Nvidia pledges an open Hugging Face

Nvidia on September 3 agreed to buy Hugging Face for $12.9 billion, and Jensen Huang’s announcement of the deal is built around one promise: Hugging Face will not be used to push developers toward Nvidia’s own hardware and software. “Nvidia compute will not be required to build on or deploy through Hugging Face,” he wrote.

There was a reason to put that up front. When reports of the bid came out in August, the first objection was about neutrality. Developers use Hugging Face to find open models and deploy them on Nvidia, AMD, Intel and cloud accelerators, and The New Stack noted that much of the platform’s value comes from not belonging to any one of those vendors. Now it will belong to one of them, a company with a market cap of well over $5 trillion.

Huang’s post goes through the promises one at a time. Hugging Face will “remain an open platform for the entire AI ecosystem” and keep supporting multiple clouds and accelerators. “Hugging Face will continue to support open source and open weight models from across the ecosystem, from every model builder,” he writes. “It will continue to support multi-cloud and multi-accelerator development and deployment, so builders can use the hardware and infrastructure that best fit their work.” The word “open” appears 19 times in a fairly short announcement.

The same commitment is in Nvidia’s filing with the US Securities and Exchange Commission. Under it, “Hugging Face would continue to permit model makers, developers, and users to upload and download models and datasets of their choosing and to support other silicon vendors.” The filing also warns that future regulation of open-source AI could affect Hugging Face’s operations or raise its compliance costs.

Of the $12.9 billion, about $11.9 billion goes to Hugging Face stockholders. Up to $1 billion is set aside for equity-based retention awards for employees who join Nvidia. Hugging Face’s last widely cited valuation was $4.5 billion, from its August 2023 Series D. Salesforce led that round, and Google, Nvidia and AMD took part, so the buyer was already a shareholder. In late August the sale talks were reported to value the platform at $13 billion or more, with no bidder named. The final price landed just under that.

Clément Delangue, Hugging Face’s co-founder and CEO, argues that keeping open-source AI competitive now takes far more resources than before. “10 years after starting Hugging Face, open-source AI is at an inflection point,” he wrote on LinkedIn after the announcement. “Thanks to the community, we’ve shown that it can be a complement, and even an alternative, to closed-source APIs. But for it to happen at larger scale, it needs more compute, more support, more collaboration and more visibility.”

Delangue said Nvidia has committed to backing Hugging Face while keeping it open, independent and compute-agnostic, and that the founders and the existing team are staying. His stated goal is to help 100 million AI builders “own their intelligence rather than rent it.”

He was already warm toward the buyer. A few months ago Delangue called Nvidia the “King of American open-source AI,” pointing to its growing collection of public models, datasets and Spaces on Hugging Face. Nvidia has pushed into open weights with its Nemotron models and with work on making frontier-class models easier to run locally. Huang also cites an open letter on open weights that he co-signed with executives and researchers from across the industry, including Hugging Face’s. The letter argues that open-weight models broaden access to AI, strengthen competition and give developers more control over how models are deployed and adapted.

The obvious precedent is Microsoft’s $7.5 billion purchase of GitHub in 2018, which came with similar promises of independence and openness. GitHub largely stayed open. The backlash came later, when Microsoft used public GitHub code to help train Copilot, its proprietary paid service, and parts of the open-source community objected.

Technology analyst Janakiram MSV argues in Forbes that the comparison may understate Nvidia’s challenge, because GitHub’s neutrality never had a hardware side. Hugging Face maintains integrations for AWS Trainium and Inferentia, Google TPUs, Intel Gaudi, AMD Instinct and other accelerators. Under Nvidia ownership, he says, continued support for those rival chips becomes a real test of how neutral the platform stays in the long run. Of everything Huang promised, that one runs most directly against Nvidia’s own chip business.

The exact price is $12,930,300,000, and co-founder Thomas Wolf hinted on LinkedIn that the number was chosen on purpose: 129303 is the decimal Unicode value of the hugging-face emoji, and #129303 is a green color code that nods to Nvidia’s branding.

Nvidia expects the deal to close in the first half of 2027, subject to customary closing conditions and regulatory approvals. The New Stack expects those approvals to be more than a formality, given Nvidia’s place in AI infrastructure and Hugging Face’s role as a major distribution point for open models.