AI · Nov 12, 2026
Jensen Huang’s exit gives Nvidia its first board chair, and that governance change is the real storyNvidia’s open-model push loses its champion, and the Hugging Face deal becomes the test
· Updated · Signal Desk newsroom
Analysis of reporting by TechCrunch. How we work

Nvidia’s former CEO argued for open models, built Nemotron, backed Reflection AI and Thinking Machines Lab, and led a 25-company letter against over-restricting open weights. With an interim CEO and a pending $13 billion Hugging Face deal, the question is whether openness was strategy or personality.
Key takeaways
- He framed open models as good for safety, innovation and sovereignty and pushed Nvidia’s Nemotron open-weight family.
- He led an open letter signed by 25 companies, including Microsoft, Meta, IBM and Palantir, urging Washington not to over-restrict open-weight models.
- Nvidia hasn’t finalized its $13 billion Hugging Face acquisition, and interim CEO Colette Kress hasn’t signaled her approach.
- An insider told TechCrunch they don’t expect the change to affect Nemotron development.
- Nvidia also agreed to pay $6 billion for a non-exclusive license to Poolside’s technology.
What the open-source community lost
The Nvidia CEO resigned after admitting an affair with an alleged Chinese spy (our coverage), and TechCrunch argues open-source AI just lost one of its most powerful allies. He framed open models as good for safety, innovation and sovereignty, pushed Nvidia’s Nemotron open-weight family (recent releases), invested in Reflection AI and Thinking Machines Lab, and led an open letter signed by 25 companies, including Microsoft, Meta, IBM and Palantir, urging Washington not to over-restrict open-weight models. Researcher Nathan Lambert posted that his departure would be very bad for the U.S. open-model cause.
Open models let universities, startups, governments and smaller companies build without permission from a handful of labs. It matters that the most influential American counterweight to closed providers was, in practice, one person with an enormous company behind him.
A debate losing balance
TechCrunch frames him as the strongest counterweight to Anthropic and OpenAI, which have argued for tighter restrictions on advanced models on safety grounds. Those arguments aren’t wrong in every case, but the policy debate loses balance just as Washington weighs how to respond to allegations that Chinese labs take IP from American ones (coverage). The scandal hands ammunition to anyone arguing for tougher rules, because his push to loosen chip export controls is now under a cloud (analysis).
Incentives may outlast the founder
Open source wasn’t just idealism at Nvidia. As big customers like Google and OpenAI build their own chips, Nvidia has every reason to grow a wide base of organizations that train, tune and run their own models. More players means more GPUs and less dependence on a few labs. An insider told TechCrunch they don’t expect the leadership change to affect Nemotron, and Hugging Face CEO Clem Delangue says his company will keep fighting for open source whatever happens (his response).
Still, founder priorities often get trimmed when a company shifts into risk-management mode, and an interim CEO steering through a national-security scandal has reason to avoid controversy. The Hugging Face deal is the clearest test (why Nvidia bought it). If it stalls, caution is winning. If it closes on schedule, openness is company policy, not one man’s project.
The harder question
Restrictions on Chinese open-weight models could, as TechCrunch notes, weaken competition for American companies whose most powerful players keep models closed. A personal scandal shouldn’t become the reason the U.S. narrows its own open ecosystem. Policy should rest on evidence about risk, not on who is currently embarrassed. Open infrastructure such as llama.cpp’s new stable release shows how much now depends on that ecosystem (coverage). More in our NVIDIA coverage.
Who feels it
- Open-model developers
- Nemotron reportedly continues, but the community loses its most visible champion.
- Hugging Face
- Its pending acquisition depends on how interim leadership weighs risk.
- Policymakers
- Arguments for restricting open weights, especially from China, gain momentum.
- Closed-model labs
- Less organized resistance to tighter model restrictions.
What to watch
- Whether the $13 billion Hugging Face acquisition closes.
- Any statement from Colette Kress on open models.
- New Nemotron releases.
- Washington moves on Chinese open-weight models.
FAQ
Will Nvidia keep supporting open-source AI?
An insider told TechCrunch they don’t expect the leadership change to affect Nemotron, but interim CEO Colette Kress hasn’t addressed it publicly.
Is Nvidia’s Hugging Face acquisition still happening?
It hadn’t been finalized at the time of the CEO’s exit; its completion is seen as a test of Nvidia’s commitment.
What was the open-weight letter Nvidia led?
A letter signed by 25 companies, including Microsoft, Meta, IBM and Palantir, urging Washington not to over-restrict open-weight models.
Companies: NVIDIA
Also covering this
AI · Nov 12, 2026
Nvidia’s rivals won’t win overnight, but Huang’s exit speeds up every custom-chip programAI · Nov 12, 2026
Beyond Nvidia: the open-weight, agent-safety and financing causes that just lost their organizerAI · Nov 12, 2026
Who is Colette Kress? Nvidia’s interim CEO is a finance veteran built for a trust crisisMore on NVIDIA
AI · Oct 28, 2026
RTX Spark laptops get open speech models and a dial for background agentsAI · Nov 5, 2026
NVIDIA’s four open Cosmos 3 Nano models let robot and driving teams simulate worlds that reactAI · Nov 9, 2026
Nvidia’s Korean startup bets span interconnect, open models, robot hands and document AIAI · Nov 8, 2026
Nvidia’s Dynamo recipes speed four Korean open models to production, tuned for B200sAI · Nov 5, 2026
Nvidia profiles Chris Alexiuk, the YouTube teacher making the case that open AI means weights, data and recipes