Jensen Huang's resignation as Nvidia CEO on November 12, 2026 raises one question for developers: does Nvidia's push for open AI models outlast the man who drove it? Nobody knows yet. No Nvidia announcement has changed its open-model releases, and CFO Colette Kress is interim CEO. But open-source advocates are worried, and the biggest open-source bet on the table, a pending $12.9 billion purchase of Hugging Face, is still unfinished.
Our earlier posts cover the resignation itself and the wider fallout. This post covers one angle those posts do not: what is at stake for open-weight AI, using the reporting from TechCrunch on the open-source questions and Hugging Face's reaction.
TL;DR
| Item | Status as of Nov 12, 2026 |
|---|---|
| CEO | Huang resigned; Colette Kress is interim CEO |
| Hugging Face deal | $12.9B, announced in September, scheduled to close in H1 2027, not closed |
| DOJ | Asking about competition effects, per an October Reuters report cited by TechCrunch |
| Nemotron | Open-weight family; 120B enterprise-agent model and two Bolt models added in October |
| Poolside | $6B non-exclusive license announced in August |
| Policy | July letter signed by 25 companies urging no over-restriction of open-weight models |
| What changed | No announced change to any of these |
Why people link Huang to open models
TechCrunch calls Huang a longtime advocate of open-source and open-weight AI. It lists several moves tied to him in 2026:
- A policy letter in July. It was his first post on X, per TechCrunch, and was signed by 25 companies including Microsoft, Meta, Hugging Face, IBM and Palantir. It urged the US government not to over-restrict open-weight models and said "Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty."
- Nemotron. The open-weight family built for long-running agent systems and reasoning workloads. In October Nvidia added a 120-billion-parameter model for enterprise agents and two lightweight Bolt models for PCs. We covered them in the Nemotron 3.5 Super VL post and the Bolt E4B edge model post.
- Poolside. In August Nvidia announced a $6 billion deal, described by TechCrunch as a non-exclusive license to the coding startup's technology. The Wall Street Journal, as cited by TechCrunch, framed it as helping build a US open ecosystem to compete with Chinese firms such as DeepSeek and Kimi K3.
- Backing US open labs. TechCrunch says Huang backed Reflection AI and Thinking Machines Lab to strengthen American open models. Our post on the Reflection AI acquisition talks covers the Reflection side.
- Hugging Face. The September acquisition announcement, covered in Nvidia to acquire Hugging Face for $12.9 billion.
At CES in January, Huang said open-source models have "really revolutionized artificial intelligence," per TechCrunch. The pattern is a CEO who spoke for open models in public and put money behind them.
What the open-source community is saying
Two reactions stand out in the reporting.
Clement Delangue, Hugging Face CEO. He posted on X on November 12 that it was a "really sad day," said he wished the situation had not happened, and pledged that Hugging Face will keep pushing for open-source AI "harder than ever." TechCrunch reads the post as a sign of worry among advocates and an effort to reassure them. Hugging Face and Nvidia are not one company yet. Nvidia has said Hugging Face will keep supporting open-source work, per TechCrunch, which also notes that the deal would give Nvidia control over a major distribution channel for AI models.
Nathan Lambert, researcher. TechCrunch quotes his November 11 post: "Jensen no longer being CEO would be very very bad for the open model cause in the US." He wrote that before the resignation was official, which shows how tightly the community tied the CEO to the cause.
Both are opinions from people with a stake. They tell you how the community feels, not what Nvidia will do.
The Hugging Face deal is the key test
The pending purchase is the biggest open-source decision on Nvidia's books. TechCrunch gives these facts:
- Announced in September at $12.9 billion, the second-largest Nvidia acquisition after its December 2025 purchase of Groq assets valued at $20 billion.
- Scheduled to close in the first half of 2027.
- Under questions from the Department of Justice about competition, per an October Reuters report; see our post on the DOJ's antitrust questions about distribution.
- Nvidia did not answer requests for comment on the deal's status or on any talks with Kress.
A deal this large does not close on one executive's word. It depends on regulators, boards and integration plans. A new CEO could keep it, reshape it or slow it. Today the public record shows none of those outcomes.
