Nvidia CEO Jensen Huang has thrown himself into the middle of a growing policy war over Chinese artificial intelligence, arguing that American companies should be free to use open-source models from China, a position that puts him directly at odds with both the Trump administration and his own largest customers.
The debate was triggered by the release of Kimi K3, an open-weights model from Beijing-based Moonshot AI that delivers near-frontier performance at a fraction of the cost of leading American models. The launch rattled chip and AI stocks and revived a long-simmering question in Washington: should US companies be allowed to download and use Chinese AI models on their own infrastructure?
Treasury Secretary Scott Bessent told Fox Business that the administration is actively examining Chinese models for evidence of stolen US intellectual property and considering sanctions. OpenAI and Anthropic, both major Nvidia customers, have separately warned officials that Chinese competitors have extracted capabilities from their models and that Beijing’s growing open-source ecosystem threatens American AI leadership.
Huang dismisses the alarm.
“These Chinese models are excellent. Open-source models that are excellent should be used,” he said, speaking at the China International Supply Chain Expo. “Free AI should be great for hardware. Free AI should be great for chips. Free AI should be great for data centers.”
The argument is not purely ideological. More AI usage, regardless of origin, means more demand for Nvidia’s accelerators. Huang sees a rising tide that lifts his company’s hardware sales, not a competitive threat.
But his more interesting argument is about security. Huang rejects the premise that downloaded Chinese models create a backdoor into American systems, companies can inspect, customize, and run them inside secure environments, he points out. Openness, in his view, improves security rather than weakening it, because outside researchers can audit the model’s behavior and identify weaknesses.
“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,” Huang argued. It is a direct response to the push for consolidated, proprietary AI infrastructure that companies like OpenAI and Anthropic represent.
On the question of intellectual property theft, Huang acknowledged that misconduct should be addressed, but targeted at the specific violation, not the model itself. “Distillation, learning from AI, learning from other sources of knowledge, is fundamental to intelligence,” he said, pushing back on the idea that training on a competitor’s outputs constitutes theft per se.
The divide exposes a fundamental fracture in the US AI ecosystem. The government and frontier AI labs want containment, legal and technological barriers to prevent Chinese models from gaining a foothold in American enterprises. Huang wants openness, because openness drives adoption, and adoption drives hardware spending. The two visions are in direct conflict, and the outcome will shape not just the AI model market but the multi-trillion-dollar data center buildout that supports it.
Sources: Jensen Huang Says U.S. Firms Should Use Chinese AI Models (Yahoo Finance / Quartz, July 22, 2026); Jensen Huang argues American companies should be allowed to use Chinese AI models (Tom’s Hardware, July 23, 2026)