Nvidia CEO Jensen Huang defends Chinese open-source AI models

Nvidia CEO Jensen Huang is pushing back against calls to restrict Chinese AI models in the United States, arguing that openness strengthens rather than threatens American competitiveness and security.

In an interview with Axios conducted in Fort Worth, Texas, Huang said American companies should be allowed to use Chinese open-source models like Kimi K3 from Beijing-based Moonshot AI. His position directly contradicts Trump administration officials and major AI labs including OpenAI and Anthropic, which have lobbied for restrictions on Chinese models they claim contain stolen intellectual property.

"These Chinese models are excellent," Huang said. "Open-source models that are excellent should be used."

Why it matters

Nvidia supplies the chips powering most AI development worldwide, giving Huang unique visibility into how the technology ecosystem actually functions. His defense of Chinese models carries weight because restrictions could reshape global AI competition and determine whether open-source development remains viable—with major implications for innovation speed, cost structures, and which companies dominate the next computing era.

The Kimi panic

Kimi K3's release triggered intense market volatility, combining frontier-level performance with dramatically lower prices and open weights that developers can download and modify. The announcement contributed to sell-offs of Nvidia and other chip stocks as investors questioned whether efficient, cheaper models would undercut demand for expensive AI infrastructure.

Huang believes the market reaction misses the point. "Free AI should be great for hardware," he said. "Free AI should be great for chips. Free AI should be great for data centers." His argument: accessible models expand the total addressable market by bringing AI to more users and businesses, ultimately driving greater demand for computing power.

Security through openness

On national security concerns, Huang rejected what he called the "misconception" that downloaded Chinese models create backdoors to Beijing. He argued that companies can customize open models and control their access within secure environments.

More fundamentally, Huang contended that openness improves security by allowing outside researchers to inspect models, identify vulnerabilities, and build defenses. "If everything just becomes one single model, one single point of attack, one single source of failure, I think the world is much, much more vulnerable," he said.

He applied this logic to American models as well, calling for Anthropic to make its restricted cyber model, Claude Mythos, available more broadly rather than limiting access.

Intellectual property questions

Huang's comments came hours after Treasury Secretary Scott Bessent told Fox Business that the administration is examining Chinese models for stolen U.S. intellectual property and considering sanctions. Bessent cited "watermarks" of U.S. models found inside Chinese ones.

Huang drew a distinction between legitimate learning and misconduct. "Distillation, learning from AI, learning from other sources of knowledge, is fundamental to intelligence," he said, while acknowledging companies should face consequences for violating privacy or contracts.

A different race

Huang also challenged Washington's framing of AI competition as a winner-take-all sprint. He dismissed the idea of "a race with an endpoint," noting that both the United States and China will continue using AI indefinitely. "There's no scenario where China runs U.S. companies off the road," he said. "Zero possibility."

The details were first reported by Axios in an exclusive interview with Mike Allen.