Huang’s departure puts Nvidia’s open-model stance in focus

Huang resigned Thursday after admitting to an affair with Alice Huang, a Megaspeed executive described in the source as an alleged Chinese spy. The scandal has renewed scrutiny of his calls for softer U.S. chip export controls on China and raised national-security concerns. Nvidia has named Chief Financial Officer Colette Kress interim CEO; the company has not said how she will approach the proposed Hugging Face deal or the open-model community. TechCrunch said it had contacted Nvidia for comment.

Huang had promoted open source AI as a force for safety, innovation, diffusion and sovereignty. He also made it part of Nvidia’s product and investment strategy. The company developed the Nemotron family of open-weight models for long-running agentic systems and reasoning workloads, and Huang backed U.S. AI startups Reflection AI and Thinking Machines Lab. Nvidia has also agreed to pay $6 billion for a non-exclusive license to technology from AI coding startup Poolside.

The company has yet to finalize a proposed $13 billion acquisition of Hugging Face. Its CEO, Clem Delangue, wrote on X that the platform would continue to support open source regardless of what happens next. Open-model researcher Nathan Lambert said Huang’s departure would be “very very bad for the open model cause in the US.”

Earlier this year, Huang signed an open letter with 25 companies, including Microsoft, Meta, IBM and Palantir, urging the U.S. government not to impose excessive restrictions on open-weight models, many of which come from China. The letter came amid debate over allegations that Chinese AI labs had stolen intellectual property from U.S. companies, including Anthropic and OpenAI. Those firms have reportedly urged Washington to strengthen protections. Their calls for restrictions, framed around safety, have clashed with the position of open-model advocates.

A leadership change could weaken the community’s influence in Washington, while tougher limits on Chinese models could reduce competition for U.S. companies whose most powerful models remain closed. But Nvidia has a commercial reason to keep the ecosystem growing: as major customers such as Google and OpenAI develop their own chips, a wider base of developers and organizations training, fine-tuning and running models could reduce Nvidia’s reliance on a few frontier labs and create more demand for its GPUs.