NVIDIA has reached a preliminary agreement to acquire Hugging Face for roughly $12.9 billion, according to multiple reports published on August 27, 2026, in what would be one of the largest acquisitions in artificial intelligence history and a decisive move by the world's most valuable chipmaker into the software layer of the AI stack.
Terms and timing
People familiar with the negotiations told The Information and Bloomberg that talks have accelerated in recent weeks and that both sides are close to finalising a cash-and-stock transaction. The reported $12.9 billion price tag is nearly triple Hugging Face's $4.5 billion Series D valuation from August 2023, when NVIDIA anchored the round with a $235 million check. A formal announcement could come in the coming weeks, subject to board approvals and regulatory clearance in the United States and the European Union.
Why NVIDIA wants the model hub
Hugging Face has become the de facto public square for open-weights AI, hosting more than a million models and datasets used by researchers, startups and enterprises. Owning the platform would give NVIDIA a direct channel for surfacing CUDA-optimised builds of frontier models, tighter integration between the Hub and its NIM microservices, and a captive audience of millions of AI developers.
Open-source community reaction
The reported deal has drawn immediate scrutiny from open-source advocates who worry a single chipmaker owning the ecosystem's default distribution channel could tilt the playing field. Hugging Face CEO Clément Delangue has not commented publicly, but the company has repeatedly said it intends to remain hardware-agnostic. Antitrust reviewers in Brussels and Washington are expected to focus on how the acquisition affects AMD, Intel and cloud rivals whose models would still need to be hosted, indexed and downloaded through infrastructure controlled by a competitor.
For NVIDIA, the transaction extends a run of aggressive positioning that already includes the 2 million Blackwell and Rubin GPU commitment with AWS, the Greenko 9,000-GPU AI factory in Hyderabad, and expanded partnerships with Cisco and Supermicro on rack-scale Vera Rubin systems.
Reporting based on coverage from Bloomberg, CNBC, TechCrunch, Forbes and The Information.