Did NVIDIA buy Hugging Face? What the reported $12.9 billion deal means
Reports say NVIDIA has agreed to buy Hugging Face, the 'GitHub of AI', for about $12.9 billion. Neither company has confirmed it, and it would face a long antitrust review. Here is what is actually known, and why it worries open-source AI.
Multiple reputable outlets report that NVIDIA has agreed to acquire Hugging Face, the open-source hub often called the "GitHub of AI," for about $12.9 billion. The important caveat first: this is reported, not confirmed. Neither company has said anything publicly, no signed final agreement has been reported, and the deal has not been through any regulatory review. If it is real, it is a landmark: NVIDIA would own the neutral distribution layer where most of the world's open AI models are shared, on top of already dominating the chips they run on. That is exactly why it is controversial, and why an antitrust review that could run past a year stands between the report and reality.
At the end of August, a story broke that would reshape the AI landscape if it holds up: NVIDIA, the company whose chips power almost all modern AI, has reportedly agreed to buy Hugging Face, the platform where the open-source AI world lives. The headline number is roughly $12.9 billion. Before anyone treats it as done, though, it is worth being precise about what is actually known versus what is still an unconfirmed report, because right now the gap between the two is the whole story.
Is the deal actually confirmed?
No. This is the part that matters most, and the part fast-moving coverage often blurs. As of now:
- The story is reported, sourced to people familiar with the matter, not announced by either company.
- Neither NVIDIA nor Hugging Face has confirmed or denied it. NVIDIA did not address it on its most recent earnings call.
- Reporting indicates no signed final agreement yet, and at least one outlet notes the deal "could still fall apart."
- It has not passed, or even formally entered, antitrust review.
The specific "NVIDIA agreed, $12.9 billion" claim was first reported by The Information in late August, then followed by CNBC, Fortune, Bloomberg, TIME and others, most of which carefully hedge it as reporting. So: credible, widely repeated, but firmly in "reported, not confirmed" territory. Everything below assumes the reports are accurate.
The price, and what it signals
The reported ~$12.9 billion would value Hugging Face far above its $4.5 billion valuation from its 2023 funding round, roughly a threefold jump in about three years. It is also a striking reversal: in late 2025 Hugging Face reportedly turned down a $500 million investment from NVIDIA that valued it near $7 billion, because it did not want a single dominant investor able to sway its decisions. With annual revenue reported around $150 million, a $12.9 billion price is roughly 86 times revenue, which tells you this is a strategic bet on position, not a bet on current earnings.
What Hugging Face is, and why it matters
If you are outside AI, the name may mean nothing; inside AI, it is infrastructure. Founded in 2016, Hugging Face became the "GitHub of AI": the default place to share, discover and download open models and datasets. It hosts more than two million models, serves around 13 million developers, and maintains the Transformers library that is the near-universal way to run and fine-tune open models. Its whole brand is neutrality, any developer can pull any model and run it on any hardware. That vendor-agnostic position is precisely what an NVIDIA acquisition would call into question.
Why NVIDIA would want it
NVIDIA already sells the chips almost every model runs on. Buying Hugging Face, as analysts read it, would add the distribution layer on top:
- Own the front door. Hugging Face is where developers discover and pull open models. Controlling it complements NVIDIA's hardware with the software channel that feeds it.
- Hedge against its own biggest customers. As large labs build their own chips, a thriving ecosystem of thousands of smaller open-model builders, who still run on NVIDIA GPUs, keeps NVIDIA central.
- Feed the GPU flywheel. Models pulled from the hub still have to be hosted and run, usually on NVIDIA hardware; ownership lets NVIDIA steer defaults toward its own tools and accelerators.
- A route into hosted inference, since Hugging Face already rents compute to run models.
The worry: what happens to a neutral hub
The loudest reaction is concern about neutrality. A neutral repository lets any developer pull any model and run it anywhere; a vendor-owned one could, over time, steer that flow toward the owner's hardware, cloud and tooling. As one framing put it, the models stay openly licensed, but the front door changes hands. Critics fear the hub drifts from a shared commons toward a corporate channel optimised for NVIDIA, and rival chipmakers like AMD and Intel may push their own platforms harder in response. The counterargument, and analysts are genuinely split, is that NVIDIA has funded open-source AI before and has a self-interest in keeping the hub broadly open, since a locked-down hub would lose the very developers that make it valuable.
