Hugging Face CEO Vows to Defend Open AI After Huang Exits Nvidia
Nvidia is buying Hugging Face, the main catalog of open neural networks, for $12.9 billion, but Jensen Huang, the most prominent advocate of open models at Nvidia, has stepped down as CEO. Hugging Face head Clement Delangue called it a "very sad day" and promised to fight for open AI even harder. The deal is expected to close in the first half of 2027, so projects built on open models should keep copies of the weights they need on their own infrastructure.
AI-processed from TechCrunch; edited by Hamidun News
Imagine the main warehouse where developers get ready-made "parts" for their products is bought by a large company, and the person who promised to keep that warehouse's doors open suddenly leaves his post. That is roughly the situation surrounding Hugging Face. It is the largest catalog of open neural networks: models and their weights (the trained "settings" without which a model does not work) are uploaded there, and developers around the world download them from there. In September, Nvidia announced it was buying the platform for $12.9 billion. On Thursday, November 12, 2026, news broke that Nvidia CEO Jensen Huang was stepping down, and Hugging Face head Clement Delangue publicly promised not to abandon the fight for open AI.
What Delangue said
Delangue's post on X is emotional and short. He wrote that "this whole story is disgusting," that he wishes nothing like this had happened, and that it is "a very sad day." But then he added a caveat that makes the post worth attention: the team will fight for open AI "even harder," because the cause is too important to slow down.
The text reads as an attempt to reassure the community. Huang has long been considered one of the main defenders of open models and open weights. When a person who personally championed this line leaves, supporters of open source have a simple question: will Nvidia keep the same course without him? There is no answer yet. TechCrunch asked Hugging Face for comment on the status of the deal and on whether they had already spoken with Nvidia's interim CEO, Colette Kress, but received no reply.
What is known about the deal
The $12.9 billion purchase is Nvidia's second-largest in its history. Only its $20 billion asset deal for chipmaker Groq in December 2025 was bigger.
From the start, the deal had two layers. On one hand, Nvidia promised to maintain open support for the platform. On the other, the purchase gives it control over one of the main distribution channels for models. In October, as Reuters reported, the US Department of Justice began examining how the deal would affect competition in the AI market. The closing is still scheduled for the first half of 2027.
How Nvidia promoted open AI
Huang's position was not limited to Hugging Face:
- In July, he wrote on X for the first time and published a letter on the importance of open models. It was signed by 25 companies, including Microsoft, Meta, Hugging Face, IBM, and Palantir. The authors asked the US government not to restrict open weights too tightly. Their argument: a model that anyone can download, inspect, modify, and run on their own infrastructure makes AI more accessible, strengthens competition, and protects customers from vendor lock-in.
- In August, Nvidia announced a $6 billion deal with coding startup Poolside to build an open AI ecosystem in the US and compete with China's DeepSeek and Kimi K3, as The Wall Street Journal reported.
- Last month, Nvidia expanded its open Nemotron model family. It added a large 120-billion-parameter model for enterprise agents and two lightweight Bolt models that run on ordinary PCs.
- In January, at the CES show in Las Vegas, Huang said open models had "truly revolutionized" artificial intelligence.
What developers should do
Nothing has been closed or changed yet: the deal is not finished, Delangue promises to continue the course, and Nvidia has already invested billions in open models. But the risk has become more visible. The platform that hosts your models, datasets, and pipelines may change owners and priorities, and the future owner no longer has a leader who personally defended openness.
The practical conclusions are simple. Keep copies of critical weights and configurations yourself or in independent storage. Do not hard-wire production systems to a single hub. Watch the regulators: the Justice Department review and the closing terms of the deal will show whether Hugging Face remains a neutral platform for the whole industry.
- Nov 21, 2026Nvidia RTX Spark Mini PCs With 128 GB Memory Can Run 120B-Parameter AI Models, Starting at $4,399
- Nov 21, 2026Huang Steps Down at Nvidia, and AI Startups Lose Their Biggest Investor: 105 Rounds in a Year
- Nov 21, 2026LangChain Raises $300M at $3B Valuation, Double Last Year's
- Nov 21, 2026Dyania's AI on Nvidia Servers Reads Records of 7,800 Cleveland Clinic Patients
Key Questions Answered
What Delangue said?
Delangue's post on X is emotional and short. He wrote that "this whole story is disgusting," that he wishes nothing like this had happened, and that it is "a very sad day." But then he added a caveat that makes the post worth attention: the team will fight for open AI "even harder," because the cause is too important to slow down. The text reads as an attempt to reassure the community. Huang has long been considered one of the main defenders of open models and open weights.
What is known about the deal?
The $12.9 billion purchase is Nvidia's second-largest in its history. Only its $20 billion asset deal for chipmaker Groq in December 2025 was bigger. From the start, the deal had two layers. On one hand, Nvidia promised to maintain open support for the platform. On the other, the purchase gives it control over one of the main distribution channels for models. In October, as Reuters reported, the US Department of Justice began examining how the deal would affect competition in the AI market.
How Nvidia promoted open AI?
Huang's position was not limited to Hugging Face: In July, he wrote on X for the first time and published a letter on the importance of open models. It was signed by 25 companies, including Microsoft, Meta, Hugging Face, IBM, and Palantir. The authors asked the US government not to restrict open weights too tightly. Their argument: a model that anyone can download, inspect, modify, and run on their own infrastructure makes AI more accessible, strengthens competition, and protects customers from vendor lock-in.
Test Frontier LLMs & Autonomous Agents in One Workspace
Run Claude Opus 5.5, GPT-6, Gemini 3.1, and DeepSeek side-by-side in Nomad AI (nomadnet.ai) — unified model routing, deep research, and multimodal workflows under one subscription.
The AI world, distilled — once a week
Seven stories that actually mattered, hand-picked. No noise, no reposts, no press releases.
Done! Check your inbox for a confirmation.