Hugging Face CEO Delangue renews warning over AI power concentration as Anthropic IPO nears
Key Takeaways
- •Hugging Face CEO Clement Delangue warned on X that the concentration of power, capabilities and wealth is the biggest risk in AI, arguing that stronger open-source efforts are needed to prevent a few wealthy firms from controlling the field.
- •Yann LeCun, Meta's former chief AI scientist, endorsed Delangue's warning on Monday, saying he has made the same argument for years.
- •Anthropic declined to sign an industry letter urging lawmakers not to rush restrictions on open-weight models, and CEO Dario Amodei later rejected accusations of seeking a ban, calling open models without dangerous capabilities a public good.
- •Nvidia agreed on September 3 to buy Hugging Face for $12.93 billion, with CEO Jensen Huang saying the platform will remain open to the whole AI industry, though the deal still requires regulatory review.
- •Anthropic's updated IPO filing on November 10 plans to offer 825 million shares at $70 to $78 each, potentially raising up to $64.4 billion at a valuation of roughly $916 billion to $1.02 trillion.

Hugging Face CEO Clement Delangue has renewed his warning that power and wealth in artificial intelligence are becoming concentrated in a handful of companies, and Meta's former chief AI scientist Yann LeCun agrees. The remarks came on X on Sunday, just as Anthropic prepares to go public, adding to an open fight over whether AI should be open or closed.
What Delangue is warning about
Posting on X on Sunday, Delangue wrote that the “concentration of power, capabilities and wealth is the biggest risk in AI.” He argued that without much stronger open-source efforts, the industry could become one in which a few wealthy firms control AI and everyone else is left behind.
It was not the first such warning. On June 26, Delangue repeated the point about the concentration of “power, capabilities and economic wealth,” calling for “more rebels” against the large companies and governments that already control much of the field. On Monday, LeCun replied to the post, writing: “I've said this for years too.”
The debate turns on a technical distinction: open-weight models are files that anyone can download, run and modify, while closed models can only be used through a company's own app or service. Hugging Face hosts millions of open models; Anthropic sells closed ones. For the businesses and developers building on these tools, the split shapes cost, control and where their data lives — and it has turned the open-versus-closed question into a recurring fight over AI policy.
Why Anthropic sits at the center of the debate
Over the summer, Hugging Face signed an industry letter urging lawmakers not to rush new restrictions on open-weight models. Nvidia (NASDAQ: NVDA), Microsoft (NASDAQ: MSFT), Meta (NASDAQ: META), IBM (NYSE: IBM) and Mistral also backed the letter, but Anthropic did not sign. The company was then accused of wanting open models restricted to protect its business.
Anthropic CEO Dario Amodei rejected that characterization on July 27, writing that the company has “never advocated for a ban on open-weights models” and that open models without dangerous capabilities are “a public good.” Rather than a ban, he favors keeping advanced chips out of China, cracking down on large-scale copying of AI models, and safety tests for any very capable model, whether open or closed. He also disagreed with the industry letter's claim that openness helps defenders more than attackers.
Delangue's company has since changed hands. On September 3, Nvidia agreed to buy Hugging Face for $12.93 billion, with Nvidia CEO Jensen Huang saying the platform will remain open to the whole AI industry. If completed, the acquisition would put Hugging Face's millions of hosted open models under Nvidia, which was also among the letter's signatories. The deal still requires regulatory review before it closes.
Anthropic's road to a public listing
Anthropic, meanwhile, is preparing to sell shares to the public. On November 10, it updated its filing to say it plans to offer 825 million shares at $70 to $78 each. That could raise as much as $64.4 billion and push the company's valuation to roughly $916 billion to $1.02 trillion. Some reports suggest the figure could reach as high as $1.4 trillion when future shares are included. The final price has not been decided. In a standard IPO, the range published in a filing is preliminary, with final terms typically set shortly before shares begin trading.
Reports state that about 4,250 of Anthropic's roughly 5,300 employees could each end up with more than $1 million following the listing, although most staff cannot sell right away because of share lockups of up to 365 days — a routine IPO mechanism that bars insiders from selling shares for a set period after a listing. Goldman Sachs (NYSE: GS), Bank of America (NYSE: BAC), JPMorgan Chase (NYSE: JPM), Wells Fargo (NYSE: WFC) and Bank of New York Mellon (NYSE: BK) are reportedly competing to manage that money.