The regulatory hurdle
Unlike NVIDIA's recent structured investments in smaller AI firms, an outright acquisition of Hugging Face would trigger a mandatory antitrust review, which analysts expect could take 12 months or more, with a real possibility of conditions or an outright block. NVIDIA is already under antitrust scrutiny over how it allocates GPUs. The central question regulators are expected to ask is whether NVIDIA would favour its own hardware inside Hugging Face's hosting and inference, disadvantaging rival chips. In other words, even if the reports are accurate, this is not closing tomorrow.
What to watch next
Four signals will tell you whether this moves from rumour to reality: an official statement (or denial) from either company; a signed agreement; a regulatory filing; and, crucially for developers, any commitment to keep the hub neutral and open. Until the first two happen, treat it as a very well-sourced report, not a done deal. For the bigger picture, see our take on the open-versus-closed AI fight and whether AI is a bubble, and browse our full AI coverage.
Frequently asked questions
Did NVIDIA buy Hugging Face?
Not confirmed. Multiple outlets report that NVIDIA has agreed to acquire Hugging Face for about $12.9 billion, but neither company has confirmed it, no signed deal has been reported, and it has not cleared any regulatory review.
How much did NVIDIA reportedly pay for Hugging Face?
About $12.9 billion, according to reports (some round it to "nearly $13 billion"). This is the reported figure, not an officially confirmed price.
Is the Hugging Face deal official?
No. As of now it is reported and said to be agreed in principle, but there is no public confirmation and reportedly no signed final agreement; one report notes it could still fall apart.
What is Hugging Face?
Often called the "GitHub of AI," it is the leading open-source hub for AI models and datasets, hosting more than two million models, serving around 13 million developers, and maintaining the widely used Transformers library.
Why does NVIDIA want Hugging Face?
To own the distribution layer where developers find and download open models, reinforcing its dominance in the chips those models run on and hedging against big customers building their own hardware.
What was Hugging Face worth before this?
About $4.5 billion at its 2023 funding round. It reportedly declined a $500 million NVIDIA investment at a roughly $7 billion valuation in late 2025.
Will Hugging Face stay open-source and neutral under NVIDIA?
That is the biggest open question and the main controversy. Open licences on the models would not change, but critics worry the hub's neutral "front door" could be steered toward NVIDIA's hardware and cloud. No commitments have been announced.
Could regulators block the NVIDIA–Hugging Face deal?
Possibly. An outright acquisition would face a mandatory antitrust review that analysts expect could take a year or more, with a real chance of conditions or a block, especially as NVIDIA is already under antitrust scrutiny.
Have NVIDIA or Hugging Face commented?
No. Neither company has publicly confirmed or denied the reports, and NVIDIA did not raise it on its recent earnings call.
When would the deal close?
Unknown. Even if the reports are accurate, an antitrust review could push any close beyond a year, so nothing is imminent.
OpenAI's 'Jalapeño' chip beat Nvidia on efficiency: what it actually means
OpenAI showed off its first in-house AI chip and said it does more work per watt than Nvidia's Blackwell. The claim is real, but the fine print matters: these are OpenAI's own benchmarks, against Nvidia's older memory, from a chip that barely ships yet.
Is AI a bubble? The money is real, and so is the circle it moves in
The AI build-out is one of the biggest infrastructure bets in history, funded by a web of deals in which the same companies invest in, sell to and buy from each other. Critics call it round-tripping; defenders call it vendor financing. Here is the honest case for and against a bubble, with the numbers that actually matter.
Nvidia and Cisco just backed a startup that wants to build AI data centres in orbit
Starcloud added a 250 million dollar extension to its round, valuing the orbital-data-centre startup at 2.3 billion, with Nvidia and Cisco Investments joining. The pitch: meet AI's power demand in space. It runs one Nvidia GPU in orbit so far, and the plan rides on cheaper Starship launches.