TL;DR
Jensen Huang announced on 3 September that Nvidia has agreed to acquire Hugging Face for $12,930,300,000. However, agreed is not closed, and the announcement says nothing about regulatory review, conditions or timing.
In plain English
Agreed versus closed — an agreed acquisition is a deal both sides have signed up to. It still has to clear regulators and any conditions before it actually completes. Deals of this size sometimes do not.
Open weights — a model whose trained parameters you can download, so you can run it on your own machines rather than only through someone else’s API.
Inference provider — a company that runs a model for you and charges per use. The Hub lists several, and whether that list stays plural is one of the things to watch.
On 28 August I wrote that CNBC’s report of an Nvidia–Hugging Face acquisition was not a confirmed transaction, and that treating it as a completed deal would be premature. On 3 September Jensen Huang announced it under his own byline: “NVIDIA has agreed to acquire Hugging Face for $12,930,300,000.”
If your work depends on Hugging Face — and if you use open-weight models it probably does, whether you think about it or not — the things to watch are not the price or the personalities. Watch whether AMD, Intel and Apple silicon runtimes stay first-class on the Hub. Watch whether inference-provider listings stay genuinely plural rather than acquiring a default. Watch whether the free tier for model hosting changes shape. Watch whether the licence and provenance metadata that makes the Hub usable stays as good for models Nvidia has no stake in. None of those require insider knowledge; all of them are visible from the outside, and all of them test Huang’s commitment that “NVIDIA compute will not be required to build on or deploy through Hugging Face”.
For scale, the numbers in the announcement: more than 18 million developers, researchers and creators; more than 3 million models, 500,000 datasets and 1 million applications; more than 200,000 companies using the platform. Nvidia also states it is already the largest contributor of open models and data to Hugging Face, with more than 500 models and 250 open datasets published there. That last point is the strongest argument that the acquisition is a continuation rather than a turn — and it is Nvidia’s own claim about Nvidia, so weigh it accordingly.