Nvidia Acquires Hugging Face for $12.93 Billion, Consolidating Its Grip on AI Infrastructure

Nvidia confirmed on September 3, 2026 that it has acquired Hugging Face for $12.93 billion TechCrunch. The deal closes a negotiation arc that became public on August 27, when Reuters and Bloomberg both reported an imminent agreement valued at roughly $13 billion Reuters, Bloomberg. Bloomberg's Ian King and Rachel Metz subsequently placed the potential figure at about $14 billion Bloomberg/YouTube. The confirmed price landed between those markers.

Hugging Face operates what has become the default distribution layer for open machine learning. Think of it as the app store and shared library for AI models: the platform hosts three million models, one million applications, half a million datasets, and counts over 18 million developers among its users TechCrunch. Founded in 2016, the company had raised just over $395 million in total funding. Its last round, in 2023, brought in $235 million led by Salesforce Ventures, with participation from Google, Amazon, IBM, and Nvidia itself TechCrunch.

The Financial Times reported that Hugging Face rejected a $500 million acquisition offer from Nvidia last year TechCrunch. The gap between that figure and the $12.93 billion final price reflects both the trajectory of Hugging Face's business and the intensity of demand for AI infrastructure and distribution assets over the intervening period.

The Information reported that Hugging Face is generating approximately $150 million in annualized revenue TechCrunch. In a July 2026 interview, CEO Clem Delangue said the platform's growth rate was bringing it close to profitability TechCrunch. Nvidia, for its part, had already established a footprint on the platform: the company has released more than 500 models and 250 open datasets through Hugging Face TechCrunch.

Jensen Huang stated that Hugging Face will remain an open platform serving the entire AI ecosystem, and that Nvidia compute will not be required to build on or deploy through Hugging Face TechCrunch. Delangue said Hugging Face approached Huang seeking more compute, support, collaboration, and visibility TechCrunch, Delangue/X. The framing from both sides emphasizes continuity and platform neutrality.

The neutrality question is where the AI community's attention will focus. Hugging Face's value to the ecosystem comes substantially from its status as a vendor-agnostic hub, meaning developers can host models trained on AMD chips, deploy on Google Cloud TPUs, and build tooling that never touches Nvidia's CUDA software stack. Nvidia's assurance of openness addresses this directly. Whether it holds under full ownership by the world's dominant GPU maker is another matter. Structural enforcement, through governance, API contracts, or platform policies, will determine whether the assurance holds in practice.

Hugging Face has been expanding beyond pure ML infrastructure. In April 2025 the company acquired Pollen Robotics, a maker of open-source robots, and signaled plans to sell open-source robotic hardware through the platform Hugging Face Blog. That deal was Hugging Face's fifth acquisition, following additions including Gradio and XetHub Hugging Face Blog. More recently, the official Hugging Face blog landing page featured a post titled "Inside three.ws: The Open-Source Stack That Gives AI Agents a Body, a Brain, a Wallet, and a Job," published August 29, 2026 Hugging Face Blog. The trajectory points toward platform ambitions extending into embodied AI (AI that controls physical robots), agentic systems (AI that takes actions autonomously), and financial transaction infrastructure for those autonomous agents.

The acquisition gives Nvidia ownership of the primary distribution channel for open ML models and datasets while the company's silicon already powers the majority of large-scale training and inference workloads worldwide. The vertical integration is straightforward: Nvidia builds the GPUs, provides the compute platforms, and now controls the repository where most open models are hosted, evaluated, and downloaded. The revenue multiple is steep relative to Hugging Face's $150 million annualized revenue, but the strategic asset is not the revenue line. It is the developer relationship. Eighteen million developers pass through Hugging Face's platform. Nvidia has now purchased that relationship outright.

The broader context here is that the $12.93 billion price, the rejected $500 million offer from a year prior, and the revenue figure together sketch a company whose leverage increased sharply as the AI buildout accelerated. Hugging Face held out, and the hold-out paid off. Whether the platform's openness survives the transition from independent entity to Nvidia subsidiary will shape how the open-source AI community organizes itself going forward. Alternatives exist, but none carry Hugging Face's gravity.