AI · Oct 28, 2026
RTX Spark laptops get open speech models and a dial for background agentsOpen-source AI lost its loudest Washington ally, and a petition won’t bring him back
· Updated · Signal Desk newsroom
Analysis of reporting by MIT Technology Review. How we work

More than 115,000 names, including investors and Hugging Face leaders, signed a petition to reinstate Jensen Huang. MIT Technology Review argues the bigger threats to open models are China chip findings and a new Commerce rule.
Key takeaways
- A petition to reinstate Huang as Nvidia CEO passed 115,000 names, including Elon Musk, Nvidia researchers, and figures from Hugging Face, Andreessen Horowitz and Sequoia.
- Huang had become open-source AI’s most influential advocate to policymakers.
- Ex-OpenAI researcher Will DePue published unverified findings suggesting Chinese AI companies had more access to Nvidia Blackwell chips than estimated.
- A new Commerce rule targets Chinese companies training on Nvidia chips accessed remotely in countries like Malaysia, which could slow Chinese open models.
- Nvidia agreed in September to buy Hugging Face for nearly $13 billion; open models also serve its chip-sales interest.
Who signed, and why
A petition to reinstate Jensen Huang as Nvidia CEO began circulating on a Friday and by Monday had more than 115,000 names, MIT Technology Review reports. Elon Musk and several influential Nvidia researchers signed. So did executives from Hugging Face and the venture firms Andreessen Horowitz and Sequoia. James O’Donnell’s point is that these supporters share an interest: open-source AI, models anyone can download and run without a big tech company in the middle.
Huang had become their ringleader in Washington, defending open models against concerns that they lack guardrails and hand AI leadership to China, home of the most powerful and popular open models. Losing him matters to the movement less for the chips than for the access. As the piece notes, open-source advocates’ worry about concentrated AI power would sound at home on the left, if so many weren’t also investors who profit when startups build on cheap models. Both things can be true.
The case against a comeback got stronger
The petition was always a long shot, and the reporting makes it longer. On the day Huang resigned, ex-OpenAI researcher Will DePue published findings suggesting Chinese AI companies had far more access to Nvidia Blackwell chips than estimated. DePue says senior OpenAI and Anthropic researchers verified the work. No independent analysis has confirmed it, and the findings don’t show Nvidia helped anyone obtain chips; smuggling was already known to extend beyond Nvidia’s control. But they put Huang in an awkward spot as the champion of open source as an American advantage. Treasury Secretary Scott Bessent also said Nvidia shouldn’t pick an insider to replace him. WIRED has since reported Nvidia’s board chair saying the board doesn’t intend to reinstate Huang.
That fits what our desk has tracked through the saga, from Hugging Face’s CEO vowing to keep fighting for open source to the bigger questions beyond the scandal. The personal story is over. The policy story isn’t.
The rule that could matter more than the CEO
The most important detail is a new Commerce Department rule announced Monday. It aims to stop Chinese companies from training models on Nvidia chips they don’t own but rent remotely in data centers in places like Malaysia, a workaround that helped Chinese open models get so good. If it works, the models popular on Hugging Face could stop improving as quickly. That’s a direct hit to the open ecosystem, whatever happens at Nvidia.
Open models are worth defending. They let researchers, universities and small companies build without permission. But the movement made a strategic error by tying its political case to one charismatic executive whose company profits from every model trained. Nvidia’s nearly $13 billion Hugging Face deal makes that conflict sharper. The open-source case needs advocates without a chip to sell, and it needs an answer on China that isn’t just “trust us.”
Who is exposed
- Open-model developers
- Loss of a powerful policy voice, plus a Commerce rule that may slow Chinese open models they depend on.
- Hugging Face
- Its pending acquisition by Nvidia ties its future to Nvidia’s leadership and policy standing.
- US policymakers
- Unverified Blackwell findings strengthen hawks pushing tighter controls on open models.
- Nvidia
- Its open-source advocacy now looks more like self-interest, with China questions unresolved.
What to watch
- Independent verification or rebuttal of DePue’s Blackwell findings.
- Implementation details and enforcement of the Commerce remote-access rule.
- Who emerges as open-source AI’s main voice in Washington.
- Progress on Nvidia’s Hugging Face acquisition under new leadership.
- Release cadence of top Chinese open models in coming months.
FAQ
How many people signed the petition?
More than 115,000 names, according to MIT Technology Review, including Elon Musk and several Nvidia researchers.
Did Nvidia help China get Blackwell chips?
There’s no evidence of that. Will DePue’s findings suggest wider Chinese access, but they haven’t been independently confirmed and don’t show Nvidia’s involvement.
What does the new Commerce rule do?
It aims to stop Chinese companies from training on Nvidia chips accessed remotely in other countries, such as Malaysia.
Companies: NVIDIA
More on NVIDIA
AI · Nov 5, 2026
NVIDIA’s four open Cosmos 3 Nano models let robot and driving teams simulate worlds that reactAI · Nov 16, 2026
Rubin HPC is Nvidia’s bet that science agents will be bottlenecked by simulation, not reasoningAI · 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 B200s