NVIDIA CEO Huang Renxun publicly opposed the United States’ ban on Chinese open-source AI. He said that Chinese models such as Kimi perform excellently, and that U.S. companies “should of course use them.” Restricting these models could, in turn, weaken U.S. security.

Huang Renxun believes companies can put downloaded models into secure sandboxes and control their own data and access permissions. Open weights also make it easier for external researchers to check for vulnerabilities. Compared with everyone relying on the same model, this approach makes it less likely to cause a single point of failure.

He also doesn’t think Kimi will squeeze out OpenAI and Anthropic. Low-priced, even free, models would expand the AI market and drive more demand for chips, data centers, and computing power. He said outright that free AI is good for hardware. #Gate事件合约首发狂欢