Nvidia to acquire Hugging Face for $12.9 billion

Less than a year ago, Hugging Face turned Nvidia down. In 2025 Nvidia proposed a $500 million investment that would have valued the startup at $7 billion, and Hugging Face rejected it, reportedly because it did not want a single dominant investor swaying its decisions ^[1]. Now Nvidia is reportedly agreeing to buy the whole company for $12.9 billion, valuing it above $13 billion - though as of publication the talks had not produced a signed agreement and could still fall through ^[2]. If it closes, this would be Nvidia's largest acquisition ever, dwarfing the $6.9 billion it paid for Mellanox in 2020 ^[2]. The jump in Hugging Face's own numbers helps explain the shift: annual revenue is now around $150 million, up from about $100 million just two months earlier, and the company is nearing profitability ^[3]. That is a long way from the $235 million Series D that valued the company at $4.5 billion back in August 2023, a round that included Nvidia itself alongside Google, Salesforce, and Amazon ^[4]. In three years, Hugging Face's valuation has nearly tripled, and the buyer that once got turned down for a minority stake is now reportedly buying control outright.

The strategic logic is less about Hugging Face's revenue and more about what it protects. A thriving open source model ecosystem gives developers and enterprises real alternatives to closed labs like OpenAI and Anthropic - and crucially, running those open models still requires Nvidia hardware, keeping demand tied to Nvidia's chips regardless of which lab wins the frontier-model race ^[2]. That matters more now because Nvidia's own biggest customers - OpenAI, Google, Amazon, and Anthropic - are all developing proprietary AI chips that could erode Nvidia's long-term data center dominance; strengthening the open model ecosystem creates a counterweight to those closed, vertically integrated rivals ^[5]. Forrester's Charlie Dai frames it as a layer play: Nvidia has already built an extended ecosystem through GPUs, CUDA, networking, inference software, and AI frameworks, and owning Hugging Face would extend that control to the developer, model distribution, and community layers on top of it, gaining a powerful developer channel and ecosystem asset ^[6]. There is also a more opportunistic angle - the deal could give Nvidia a path back into cloud computing after it scaled back its DGX Cloud ambitions, and a way to resell unused computing capacity ^[7]. On X, reaction to the price tag quickly folded into a broader narrative about Nvidia's spending habits this year, with one widely shared post describing the Hugging Face deal as just the latest in a run of investments - alongside a stake in Perplexity, a licensing deal and stake in Poolside, and continued backing for OpenAI, xAI, and Safe Superintelligence - that reads less like isolated bets and more like a company buying influence across the entire AI stack.

The biggest source of unease is that Hugging Face is not just Nvidia's shop window - it actively supports Nvidia's competitors. The platform maintains libraries like Optimum AMD and Optimum Intel that let developers run models on non-Nvidia hardware including AMD chips, Intel chips, Google TPUs, and AWS's own chips. Under Nvidia ownership, there is a real question of whether those libraries keep getting maintained, or whether search results and recommended defaults quietly start favoring CUDA-optimized models, eroding the platform's value as a neutral hub for the whole industry ^[8]. That skepticism isn't confined to Wall Street or the regulatory world - developer communities that rely on Hugging Face for hardware-agnostic tooling reacted with open concern as soon as the deal broke, some invoking the 'embrace, extend, extinguish' pattern associated with past corporate stewardship of open-source projects and drawing comparisons to how other widely used open platforms lost community trust after being folded into a single company's control. The core tension is the same one Forrester's analysis raises: the acquirer of an open ecosystem has every incentive, in the short term, to bend it toward its own hardware ^[6].

The deal is expected to face full antitrust review in multiple jurisdictions: a mandatory HSR filing with the FTC and DOJ in the US, scrutiny in the EU where active AI Act enforcement compounds existing friction between Nvidia and regulators over its Run:ai deal, and likely a UK review as well. The central question in each case is the same vertical relationship - what happens when the dominant AI chipmaker also owns the dominant open model distribution layer ^[9]. Analysts themselves are not aligned on whether that outcome is bad for the industry. Some, echoing Forrester's read, see the deal raising competitive pressure on hyperscalers, model vendors, and alternative hardware players, potentially forcing rivals to accelerate investment in independent model repositories and developer platforms of their own ^[6]. Others argue the opposite: that Hugging Face gaining Nvidia's capital, compute, and enterprise support could accelerate open model adoption broadly. As Tekonyx's Sid Nag puts it, analysts are simply split on whether this is a smart move for the industry, let alone for Nvidia itself ^[10].