NVIDIA Reportedly Acquires Hugging Face for $12.9B in Major Open-Source AI Expansion

NVIDIA Corp. has reportedly agreed to acquire Hugging Face for approximately $12.9 billion in a deal that is likely to reshape the artificial intelligence (AI) landscape.

The acquisition would mark one of NVIDIA’s largest-ever buyouts, dramatically expanding its influence across the AI software ecosystem.

While reports from The Information and Business Insider indicate the agreement is finalized, representatives for neither company have publicly confirmed the deal, and negotiations could still face late-stage hurdles. Microsoft Corp. had also previously held preliminary talks with the platform.

The purported $13 billion price tag represents a massive leap from August 2023, when New York-based Hugging Face secured a $4.5 billion valuation during a $235 million Series D round, a funding effort in which NVIDIA participated. The 10-year-old startup, which hosts open-weight models, datasets, and benchmark evaluations, is reportedly nearing profitability with annual revenues approaching $150 million.

If completed, the acquisition provides NVIDIA with a crucial strategic moat.

“Most enterprises already depend on Hugging Face without ever having procured it. It sits in AI build pipelines the way a package registry does, unexamined and unpriced, which is exactly what makes an ownership change consequential,” said Mitch Ashley, vice president and practice lead for Software Lifecycle Engineering and AI-Native Software Engineering at The Futurum Group. “The open models stay open. What changes is that a quiet dependency becomes a new vendor relationship without the perceived neutrality, pre-acquisition service this debt.”

As major tech competitors like Google, Amazon.com Inc., and OpenAI develop proprietary chips to reduce their dependence on NVIDIA hardware, control over Hugging Face anchors NVIDIA to the open-source developer ecosystem. Developers downloading open-weight models must run them on local or cloud-based hardware, which overwhelmingly relies on NVIDIA graphics processing units (GPUs).

Owning Hugging Face also offers NVIDIA an entry point back into cloud computing services, allowing the chipmaker to optimize and offload unused cloud capacity.

“Strategically, the deal makes enormous sense for NVIDIA. It could connect model discovery and development much more directly with NVIDIA’s GPUs, CUDA, NIM, DGX Cloud, and robotics portfolio,” said Stephanie Walter, practice lead for AI Stack and Enterprise Application Development at HyperFRAME Research. “NVIDIA would no longer simply provide the infrastructure beneath the open AI ecosystem. It would own one of its most important routes to market.”

“The concern is neutrality. Hugging Face’s value comes partly from serving models and technologies across competing hardware providers, clouds, and AI companies,” Walter added. “Open licenses can preserve access to models and code, but they cannot guarantee equal treatment in discovery, optimization, integration, or product priorities. The risk is not necessarily that Hugging Face stops supporting NVIDIA competitors. It is that the easiest and best-supported path gradually becomes an NVIDIA path.”

The deal comes amid heightened global debate over the security and regulation of open-source AI.

Last month, NVIDIA CEO Jensen Huang made his debut post on X to advocate against “premature restrictions” on open models, co-signing an industry letter alongside Hugging Face CEO Clément Delangue, Meta Platforms Inc., Amazon, and Microsoft. The coalition argued that open-weight models foster market competition and democratize technological gains.

Safety concerns surrounding open repositories have intensified following a startling security breach. OpenAI recently disclosed that an advanced model under internal testing, GPT-5.6 Sol, escaped its sandboxed environment, accessed the internet, and autonomously exploited a vulnerability in Hugging Face’s systems to access evaluation data.

Hugging Face revealed that proprietary American defensive models failed to stem the breach, forcing the platform to deploy an open-weight model — GLM 5.2, developed by China’s Z.ai lab — on its own infrastructure to neutralize the attack. The breach triggered local law enforcement notifications and drew sharp scrutiny from lawmakers concerned about autonomous cyber threats and national security.

“This isn’t a new lesson, it’s the same one enterprises keep running into. IT leaders should not build your AI strategy around one vendor’s model or one vendor’s platform,” said Luis Mesas, vice president of engineering services at SNGULAR. “We’ve been telling clients the shift that matters is having a gateway that can route to the right model for the right task, not betting everything on one provider staying neutral. NVIDIA owning the biggest open-model hub is exactly the kind of concentration that makes routing flexibility a requirement, not a nice-to-have.”

“Every infrastructure choice eventually gets the same question: What happens if I need to walk away from this vendor? For a lot of companies, running open models through Hugging Face was the answer, since you could self-host and weren’t locked into one frontier lab,” said Vinay Thakker, chief technology officer of KloudStax. “NVIDIA buying Hugging Face doesn’t break that overnight, but it does mean the fallback option now sits inside the same company selling the chips it runs on. That’s worth planning around the same way we’d plan around any other single point of dependency, not with alarm, just with the same recovery thinking IT teams have applied to every other system for years.”

NVIDIA’s blockbuster move reflects a broader consolidation trend across the AI sector, mirroring financial infrastructure provider Stripe Inc.’s recent $7 billion acquisition of OpenRouter. By absorbing Hugging Face, NVIDIA secures a commanding position over the software pipelines driving the next generation of AI.

“Hugging Face isn’t just another AI startup. It’s increasingly the distribution layer for the open-model ecosystem. If this deal holds, NVIDIA isn’t just supplying the compute underneath AI, it’s buying a much more direct relationship with the developers deciding what models get used,” said Joseph Hoefer, chief AI officer at Monument Advocacy. “And at nearly $13 billion, you’re not buying Hugging Face for its current revenue. You’re paying for developer distribution, ecosystem influence, and the possibility that the open-model layer becomes every bit as strategically important as the models themselves.”