NVIDIA CEO Jensen Huang stated on July 21 that China's open-source AI models are "excellent," and American companies "should utilize them." He also warned that restricting open-source models could potentially weaken the United States' competitiveness in the field of artificial intelligence.
According to a report by the American news website Axios on the 23rd, Huang Renxun said in an interview with the media, "These Chinese models are very excellent. Excellent open-source models should be utilized." He believed that the United States has no reason to fear the development of open-source AI from China.
Huang Renxun's stance is considered to be in stark contrast to the positions of the U.S. government and some AI companies. American AI companies, including those associated with NVIDIA, have warned Washington that Chinese competitors could potentially obtain U.S.-made models through techniques such as "distillation," and have threatened to undermine America's leading position in the AI field.
Huang Renxun does not agree with this. He stated that open-source models will not cause American companies to be "driven out of the market by China". "This is absolutely impossible, there’s no possibility of it happening."
Huang Renxun said that the market previously misjudged the impact of Chinese AI models. Regarding the market concerns caused by the Kimi K3 model launched by Chinese company Yuezhin Meinan, he stated that Wall Street and US policymakers have "misunderstood the implications once again."
In his view, cheaper and more open AI models will encourage more companies and individuals to use artificial intelligence, ultimately increasing the demand for NVIDIA chips, data centers, and computing resources. ‘Free AI should benefit hardware, chips, and data centers.’
Reports indicate that the Kimi K3 model has recently caused a new wave of excitement in the AI industry. Its features include performance close to cutting-edge models, lower costs, and an open weight structure. Some concerns suggest that such low-cost models could weaken the commercial rationale behind large-scale AI infrastructure construction in the United States.
Regarding the US's restrictions on Chinese AI models under the pretext of safety, Jensen Huang also raised doubts.
He pointed out that an open model does not mean there is a so-called ‘backdoor’. Enterprises can control risks through methods such as secure isolation environments. On the contrary, openness allows more researchers to discover vulnerabilities and improve protection measures.
"If the world has only one model, one attack point, and one source of failure, I believe the global situation will be even more fragile," said Jensen Huang.
When Huang Renxun made the above remarks, US Treasury Secretary Steven Bisonter claimed that the US is evaluating whether Chinese AI models involve so-called 'stealing US corporate intellectual property' and is considering taking sanctions.
Huang Renxun responded that AI's continuous improvement through learning knowledge from various sources is the foundation of intelligent development. Measures should be taken against specific violations, rather than restricting the model itself.
Yahoo Finance quoted analysts as saying that Chairman Huang's public discussion of the importance of Chinese AI models is related to NVIDIA's commercial interests and may also touch on issues related to U.S. technology competition with China. His stance aligns with NVIDIA's long-term business logic. For NVIDIA, allowing more companies to run AI models at low costs means a larger computing power market.
