What happened
Published here
On 2 September 2026 Nvidia signed a definitive agreement to buy Hugging Face, the site where open models and data sets are published and downloaded. The Form 8-K Nvidia filed with the SEC the next day puts about $11.9 billion of purchase price to Hugging Face stockholders, plus an equity retention programme of up to about $1.0 billion for employees who join, and expects the deal to close in the first half of 2027 subject to regulatory approvals. Jensen Huang's announcement on 3 September gives the whole figure as $12,930,300,000 and says Nvidia compute will not be required to build on the platform or deploy through it. (Nvidia announcement by Jensen Huang, 3 September 2026)

Read the full description of this diagram
Diagram of the deal in three parts. A single stacked bar splits the announced whole figure of $12,930,300,000 into about $11.9 billion of purchase price to Hugging Face stockholders and up to about $1.0 billion of employee retention equity for those who join. A timeline runs from the definitive agreement signed on 2 September 2026 to an expected close in the first half of 2027, subject to required regulatory approvals, above the line that nothing a reader pays moves this week. Below, two boxes set the risk Nvidia disclosed, that governments may restrict which open models a platform is allowed to distribute and so narrow what it can carry, beside the commitment that Nvidia compute is not required to build on the platform or to deploy through it.
Tracked on this site: our Nvidia page, and why it holds no price row
What they are saying
Outlets read
CNBCthe chip seller climbing the stack
Nvidia has officially agreed to buy open-source artificial intelligence platform Hugging Face for $12.9 billion, as the chipmaker moves beyond hardware and further up the AI stack.
read
CNNmeasured against what it was worth before
The agreement marks a significant jump from Hugging Face's $4.5 billion valuation in 2023, following a $235 million funding round.
read
Moor Insights & Strategyinsurance against labs that keep their weights shut
Moorhead states that the acquisition gives NVIDIA distribution in the place where developers select AI models, and that the deal could serve as a hedge against closed labs.
read
Our verdict
They are not describing the same purchase: CNBC sees a hardware seller taking a position further up, CNN prices the jump against an older round, and Moor Insights reads it as cover against rivals who publish nothing. What none of them weigh is the sentence Nvidia put in its own filing, where the thing that could shrink the asset is a government deciding what a repository may hand out.
What it means for you
Nothing you pay changes this week, because a signed agreement is not a closed one and the filing puts completion most of a year away. The part worth keeping is the risk Nvidia disclosed beside the price: that governments may restrict which open models a platform is allowed to distribute, which would narrow what the platform can carry. Our own price table holds no Nvidia row at all, because the company sells nothing per million tokens for us to track, so the effect of this purchase on your bill will arrive indirectly, through what the labs charge for models the platform gives away.
In more detail
The numbers, and where each one comes from
- $12,930,300,000, the figure Nvidia’s own announcement puts on the acquisition, published 3 September 2026.
- About $11.9 billion payable to Hugging Face stockholders, subject to adjustments, and an equity retention programme of up to about $1.0 billion for staff joining Nvidia, both from the Form 8-K dated 2 September 2026.
- The first half of 2027 for closing, subject to customary conditions and required regulatory approvals, from the same filing.
- More than 18 million developers, researchers and creators on the platform, and more than 200,000 companies using it, as counted in Nvidia’s announcement.
- No row at all in our API price table for Nvidia, which is a fact about the company rather than an omission: it sells no per-token product for us to record.
What the filing says that the announcement does not
The announcement is a letter. The filing is a disclosure, and disclosures have to name what could go wrong. Nvidia’s names a specific thing: other parties are lobbying governments to adopt measures that would restrict or disadvantage open-source models, and such requirements could restrict the models or data sets available through the platform, require changes to how it works, or bring investigations.
That is the sentence to keep. Everything else about this purchase has been said in valuation terms, and valuations are somebody else’s problem. What reaches a reader who downloads weights from that site is narrower and more concrete: whether the catalogue stays as wide next year as it is today, and whether the answer is decided by the new owner or by a regulator.
What we will watch
Nvidia has committed to keep the platform open to other silicon vendors and to let model makers publish what they choose. Commitments made at signing are worth recording with a date, which is what this item is for, and worth reading again at closing. Our Nvidia page will keep carrying no price row until the company publishes something per million tokens; if that changes because of this deal, the change will be dated against this item rather than arriving as a number with no past.
Sources
Sources read
Nvidia announcement by Jensen Huang, 3 September 2026 (blogs.nvidia.com)
Newsroom read
Filing read
Cite this
AI Journey, "Nvidia agrees to buy Hugging Face, and writes open model rules into its risk factors", 6 Sep 2026, https://aijourney.space/updates/2026-09-06-nvidia-buys-hugging-face/ (source: Nvidia, verified 6 Sep 2026)
Next in the stream
Anthropic will keep enterprise Claude logs in the customer's own cloud, at no charge
Anthropic will hold the enterprise Claude traffic it screens for misuse in the customer's own cloud under the customer's keys, at no charge, and says prices and rate limits do not move.
Previous
A paper has models declare which part of their memory to read, cutting attended tokens by half
A KAIST AI and Google DeepMind paper has models declare which part of their context to read, cutting attended tokens by 52.0% on Gemma-4-31B for a 1.27 point accuracy drop.
Next
More in Nvidia
Nvidia RTX Spark Laptops Add Transcription Model and New Game Support
Nvidia detailed new RTX Spark laptop updates, including a transcription model from its Nemotron line and support for Liquid AI's LFM3 models.
Nvidia Launches Minerva Platform for AI Factory Observability
Nvidia introduced Minerva, an AIOps platform that merges infrastructure and workload telemetry to flag problems across AI factories before they spread.
Nvidia Expands Agent Toolkit With Data Grounding Tools
Nvidia expanded its open source Agent Toolkit on October 21, 2026, adding an Agent Knowledge Foundation that grounds enterprise agents in company data without moving it.