NVIDIA Confirms Buying Hugging Face for $12.93bn With Open

NVIDIA confirmed on Thursday, September 3, 2026, that it is buying Hugging Face for $12.93 billion, unconfirmed, ending a week of speculation about the chipmaker’s largest-ever software acquisition.

According to the outlet, the deal hands NVIDIA control of the platform where much of the open-source AI community stores and shares its models, while leaving the platform itself structurally untouched.

Hugging Face: The Before: A Decade of Independent Growth

For ten years, the platform operated as a neutral home base for the open AI ecosystem. It hosted three million models, one million applications, and half a million datasets, attracting 18 million developers and 200,000 companies that built on top of its tooling.

Crucially, it ran across competing accelerators and cloud providers, including hardware that directly rivals NVIDIA’s own GPUs. That neutrality was the entire point. Developers picked their own frameworks, chose their own clouds, and routed inference through whichever provider made sense for the workload.

Since founding, the company had raised $395 million, with the most recent round—a $235 million raise in 2023 led by Salesforce Ventures—valuing it at $4.5 billion.

The Catalyst: A $12.93bn Confirmation

Jensen Huang moved quickly to confirm the deal and address the obvious concern: would NVIDIA lock the platform to its own silicon? The NVIDIA CEO posted on LinkedIn that “Hugging Face will remain an open platform for the entire AI ecosystem,” adding that developers “will choose the models they want, the frameworks they want, the clouds and inference service providers they want and the computing platforms they want.” That last clause matters most, because it explicitly preserves the multi-accelerator model that competitors like AMD and Intel rely on. The Anthropic NVIDIA Lambda $35B cloud partnership and the AWS and NVIDIA Lock on GPU supply.

The After: Same Platform, New Owner

In practical terms, little changes for users on day one. The roadmap, the API, the model registry, and the inference endpoints all remain live. NVIDIA gains a distribution channel for its hardware and a gravitational pull on the open-source community that competitors will struggle to match. The acquisition also slots neatly into the broader NVIDIA AI stack that NVIDIA CEO Jensen Huang has been assembling since early 2026.

Before vs After at a Glance

Verdict: Who Wins, Who Watches

If you are a developer building on the platform, the open-access promise means your workflows stay intact for now. If you are an enterprise procurement team evaluating AI infrastructure, the deal reshapes the landscape — NVIDIA now controls the default starting point for open model distribution, complementing its gaming-grade pipelines like NVIDIA DLSS 5 Arrives on the consumer side. Watch for two signals in the next quarter: whether competing accelerators still get first-class support, and whether enterprise contracts quietly steer workloads toward NVIDIA-hosted inference.

Related Articles

• Open AI’s Rogue AI Agent Hacked More Than Just Hugging Face
• The OpenAI Models That Hacked Hugging Face Were ‘Active on the Internet’ for Days
• OpenAI: Hugging Face Breach Caused by Our Models

FAQs

Why is NVIDIA buying the platform?

NVIDIA gains the largest open-model distribution platform in AI, anchoring its hardware story to the community where most open-source training and inference already happens.

Will the platform remain open?

Jensen Huang publicly committed to keeping it open across models, frameworks, clouds, and accelerators, including hardware that competes with NVIDIA.

How much did NVIDIA pay?

The reported deal value is $12.93 billion unconfirmed, roughly 86 times the company’s $150 million in annualised revenue.

When was the deal confirmed?

NVIDIA officially confirmed the acquisition on Thursday, September 3, 2026, after a week of prior reporting from the outlet.