According to Dongcha Beating's monitoring, NVIDIA CEO Jensen Huang publicly opposed the U.S. banning Chinese open-source AI. He mentioned that models like Kimi from China have shown excellent performance, and U.S. companies "should certainly use" them. Restricting these models could instead weaken U.S. security.
Jensen Huang believes that companies can put downloaded models into a secure sandbox, allowing them to control data and access permissions. Open-sourcing the weights also makes it easier for external researchers to check for vulnerabilities. This approach, compared to everyone relying on the same model, is less susceptible to single-point failures.
He also does not think Kimi will overshadow OpenAI and Anthropic. The availability of low-cost or even free models will expand the AI market, driving demand for more chips, data centers, and computing power. He stated clearly that free AI is beneficial for hardware.
