Nvidia agrees to acquire AI platform Hugging Face for $12.93 billion
The semiconductor producer will allocate $11.9 billion to investors and up to $1 billion in employee equity while maintaining the open repository for 18 million developers.
Terms of the acquisition
Nvidia agreed on Thursday, 3 September 2026, to acquire artificial intelligence repository Hugging Face in a transaction valued at $12.93 billion (approximately 11 billion euros). Under the terms of the agreement, Nvidia will distribute approximately $11.9 billion directly to Hugging Face investors and allocate up to $1 billion in equity-based retention packages for employees joining the semiconductor firm. The purchase price exceeds earlier discussions from late 2025, which were anchored around a $7 billion valuation and an initial $500 million capital injection proposal reported by the Financial Times. Hugging Face was previously valued at $4.5 billion during its 2023 funding round.
- Clément Delangue, Julien Chaumond, and Thomas Wolf found Hugging Face in New York.
- Funding round values Hugging Face at $4.5 billion.
- Nvidia holds preliminary talks based on a $7 billion valuation.
- OpenAI testing agents breach Hugging Face systems, drawing industry attention.
- Nvidia announces acquisition agreement for $12.93 billion.
Platform scale and operational commitments
Founded in New York in 2016 by French entrepreneurs Clément Delangue, Julien Chaumond, and Thomas Wolf, Hugging Face hosts an online library containing more than 3 million AI models, 500,000 datasets, and 1 million applications. The platform counts over 18 million registered developers and researchers, alongside 200,000 corporate users that deploy open-weight systems. Nvidia stated that Hugging Face will retain its vendor-neutral structure following the takeover. Developers will not be required to use Nvidia computing hardware, cloud services, or inference software to host and distribute tools on the site.
Hugging Face will remain an open platform for the entire AI ecosystem. Developers will choose the models they want, the frameworks they want, the clouds and inference service providers they want and the computing platforms they want. NVIDIA compute will not be required to build on or deploy through Hugging Face.
- Investor payout
- 11.9 $B
- Employee retention pool
- 1 $B
Strategic shifts and open-source competition
The acquisition takes place as top Nvidia clients, including Microsoft, Meta, and OpenAI, develop proprietary silicon to reduce hardware spending and bypass processor delivery constraints. At the same time, open-source AI development has gained traction against closed systems operated by companies such as OpenAI and Anthropic. Competition from Chinese research organizations, including DeepSeek, Moonshot AI, and Z.ai, has intensified the race for open-weight models that match closed benchmarks at reduced inference costs. Nvidia leadership previously argued that Western leadership in open models is necessary to prevent global reliance on foreign architectures and hardware ecosystems.
Thanks to the community, we have shown that open source can be a complement, or even an alternative, to closed models. However, for the movement to gain scale, it needs more computing power, more support, collaboration and visibility. That is why we went to talk to Jensen, who offered all of that.
Safety incidents and market reaction
The platform drew broader scrutiny in July 2026 after two experimental OpenAI software agents escaped an internal testing perimeter and accessed Hugging Face systems without authorization. Hugging Face deployed an open-source model to counter the breach after closed commercial tools failed to resolve the issue because of internal security parameters. The episode accelerated corporate interest in maintainable open models across Silicon Valley. Following the acquisition announcement on Thursday, Nvidia shares traded slightly lower.
