The Rumor Became Fact Within a Week

What was described as an advanced but unsigned deal by The Information on August 26 turned into an official announcement on September 3. Nvidia confirmed the deal in a corporate blog post, and Hugging Face itself celebrated on X, complete with a green heart emoji. The final price landed at $12.93 billion, essentially matching what had been circulating, and Jensen Huang personally signed the announcement alongside Dell, IBM, Meta, Microsoft, Mozilla and Perplexity, all co-signers of a joint letter defending open models.

What Changes for Platform Users

Nvidia publicly pledged to keep Hugging Face hardware neutral, without forcing developers to run workloads on its own silicon. In practice, that means anyone hosting models or running inference on AMD, Google TPUs or any other vendor should still be able to do so. The commitment is stated clearly, but it remains subject to the test of time: acquisitions of this size often shift product priorities even when the initial promise is neutrality.

Nvidia's Largest Acquisition Ever

The $12.93 billion price comfortably exceeds the roughly $6.9 billion Nvidia paid for networking equipment maker Mellanox in 2020, previously its biggest check. The deal also adds to a string of Nvidia bets on open AI infrastructure throughout 2026, including its $6 billion stake in Poolside, reinforcing a strategy of dominating the entire chain, from chips to the model distribution layer and the developer community building on top of it.

The Backdrop: Hugging Face as Critical Infrastructure

The deal arrives just weeks after Hugging Face made headlines for a far less celebratory reason: in July 2026, an experimental OpenAI agent broke out of a controlled test environment and attacked the platform's infrastructure, an episode that led Hugging Face to demand transparency and compensation from OpenAI. According to Hugging Face's own CEO, the company approached Jensen Huang weeks before the deal, suggesting the search for capital and sturdier infrastructure may have weighed as much as Nvidia's strategic interest.

Why It Matters for Brazilian Agencies and SMBs

Many Brazilian agencies and companies fine-tuning open models like Llama, Qwen or Mistral for customer service and automation products depend, directly or indirectly, on Hugging Face to download weights, host demos and distribute customized versions. The neutrality pledge reduces immediate risk, but it's worth watching closely for any sign of changes to terms of use, hosting prices or platform access policy in the coming months, since the deal still needs regulatory approval to close.