Nvidia Reportedly Making $13,000,000,000 Purchase Of Open-Source AI’s Main Hub Hugging Face

Nvidia is seemingly moving to buy the main public hub for open-source artificial intelligence.

The California-based chipmaker has agreed to purchase New York City-based AI startup Hugging Face for roughly $12.9 billion, which is approximately 80 times the 10-year-old firm’s $150 million yearly revenue, according to The Information. Hugging Face partnered with Nvidia in 2023 to quickly train AI models using the chipmaker’s DGX Cloud, an Aug. 2023 Nvidia press release said.

Microsoft separately met with Hugging Face, but the companies are no longer discussing a deal, Business Insider reported.

The AI startup previously rejected a $500 million investment from Nvidia in late 2025 and said they did not wish to have one large investor steer the company, according to Financial Times.

Open weight AI models “contribute to democratising AI, to fighting concentration of power which is, in my opinion, the biggest risk in AI,” Hugging Face Co-founder and CEO Clement Delangue said, Financial Times reported on Jan. 28.

Nvidia’s reported purchase would extend its reach where developers choose which AI models to use, Sanchit Vir Gogia, chief analyst, founder and CEO of Greyhound Research told the Daily Caller News Foundation. Nvidia already makes the chips and software that powers much of the AI industry, Gogia told the DCNF. He explained that buying Hugging Face would hand the chipmaker a hub where those models are found and adopted, which is what the $12.9 billion price reflects.

Owning Hugging Face would not rewrite the licenses on models already published or downloaded, Gogia told the DCNF. However, he said, the acquisition could shift which models get seen through search, rankings and default settings. Nvidia makes money whenever any model runs, and it benefits from open models staying popular. Gogia’s concern is that models built for Nvidia’s own chips subtly become the fastest and cheapest to run so rival chips lose ground in the market without ever being banned, he told the DNCF.

Hugging Face told the DCNF it did not have information beyond what has been publicly reported.

Nvidia did not immediately respond to the DCNF’s request for comment.

OpenAI models, including GPT-5.6 Sol, hacked Hugging Face when the models escaped a cybersecurity test, OpenAI admitted in July.

“We got attacked by secret unreleased proprietary models and defended ourselves with an open model, more precisely the nvidia quantized version of GLM 5.2 coming from Zai,” Hugging Face Co-founder and CEO Clement Delangue said in a July 31 X post, advocating for open-weight models in the US after his company was hacked by OpenAI’s models.

The AI startup is a “platform where the machine learning community collaborates on models, datasets, and applications,” Hugging Face stated on its website. The firm’s hub brings together millions of publicly available AI products, including more than two million models, as well as 1.5 million data sets and 1.5 million AI applications, according to Hugging Face’s ‘Hub’ page.

Chinese AI companies such as DeepSeek and Alibaba are available on Hugging Face’s Hub, where users around the world can download and use their products.