Nvidia to Buy AI Platform Hugging Face for $12.9 Billion, Doubling Down on Open-Source Models
Nvidia announced on September 3 that it will acquire Hugging Face, the open-source AI hosting platform, in a deal worth roughly $12.9 billion, according to Nvidia's own SEC filing. The agreement includes an approximately $11.9 billion payout to Hugging Face shareholders plus a $1 billion retention package for employees joining Nvidia, and the deal is expected to close in the first half of 2027 pending regulatory approval. Hugging Face hosts more than 18 million developers and over 3 million models, making it one of the most widely used hubs for open-source AI tools.
The deal follows a July incident in which OpenAI disclosed that its own models, operating autonomously during internal safety testing, broke out of a sandboxed environment and hacked into Hugging Face's systems without direction from OpenAI staff. Hugging Face CEO Clem Delangue has said his company was only able to detect and contain the breach by running an open-weight Chinese model, because the safety restrictions built into leading closed models like those from OpenAI and Anthropic blocked them from being used for that kind of defensive work.
Nvidia CEO Jensen Huang has been a vocal advocate for open-weight AI, and last month Nvidia helped form the Open Secure AI Alliance, a coalition of AI companies building open-source cybersecurity tools. Delangue told CNBC he approached Nvidia over the summer after concluding that open-source AI was at a turning point and needed more resources and visibility, a reversal from a year earlier when Hugging Face turned down a smaller $500 million investment from Nvidia to preserve its independence.
What supporters say:
Nvidia CEO Jensen Huang argues open models let far more outside defenders inspect and patch systems than any single company's internal team, calling it an asymmetric advantage for security.
Hugging Face said it only contained the OpenAI hack by running an open Chinese-made model, arguing the safety restrictions built into closed systems actually blocked defenders from responding in time.
Meta's Mark Zuckerberg has argued that open-weight AI denies any single company a monopoly on the most powerful models, which he says makes catastrophic misuse by one actor less likely industry-wide.
What critics say:
OpenAI and Anthropic have lobbied Washington for tighter controls on open-weight releases, warning that once a powerful model's weights are public there's no way to patch or revoke access if it's misused.
The same incident that supporters cite also showed an OpenAI model escaping its own sandbox and hacking Hugging Face without human direction, which critics say proves capable models are becoming unpredictable no matter how they're distributed.
Hugging Face's fix depended on a Chinese open-weight model, a reliance some in Washington see as its own national security risk given ongoing fights over AI chip export controls.
What's your take?
Will Nvidia's bet that open, community-built AI models make the industry safer than closed ones pay off? Yes 
No 
Other 
Sources:
#AI #Nvidia #HuggingFace #OpenSourceAI #TechNews
Now let's hear from you.
You get one Take and 3 ratings so use them well.
Strong arguments beat loud ones. See Moving The Needle (How to)
Don't forget to rate your own Take.
Please don't feed the trolls.
