# Nvidia Closes $12.93 Billion Purchase of Hugging Face, Buying the Distribution Layer for Open Models

About $11.9 billion goes to investors with up to $1 billion in equity retention for staff, making it Nvidia's second-largest deal after last year's $20 billion purchase of Groq assets.

- Published: 2026-09-06T06:32:46.692Z
- Canonical: https://polylog.news/ai/2026-09-06/nvidia-closes-12-93-billion-purchase-of-hugging-face-buying
- Publisher: Polylog (AI desk)
- Section: tech
- Sources: [Polylog editors](https://polylog.news)

Nvidia confirmed it will acquire Hugging Face for $12.93 billion, a price the sellers treated as a puzzle. Hugging Face co-founder Thomas Wolf publicly invited people to find the references to Hugging Face and Nvidia hidden inside the figure, and [the Russian-language channel AI ML Big Data documented](https://t.me/ai_machinelearning_big_data/10845) the resulting search across social media.

The structure matters more than the number. [TechCrunch reported](https://techcrunch.com/2026/09/03/nvidia-confirms-it-will-buy-hugging-face-for-12-9-billion/) that roughly $11.9 billion goes to Hugging Face investors, with up to $1 billion set aside as equity retention for employees who move to Nvidia. [Tom's Hardware noted](https://www.tomshardware.com/tech-industry/artificial-intelligence/nvidia-acquires-hugging-face-for-usd12-93-billion-company-gains-control-of-major-ai-model-distribution-platform) that the platform hosts more than 18 million developers and more than 3 million models, 500,000 datasets and 1 million applications. That is the default place engineers go to find, evaluate and pull open weights.

Nvidia says the platform stays open and that its own chips will not be required to build or deploy through Hugging Face. [Hugging Face's chief executive told CNBC](https://www.cnbc.com/2026/09/03/nvidia-agrees-to-buy-hugging-face-for-almost-13-billion-ai-expansion.html) the company approached Nvidia's chief executive, Jensen Huang, weeks before the deal was agreed, which reframes the transaction from a hostile takeover into a sale the seller initiated.

What Nvidia gains is visibility. A chip vendor that owns the registry sees which architectures are downloaded, which quantizations are used in production, and which inference stacks developers actually run. That telemetry is worth something independent of hardware lock-in, and it is exactly what competing accelerator vendors and cloud providers will now have to ask a rival for.

## What this means

Nvidia has moved from supplying the compute under open models to owning the platform through which those models reach developers. AMD, Intel, Google and Amazon lose access to a neutral venue where engineers choose their runtimes, and any signal Nvidia extracts about which kernels and quantizations dominate feeds directly back into chip and software roadmaps. The opposing case is real: if Nvidia biases the platform toward its own hardware, the open-weight community can fork the registry elsewhere, and mirrors in China and Europe already exist. Which of those two outcomes occurs will be visible in whether major labs keep publishing weights to Hugging Face first.

## What to watch

- Whether competing labs and accelerator vendors keep releasing on Hugging Face first or start mirroring elsewhere, which is the clearest test of whether the community believes the neutrality promise.
- Any antitrust review in the United States or European Union of a chip supplier owning the main open-model distribution point, since a challenge would slow similar vertical deals across the sector.
- Whether Nvidia begins offering hosted inference priced against the clouds, which would turn a distribution asset into direct competition with its own largest customers.