Huang singled out China’s work on open AI models as a strength. Open models spread quickly once released.

Huang’s call for US-China dialogue on AI safety adds another layer. Even as the two countries compete hard for talent and market share, he sees shared risks that neither can manage alone.

The next data points to track are clear enough. Watch future NeurIPS cohorts to see whether the 57% figure keeps climbing, whether the 41% versus 34% workplace gap widens, and whether US visa and export policy shifts in response.