Jensen Huang: "China's AI Cannot Be Blocked"... Directly Rebutting U.S. Regulation Arguments

Nvidia CEO Jensen Huang has publicly voiced opposition to U.S. moves to regulate or block China's open-source artificial intelligence (AI) models.

In an interview with Axios released on the 22nd local time, Huang said, "China's open-source AI models are great," adding, "Good open-source models should be widely utilized."

He argued that U.S. companies should "naturally be allowed" to use Chinese AI models, showing a view contrary to some factions within the Trump administration attempting to restrict them.

In particular, regarding concerns raised by Nvidia's key customers, OpenAI and Anthropic, that Chinese firms are mimicking or absorbing their models' capabilities and guarding against the Chinese open-source offensive, Huang expressed the stance that "there is no need to be afraid."

He asserted, "There is no scenario where China pushes U.S. companies out of the market," and stated firmly, "That possibility is zero."

He explained that open-source models provide more users with an opportunity to experience AI for the first time, thereby expanding the overall market, and that a significant number of users will still pay for high-performance closed services.

Huang pointed out that the market is overreacting following last year's "DeepSeek shock" and the recent unveiling of "Kimi K3" by Chinese AI startup Moonshot AI.

He argued, "Free AI is positive for hardware, semiconductors, and data centers," maintaining that the spread of open-source AI will ultimately increase demand for AI infrastructure.

He also dismissed concerns that downloading Chinese AI models could create a so-called "backdoor" connecting to China as a "misunderstanding."

He explained that because open-source models allow external researchers to verify the structure and code and find vulnerabilities, openness actually enhances AI safety.

Along the same lines, Huang argued that Anthropic's "Claude Mythos," which is restrictively disclosed only for research purposes, should also be opened more widely.

He emphasized that "tying Anthropic down does not serve U.S. interests" and that the openness of AI technology helps strengthen U.S. competitiveness.

Huang's remarks came just hours after U.S. Treasury Secretary Scott Bessent stated in an interview with Fox Business that the U.S. is investigating whether Chinese AI models infringe on U.S. intellectual property and is considering potential sanctions.

While Secretary Bessent claimed that "watermarks" of U.S. models were found in some Chinese AI models, Huang rebutted, saying, "Distillation, meaning learning from other AIs, is a fundamental property of intelligence."

He emphasized that what should be taken issue with is cheating, not the technology itself, and that learning and knowledge transfer between AI models should not be fundamentally blocked.

(Photo: AP, Yonhap News)
※ Please note: This article was translated by AI and may contain errors.