Nvidia to Buy Hugging Face for $12.9 Billion

Free Instant SEO Audit

Nvidia acquiring Hugging Face symbolized by glowing data cubes on a futuristic bridge

By NewSunSEO September 23, 2026

Nvidia has agreed to acquire Hugging Face for $12.9 billion, giving the chipmaker direct control of the largest open-source AI hub on the internet and a fast-growing cloud and storage business built on top of it. The deal values the ten-year-old startup at roughly 86 times its $150 million in annualized revenue and nearly triple the $4.5 billion valuation it fetched in 2023.

For developers, the acquisition brings the platform’s model hub, GPU rental layer and enterprise storage under the same roof as the world’s leading AI accelerator company. For Nvidia, it secures a distribution channel for its GPUs in the one corner of AI it does not already control.

What the Deal Covers

Hugging Face is best known as a repository for community-built models, hosting more than 3 million open-source large language models and growing toward 4 million. Beneath that surface, the company has quietly built a vertically integrated AI platform that now spans frontier model distribution, physical robotics, and an enterprise storage business that the company’s co-founder says is gaining traction faster than almost anything else the company has built.

The acquisition was reported the same day Nvidia disclosed $96.2 billion in quarterly revenue and forecast a 70 percent jump in revenue next fiscal year. The chipmaker also disclosed $18 billion committed to equity investments through 2027. At its current revenue pace, $12.9 billion represents about thirteen days of sales, a price tag for securing the open-source distribution layer that competes with closed labs at OpenAI, Google, Amazon and Anthropic.

In January, Hugging Face had turned down a $500 million investment from Nvidia at a $7 billion valuation, a signal that independence mattered more than a quick capital injection. Nvidia came back with a different offer. This time, it bought the whole thing.

From Model Hub to Cloud and Storage Company

What headlines about the deal often miss is that Hugging Face is already a cloud company, and that is precisely the layer Nvidia wanted. For years, Hugging Face has built a managed compute platform on top of its model hub. Developers can spin up Inference Endpoints, dedicated GPU APIs priced from $0.03-per-hour CPU instances up to $80-per-hour clusters of eight Nvidia H100s. They can host interactive demos on Spaces, paying by the hour for GPU hardware. They can route API calls through Inference Providers, a billing layer that connects users to third-party GPU clouds such as Together AI, SambaNova and Groq.

Most importantly, Hugging Face and Nvidia already run Training Cluster as a Service, a joint product that gives any of Hugging Face’s 500,000 organizations on-demand access to large GPU clusters, billed only for the duration of a training run. It functions as a distribution channel for Nvidia compute dressed up as a developer tool.

That channel matters because Nvidia’s own cloud business has reportedly struggled. The company scaled back its DGX Cloud offering roughly a year ago. Owning Hugging Face gives Nvidia a way back into cloud computing without starting from scratch, inheriting a platform where developers already rent GPUs, already pay for inference and already trust the brand.

Hugging Face’s cloud services, storage and subscriptions produced roughly $150 million in annualized revenue as of summer 2026, up from about $100 million two months earlier. The revenue line is still small in Nvidia terms. Nvidia is buying a distribution channel, not a top-line contribution.

The Enterprise Storage Pivot

Because the hub houses petabytes of model weights and datasets, the engineering team built an ultra-efficient blob storage system. The company now sells that storage to customers storing petabytes or terabytes of data, and the unit has drawn strong traction.

In June, AI lab Arcee became the first major American company to replace AWS S3 with Hugging Face Private Storage in a multi-million dollar commercial partnership. When customers migrate off Amazon to store data with you, you are no longer just another open-source project. You are infrastructure.

The Robotics Expansion

Hugging Face’s hardware pivot started with software. The company released an open-source library for robotics that gained traction quickly, but the team noticed a bottleneck code could not fix: hardware was very expensive, with entry-level arms running from several thousand dollars up to $50,000. For a software developer who wanted to experiment with physical AI, the barrier was insurmountable.

The first response was a robotics arm kit priced at roughly $100, which sold more than 10,000 units in its first year. The company is on track to sell 20,000 this year. That momentum pushed Hugging Face to acquire French robotics startup Pollen Robotics, which now forms the core of its physical AI division.

A second consumer robot kit, the Reachy Mini, is ready for release this summer priced between $399 and $499. On the day the Nvidia acquisition was reported, the company unveiled Microduck, a 25-centimeter open-source biped with 15 actuators and a full sensor suite including camera, speaker, LiDAR, NFC, Bluetooth and Wi-Fi. It is designed to be trained from scratch with reinforcement learning and ships with more than half a dozen pre-trained policies so it can walk, sit, crouch, roller-skate, pick up objects with its articulated beak, and recover on its own, all for $399. Order volume reached over $2.6 million after the launch announcement.

Where Open-Source AI Stands in 2026

If 2025 was the year enterprise discovered open-source AI, 2026 is the year open-weight models caught up to the frontier. GLM 5.2 by Z.ai, released weeks before the acquisition was reported, was described as surprisingly close to frontier closed models. Google’s Gemma 4 was called out as a personal favorite for its quality and small footprint, with a demo released showing live conversation with the model running locally.

An experiment with 100+ AI agents collaborating freely on an open project produced a 5x inference speedup on Gemma 4 inside vLLM, one of the more striking emergent behaviors seen from agent swarms to date. The open ecosystem continues to iterate in public, producing specialized models for vision, scientific computing, and coding that cost nothing to download and pennies to run.

The hub’s trending leaderboard has become the starting point for developers who want to know which open-weight models are genuinely useful right now. A recent partnership with Qualcomm is also pushing AI inference toward edge chips.

What Happens Next

The deal has not yet closed and remains subject to regulatory approval. The financial figures and product details reflect the most recent publicly available information as of late August 2026.

For developers, the practical question is whether the open ecosystem that made Hugging Face valuable stays open under Nvidia’s ownership, or whether the distribution layer tightens to favor Nvidia hardware and partner clouds. Hugging Face’s co-founder turned down Nvidia once before. This time, he said yes. Whether the company keeps following talent and problems independently, or becomes a channel for Nvidia compute, will define what open-source AI looks like for the rest of the decade.

FAQ

How much is Nvidia paying for Hugging Face?

Nvidia agreed to acquire Hugging Face for $12.9 billion, valuing the company at roughly 86 times its $150 million in annualized revenue and nearly triple its $4.5 billion valuation from 2023.

Why does Nvidia want Hugging Face?

Hugging Face already runs a managed cloud layer on top of its model hub, including GPU rental APIs, a Spaces demo platform and a joint Training Cluster as a Service product with Nvidia. Owning it gives Nvidia a distribution channel for its GPUs in open-source AI, the one corner of the market where it does not already compete directly with closed labs.

What does Hugging Face actually do beyond hosting models?

Hugging Face hosts more than 3 million open-source AI models, sells enterprise storage built from its in-house blob storage system, runs paid GPU inference and training services, and now sells affordable open-source robotics kits including the Reachy Mini and the Microduck biped.