Why Nvidia may keep backing open models anyway
TechCrunch offers a business argument that does not depend on one person. Major Nvidia customers, Google and OpenAI, are building their own chips. That gives Nvidia a reason to grow a wide developer base, because more groups training and running models means more GPU sales. Open models feed that base. A cautious interim CEO has little reason to cut a program that sells hardware.
The counter-risk is policy. TechCrunch notes the open-weight debate followed allegations that Chinese AI labs are stealing intellectual property from Anthropic and OpenAI, which reportedly urged Washington to act. If export and security pressure grows, a more careful leadership may avoid public fights over open weights. That is a plausible worry, not a known plan. For the legal side of open releases, read our guide on open-weight model licenses.
Nvidia's open-source year in order
Putting the dated items from the reporting in sequence shows how much of the strategy was built in 2026:
- January: At CES, Huang says open-source models have "really revolutionized artificial intelligence."
- July: Huang's first X post is the open-weights policy letter signed by 25 companies.
- August: Nvidia announces the $6 billion Poolside deal.
- September: Nvidia announces the $12.9 billion Hugging Face acquisition.
- October: Nemotron grows with a 120-billion-parameter enterprise-agent model and two Bolt models. Reuters reports the DOJ is asking about the Hugging Face deal.
- November 12: Huang resigns, and the open-model community asks what happens next.
Almost every item sits within the last six months. That is why the question feels sharp. A strategy built that fast has not yet been tested by a leadership change.
Three ways this could go
These are scenarios, not predictions, and none has any announcement behind it.
- Continuity. Kress or a permanent successor keeps Nemotron releases on schedule, the Hugging Face deal proceeds, and the company argues that open models sell GPUs. The business logic TechCrunch describes supports this path.
- Caution. Leadership keeps releasing models but stops taking public policy stands on open weights, and the Hugging Face timeline stretches under regulatory review. Delangue's pledge and Lambert's warning address this case.
- Retreat. Nvidia scales back open releases or restructures the Hugging Face deal. Nothing in the reporting points to this, and it would cut against the company's incentive to grow its developer base.
Which path unfolds will show in small signals first: a release date that holds or slips, a line in a regulatory filing, or a statement from the interim CEO.
What is known and what is not
Known (per TechCrunch): Huang resigned November 12. Kress is interim CEO. The Hugging Face deal is unfinished and scheduled for H1 2027. Nemotron, Poolside and the July letter exist as described. Delangue and Lambert made the quoted statements.
Not known: whether Nvidia will change Nemotron release plans, whether the Hugging Face deal closes on the same terms, whether Kress will sign or defend the July letter, and who the permanent CEO will be. TechCrunch also says it could not find Nvidia's comment on the open-source questions. Reports on the resignation's cause belong in our Huang scandal coverage, and we do not repeat unverified details here.
What developers and teams should do
- Keep using what is released. Weights already published under their licenses do not vanish. Download and archive the versions you depend on.
- Do not plan around a roadmap. Treat any future Nemotron model as unconfirmed until Nvidia ships it.
- Watch three signals. Kress's first public statement on open models, any change to the Hugging Face timeline, and the next Nemotron release.
- Keep a second source. If you rely on Nvidia-trained open models, test an alternative such as another open-weight family so a policy change does not strand you.
- Track the DOJ review. It can shape how Hugging Face runs as a neutral hub, which matters to every team that hosts models there.
Bottom line
Open-weight AI got a lot of weight from one executive's public stance, and now that stance has no confirmed owner. The facts today are modest: no announced change, one big deal pending, and worried advocates. Treat the next 90 days as the real test. A shipped Nemotron model or a closed Hugging Face deal would be evidence. A quiet pause would be too.
Related reading
- Jensen Huang resigns: Colette Kress named interim CEO
- Nvidia to acquire Hugging Face for $12.9 billion
- DOJ questions Nvidia and Hugging Face deal
- Nvidia and Reflection AI acquisition talks
- Nemotron 3.5 Super VL open agents
- Open-weight AI model licenses explained
This post reflects reporting available on November 12, 2026. We will update it as Nvidia comments.