Nvidia’s newest big bet was just announced: today, Thursday, September 3, 2026, the chip giant confirmed it had reached a definitive agreement to acquire Hugging Face, the open-source AI software platform used by developers and companies to build, share, and deploy artificial intelligence models.

Deal value: $12.9 billion.

The Hugging Face deal is Nvidia’s second-largest acquisition ever

On the financial terms, Nvidia will pay Hugging Face’s investors roughly $11.9 billion and, according to Reuters, is also offering a stock-based retention package worth up to $1 billion for employees joining the company.

The Hugging Face acquisition ranks as the second-largest in Nvidia’s history, behind the $20 billion purchase of chipmaker Groq’s assets last December.

Before that, in 2019, Nvidia acquired Israeli semiconductor company Mellanox for nearly $7 billion.

Nvidia’s two previous major acquisitions both involved chip companies, which underscores just how significant this latest move into Hugging Face really is.

Nvidia moves beyond AI chips, betting on software and the wider AI ecosystem

Unlike Groq or Mellanox, Hugging Face doesn’t make chips at all. It’s a software company through and through, and the deal confirms Nvidia’s ambition to expand well beyond AI chip manufacturing, the business that has defined its strategy until now.

Owning Hugging Face gives Nvidia a front-row seat on the models, tools and trends emerging across the open-weight AI landscape, including AI coming out of China.

Open-weight models (AI systems whose underlying parameters are published openly) let companies run AI on their own infrastructure, keep tighter control over their data, and customize models to their needs, without relying on APIs (Application Programming Interfaces) to communicate with another software provider, such as OpenAI or others, to get what they need.

A sharper response to Chinese rivals DeepSeek and Z.ai

By acquiring Hugging Face, Nvidia takes a clear step forward in countering competition from Chinese companies like DeepSeek and Z.ai, which have managed to build models that rival the best American ones on certain tasks, at a fraction of the cost.

The rise of these Chinese firms has fueled concern that some American companies could end up dependent on Chinese-made models.

By taking over Hugging Face, Nvidia gains a clearer view of emerging models, including open-weight ones, while also keeping tabs on the evolution of Chinese AI.

Why Hugging Face will stay an open platform, with no obligation to use Nvidia chips

To play an even more central role in artificial intelligence, Nvidia CEO Jensen Huang made a significant announcement of his own, specifying that Hugging Face will remain an open platform for the entire AI ecosystem and adding that there will be no requirement to use Nvidia chips to build models on the platform or to deploy them through Hugging Face.

The logic is straightforward: if Nvidia turned Hugging Face into a platform locked to its own chips, it would risk losing the developers and companies that rely on AMD chips, Google’s TPUs, AWS, or other systems.

Another upside of the deal is that by strengthening the open-source ecosystem, Nvidia could offset any slowdown in demand for its AI chips from customers such as Meta, OpenAI and Microsoft.

More and more of these giants are developing their own AI chips to cut their reliance on Nvidia’s processors, which are expensive and in limited supply.

Those same companies, which already use Hugging Face’s services, could increasingly turn to Nvidia not just to buy chips but to understand the AI models and software of the future, thanks to the developer activity happening on the platform it has just acquired.

The pool Nvidia can now tap into is enormous: Hugging Face says more than 50,000 organizations use the platform today, and its Enterprise client list already includes Meta, Amazon, Microsoft, Google, OpenAI, and Nvidia itself.

In essence, Nvidia built its empire selling the infrastructure that powers AI: the chips. With Hugging Face, it now aims to also control the platform where the models that run on that infrastructure are developed, shared and established.

Editor’s note

This article was originally published in Italian on money.it by Laura Naka Antonelli on September 03, 2026 as «Perché Nvidia ha deciso di comprare Hugging Face per $12,9 miliardi. La grande scommessa oltre i chip AI». It has been translated and adapted for an international audience by the Money.it International desk.