Huang's Nvidia exit puts open-source AI support in question

Jensen Huang’s resignation as Nvidia CEO on Thursday, following reports of an affair with a woman the U.S. suspects of spying for China, strips the open-source AI community of one of its strongest institutional backers. Huang framed open-weight models as a boost to safety, innovation, diffusion and national sovereignty, and under him Nvidia built the Nemotron family of open-weight models for long-running agentic and reasoning workloads, invested in open-model-focused startups including Reflection AI and Thinking Machines Lab, and committed $6 billion to license Poolside’s coding technology, a startup building open AI tools.

Supporting open source was never purely idealism for Nvidia — it is increasingly central to the chipmaker’s business strategy. As major customers including Google and OpenAI develop their own chips, a broader ecosystem of developers training and running models on Nvidia hardware reduces the company’s dependence on a handful of frontier labs. That logic underlies Nvidia’s still-unfinished roughly $13 billion deal to acquire Hugging Face, whose CEO Clément Delangue said on X the platform would keep fighting for open-source AI regardless of what happens next at Nvidia.

Huang’s influence went beyond Nvidia’s own roadmap: he organized an open letter, signed by 25 companies including Microsoft, Meta, IBM and Palantir, urging Washington not to over-restrict open-weight models — many of which originate in China — as the government weighs allegations that Chinese AI labs have stolen intellectual property from American labs like Anthropic and OpenAI. That mix of personal star power and Nvidia’s market weight made Huang the loudest counterweight to closed-model providers pushing for tighter restrictions. Interim CEO Colette Kress has not signaled how she will approach the Hugging Face deal or open-model support, and the scandal over Huang’s relationship with Alice Huang, a Megaspeed executive the U.S. suspects of espionage, has added fresh scrutiny to his record of favoring looser chip-export rules toward China.

Why it matters

Whether Nvidia keeps championing open-weight models now hinges less on ideology than on whether Kress or a permanent successor sees the same business logic Huang did — that a wider ecosystem of AI builders running on Nvidia chips matters more than any single frontier lab’s loyalty.