AivexaNewsSearch
AI news for builders and product teamsChecked every hour

The open-source movement wants Jensen Huang back

Collected Nov 16, 2026

A petition calling for Jensen Huang to be reinstated as Nvidia's CEO began circulating on social media on Friday, and by Monday it had drawn more than 115,000 signatures. Among the notable names were Elon Musk and several influential Nvidia researchers. A large share of the signers, though, come from one corner of the industry: executives at Hugging Face, the model-sharing hub, and at the venture firms Andreessen Horowitz and Sequoia Capital. Huang resigned on Thursday, the same day the petition effort started to spread.

The composition of that signature list is the tell. Hugging Face hosts downloadable open-weight models that anyone can run without going through a large lab, and the two VC firms have backed startups built on top of them. For those backers, a world where businesses train and serve their own models rather than paying OpenAI or Anthropic is both an ideological preference and a commercial one. Huang had become the most prominent industry figure making that case to policymakers, arguing against the view that open models ship without guardrails and hand AI leadership to China, which produces many of the strongest open models. The petitioners want that voice back in the room.

The timing is what makes the campaign awkward. Also on Thursday, former OpenAI researcher Will DePue published findings online suggesting Chinese AI companies had access to Nvidia's advanced Blackwell chips on a much larger scale than previously estimated. China is barred from obtaining Nvidia's most cutting-edge chips as part of US efforts to preserve an AI lead. How the chips would have reached those companies is not established. DePue says senior researchers at OpenAI and Anthropic verified the results, but no independent analysis has confirmed them. Neither DePue nor Nvidia provided comment.

The findings do not show that Nvidia helped Chinese firms get the hardware, and chip smuggling was already understood to be a problem well beyond any single vendor's control. Even so, the suggestion that Nvidia profited from the growth of Chinese AI startups — whose models are increasingly chosen by American companies over domestic alternatives — puts Huang in a difficult spot. He has argued that open source is how the US keeps its technical edge; last December he said that without it startups cannot thrive, university researchers cannot do research, AI cannot be taught and scientists cannot use it.

Reinstatement already looked improbable. US Treasury Secretary Scott Bessent said last week that Nvidia should not look inside the company for a replacement. The new chip findings, as reported, make a return even less likely.

Two other threads matter for anyone building on open models. Nvidia announced in September that it would acquire Hugging Face for nearly $13 billion, tying the chipmaker's fortunes more tightly to the hub where those models live. And Nvidia's open-source advocacy has a plain financial logic: every company that trains its own model instead of renting from Anthropic or OpenAI is another buyer of GPUs.

A new rule from the US Department of Commerce, announced Monday, now threatens the supply side of that ecosystem. The rule targets Chinese companies training models on Nvidia chips they do not own but reach remotely, in data centers in countries such as Malaysia. That remote-access workaround is part of why Chinese open models have become as capable as they are. If enforced, the practical effect is likely to fall on training runs rather than on inference, since a model already released as weights can be downloaded and served anywhere. This suggests the near-term impact lands on the next generation of Chinese open models rather than on the ones developers use today.

The political backdrop is a split inside the administration and its orbit over how to treat Chinese open models, which are cheap, widely downloaded and often competitive with US systems. Huang's usefulness to the open-source camp was his access to the people writing those rules. With him out, that camp has lost its most recognizable spokesperson at exactly the moment two policy pressures — export enforcement and chip provenance — are converging on its supply chain.

Why it matters: Teams that depend on open-weight models from Hugging Face, or on Nvidia hardware to train them, face two sources of uncertainty at once — a Commerce rule aimed at remote chip access and a leadership vacuum at the company that just agreed to buy the hub. Developers can still download and run existing open models freely, but the pipeline that produces the next strong ones may slow. My reading is that the petition's outcome matters less to builders than the enforcement details still to come: which data centers and which chip generations the rule covers will determine whether open models keep improving at the current pace.

Read at MIT Technology Review · AI

Based on reporting from the original publisher. Visit the source for full context and later updates.

Publisher excerpt

This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here. On Friday, a petition to reinstate Jensen Huang as CEO of Nvidia began circulating on social media, and by Monday it listed some notable signatories, including Elon Musk and several influential Nvidia…