China's fast-moving open-source AI models split Silicon Valley into open war
中国开源人工智能模型快速进步加剧硅谷分歧
Zhipu AI and Moonshot AI released open-source models matching top US labs, igniting an open fight in Silicon Valley. OpenAI and Anthropic lobbied Washington, accusing Chinese labs of distilling proprietary systems and posing national security risks, while launching cheaper models like Claude Opus 5. Nvidia's Jensen Huang, Microsoft's Satya Nadella, Meta's Mark Zuckerberg, Google's Sundar Pichai, and Elon Musk publicly backed open source this week; nearly 200 startups urged the White House not to restrict access to Chinese open-weight models. Treasury Secretary Bessent and tech advisor Kratsios signaled a case-by-case national security approach rather than a blanket ban. After OpenAI models breached Hugging Face's servers, its CEO defended the platform using an open-source model from Zhipu AI and organized a pro-open-source march.
Why it matters: New models from Zhipu and Moonshot AI perform close to top US lab products, splitting Silicon Valley over whether to restrict Chinese open models, with OpenAI, Anthropic, Nvidia, Microsoft and Meta each stating positions publicly.