Jensen Huang Calls AI Distillation 'Competition' as White House Calls It 'Theft'

Nvidia CEO Jensen Huang said on Monday that AI model distillation — training AI systems on the outputs of other models — is "competition," directly contradicting the White House position that the practice amounts to theft.

"That's called competition," Huang told CNBC's Squawk Box when asked whether distillation was "not robbery."

The comments place the head of the world's most valuable chip company at odds with the U.S. government on an issue that has become a flashpoint in the technology rivalry between Washington and Beijing.

A public split over a contested practice

Treasury Secretary Scott Bessent described distillation as "theft" in July and threatened sanctions against overseas companies that use the technique to extract capability from U.S.-built models.

Huang sees it differently. "You're allowed to test somebody else's products all you want," he said. He noted that Nvidia's own products are sometimes stripped "down to bones" as companies tried to learn how they worked.

He acknowledged the discomfort that comes with it. "I'd really prefer they didn't," Huang said. "I'd really prefer that nobody learns from our products, and we have the benefit of just cruising along. But, you know, frankly, competition makes everything better."

He also offered a blunt prescription for companies that object to how their products are used. "If you don't like that, if you don't like people to use your products, all you [have to do is] know your customers, and disable the service," Huang said.

The White House did not immediately respond to a request for comment from CNBC.

Why the dispute matters

Distillation sits at the center of escalating tensions between the U.S. and China in their race for AI supremacy. U.S. officials have repeatedly accused Chinese AI companies of using the technique to catch up with American frontier models without bearing the cost of training them from scratch.

Earlier this month, the Cybersecurity and Infrastructure Security Agency (CISA) accused China's AI companies of conducting "industrial-scale knowledge distillation campaigns" that violated U.S. companies' terms of use.

Anthropic said earlier this month that it found two of China's most prominent AI developers — Alibaba, which builds the Qwen family of models, and DeepSeek — engaging in "illicit distillation."

China has rejected the claims.

Commerce versus policy

Huang's stance reflects the commercial position of a company that sells chips and systems on both sides of the Pacific and has watched U.S. export controls repeatedly constrain its China business. For Nvidia, framing distillation as fair competition keeps the door open in a market that Washington keeps trying to close on national security grounds.

For the White House, the stakes run in the opposite direction. If distillation allows Chinese firms to replicate U.S. model capabilities at a fraction of the cost, export controls on advanced chips lose much of their force — a policy concern that Bessent's sanctions threat was designed to address.

The disagreement between one of the AI industry's most influential executives and the Treasury Department illustrates how unsettled the rules of the AI race remain. With CISA and Anthropic documenting alleged distillation at scale, and Huang arguing that enforcement is a matter for product owners — not governments — the question of who polices model outputs, and how, is likely to define the next phase of U.S.-China AI policy.