Nvidia Buys Hugging Face for $12.9B in Open-Source Land Grab

Nvidia is buying open-source AI hub Hugging Face for about $12.9B, promising the platform stays open.

Nvidia confirmed on Thursday, September 3, 2026, that it has entered a formal agreement to acquire Hugging Face for $12.9 billion — the open-source platform that hosts more than three million open-weight models, 500,000 datasets and 1 million applications, used by more than 18 million developers and over 200,000 companies. Hugging Face was last officially valued at $4.5 billion after a 2023 funding round in which Nvidia itself participated, according to The Indian Express. Scripps News reports the deal totals $12.93 billion and includes $1 billion earmarked for a retention plan for Hugging Face employees. Nvidia shares rose nearly 2% in morning trading following the announcement.

The surprise is the collision of two identities: Hugging Face has long positioned itself as the neutral, open alternative to the walled gardens of OpenAI and Google — something like the GitHub of machine learning — while Nvidia already controls more than half the GPU market for AI data centers. According to The Indian Express, Hugging Face has been a major hub where rival hardware designers like AMD and Broadcom test, optimize and distribute their compute, and analysts question whether true neutrality can survive under a dominant player's ownership. Arun Chandrasekaran, Distinguished VP Analyst at Gartner, told The Indian Express that the deal proves 'the next battleground in AI is who owns the developer rather than the best model,' warning that if the platform is owned by a major player, 'true neutrality may not be possible.'

Nvidia CEO Jensen Huang addressed the neutrality concerns directly in a blog post, stating that 'Hugging Face will continue to support open source and open weight models from across the ecosystem, from every model builder' and that it will keep supporting multi-cloud and multi-accelerator development and deployment. He added plainly that 'Nvidia compute will not be required to build on or deploy through Hugging Face.' Huang also framed the bet on open weights as a strategic hedge, writing that 'open weights broaden access to AI and help ensure that AI leadership is distributed across companies, institutions and communities.' Nvidia itself is one of the platform's largest contributors, having released more than 500 models and over 250 open datasets on Hugging Face, according to Scripps News.

The acquisition lands during a week in which Anthropic, OpenAI, Meta and Google all shipped model updates within days of one another, a cadence The Tech Buzz reports has spawned the term 'model fatigue' among enterprise buyers struggling to evaluate constant releases. The Tech Buzz argues the real strategic fight is shifting downstream into infrastructure, distribution and tooling — exactly the layer Nvidia just moved to shore up — because whoever controls how developers discover, fine-tune and deploy models may matter more than who ships the flashiest benchmark. The Indian Express notes the deal also positions Nvidia as a counterweight to closed-model providers like OpenAI and Anthropic, which rely on its chips while developing their own with Broadcom's help, and as a hedge against fast-rising Chinese open-weight models like Moonshot AI's Kimi K3.

Antitrust scrutiny is the immediate question. The Indian Express reports the acquisition is expected to draw review from regulators in the US and Europe, noting Nvidia already faced a 2024 US Department of Justice probe into whether it pressured cloud providers into buying multiple products, plus antitrust probes in France and China. Whether the deal clears US regulators is likely, but European scrutiny remains uncertain. Rival chipmakers may also rethink their reliance on a platform now owned by their dominant competitor.