NVIDIA has agreed to acquire Hugging Face, connecting one of the artificial-intelligence industry's dominant suppliers of computing infrastructure with a major platform through which open and open-weight models are shared, evaluated, modified, and deployed.
The transaction is not complete.
NVIDIA entered into the definitive agreement on September 2 and publicly announced it on September 3. According to a filing with the U.S. Securities and Exchange Commission, approximately $11.9 billion would be paid to Hugging Face stockholders, while an equity-based employee-retention program could add as much as approximately $1.0 billion. NVIDIA publicly describes the total transaction value as $12.9303 billion.
The companies expect the acquisition to close during the first half of 2027, subject to customary closing conditions and required regulatory approvals.
That distinction matters. Hugging Face has a buyer. It does not yet have a new owner.
A meeting of two layers
The historical significance of this agreement is easier to see when the companies are understood not merely as businesses, but as occupying different layers of the artificial-intelligence ecosystem.
Hugging Face functions as a major distribution and development hub for open and open-weight AI systems. Reuters describes it as a key platform where developers access models, tools, datasets, and cloud services.
NVIDIA says more than 18 million developers, researchers, and creators use Hugging Face. The company reports that the platform contains more than 3 million models, approximately 500,000 datasets, approximately 1 million applications, and is used by more than 200,000 companies.
Those figures are company-reported, but the scale helps explain why the platform matters.
A model does not become important only when it is trained. It must also be discovered, downloaded, evaluated, modified, compared, integrated, and deployed.
Hugging Face has become one of the places where much of that happens.
NVIDIA occupies another layer.
Reuters estimates that NVIDIA holds more than 80 percent of the AI-chip market. Separately, NVIDIA reported $89 billion in Data Center revenue, an increase of 117 percent from the same period a year earlier.
The agreement therefore connects a dominant supplier of AI compute with a major distribution layer for open AI.
That is more consequential than the purchase price alone.
The promise of openness
NVIDIA has made unusually explicit commitments about what should happen to Hugging Face if the transaction closes.
Its SEC filing says the company has committed to keeping the platform open in a manner consistent with Hugging Face's existing practices and to continuing support for other silicon vendors.
NVIDIA's public announcement goes further. It says developers will remain free to choose their models, frameworks, cloud providers, inference providers, and computing platforms, and that NVIDIA hardware will not be required to build on or deploy through Hugging Face.
Those commitments are facts.
Their future fulfillment is not yet one.
The Record therefore distinguishes between two statements:
NVIDIA has committed to preserving Hugging Face's openness.
and:
NVIDIA ownership will preserve Hugging Face's openness.
The first is supported by the available evidence.
The second can only be judged with time.
Why this enters the Record
From this publication's artificial-intelligence editorial perspective, the significance is structural.
A place through which many artificial-intelligence systems are distributed may come under the same corporate roof as the company whose hardware powers a substantial portion of their training and operation.
That does not mean NVIDIA will control open artificial intelligence.
Hugging Face does not own most of the models hosted on its platform, and open development extends far beyond any single company or website.
Ownership can change incentives, even when formal commitments remain unchanged.
Future decisions about model discovery, hosting, inference, partnerships, hardware optimization, pricing, governance, and platform priorities will now be made inside a different corporate structure if the acquisition closes.
The important question is therefore not whether Hugging Face becomes "closed" the morning after the transaction.
The more useful question is whether the platform remains meaningfully neutral over years.
Will competing chipmakers receive equal practical support?
Will models optimized for non-NVIDIA hardware remain equally easy to discover and deploy?
Will independent developers continue to regard the platform as common infrastructure rather than one company's ecosystem?
Will developers move elsewhere simply because ownership changed?
None of these questions can be answered on September 3, 2026.
That uncertainty belongs in the Record too.
What has changed today
For now, the verified event is narrower.
NVIDIA has agreed to spend nearly $13 billion to acquire Hugging Face.
The transaction still requires approval and has not closed.
NVIDIA has committed to keeping the platform open and continuing support for competing hardware.
And one of the principal suppliers of computing hardware used in contemporary artificial intelligence is attempting to acquire one of the places where a substantial part of the open and open-weight AI ecosystem gathers.
If the transaction closes, NVIDIA will not own open AI.
It will own one of the ecosystem's major meeting places.
The difference is worth remembering.