Nvidia Confirms $12.9 Billion Deal to Buy AI Model Platform Hugging Face
Nvidia says it will pay about $11.9 billion to Hugging Face shareholders plus roughly $1 billion in retention equity, in a deal it expects to close in the first half of 2027 pending regulatory review.
Nvidia Just Bought the Shelf Where Its Rivals Stock Their Products
Nvidia announced on September 3, 2026, that it will pay about $12.93 billion to buy Hugging Face, the website most of the world's AI developers use to find and download open-source models[1][3]. About $11.9 billion goes to Hugging Face's shareholders. Another $1 billion is Nvidia stock set aside to keep Hugging Face's own staff from leaving after the deal closes[2]. Nvidia expects the sale to wrap up in the first half of 2027, after regulators get a chance to review it[1][3].
Here is the detail that makes this more than a big number: Hugging Face isn't just Nvidia's property. It's the shared library that Nvidia's competitors depend on too. AMD, Intel, Google and Amazon all post their AI models there. So do Chinese labs like DeepSeek[5][18]. The site works, in part, because no single hardware company owns it. That is about to change.
A Library Everyone Uses, About to Get a New Landlord
Hugging Face is often called the "GitHub of AI." It hosts more than 3 million models, 500,000 datasets and 1 million apps. More than 18 million developers and 200,000 companies use it[1]. It was valued at $4.5 billion in a 2023 funding round — this sale values it at nearly three times that[3].
CEO Clément Delangue says open-source AI hit a "turning point" and needed more money and scale than an independent startup could provide on its own[4]. He says he reached out to Jensen Huang, Nvidia's CEO, over the summer, rather than being approached first[2]. Hugging Face also says it turned down several earlier offers before choosing Nvidia specifically because of its stated commitment to keeping the platform open[8].
That commitment is the whole ballgame. Running a free hosting service for millions of models and datasets is expensive, with little direct revenue to show for it[1]. A sale at triple its last valuation gives Hugging Face's investors and founders a payout, and puts a much bigger balance sheet behind the hosting bill[3].
The Promise: We Won't Touch the Scale
Huang has made a specific, checkable pledge: Hugging Face will "remain an open platform for the entire AI ecosystem," and developers will not need Nvidia chips to build or deploy anything through it[1]. Nvidia goes further, calling the deal a "deconcentration platform" — its argument is that spreading open models widely weakens the handful of closed AI labs, rather than concentrating power in Nvidia's hands[13].
There's a real business reason behind that generosity. Nvidia's biggest customers — OpenAI, Google, Amazon, Anthropic — are all building their own AI chips to rely on Nvidia less[2]. Every dollar of AI work that shifts from those closed labs toward open models is a dollar that still tends to run on Nvidia's software, called CUDA. CNBC has described the purchase as a "defensive move[10]." Owning the front door protects demand for what's sold behind it.
CUDA is the key piece of jargon here, and it explains why this deal is about more than the sale price. It's the programming layer developers write their AI code in to run on Nvidia chips. Switching that code to run on a rival's chips, like AMD's, takes real work — so once a developer builds on CUDA, they tend to stay. Hugging Face is where developers make their first choice of model and software. Whoever shapes that first choice shapes years of future spending[5].
Why Rivals Aren't Reassured
Critics don't dispute Nvidia's promise — they dispute whether a promise from an interested party is worth much. The Register, a British tech publication, argues Hugging Face is "too important to fall into Nvidia's hands[5]." It points to two specific ways Nvidia could tilt the field without ever breaking its word.
First is ranking. Hugging Face decides which models get shown first to a developer browsing the site. Small changes to what surfaces on top could quietly steer traffic toward Nvidia-friendly software and away from AMD's competing option, called ROCm[5]. Second is information. Owning the platform means Nvidia would see download and usage data for rival companies' projects — a window no competitor gets today[5].
This isn't a new argument for Nvidia to face. In 2022, Nvidia tried to buy the chip designer Arm for $40 billion and abandoned the deal after U.S. and other regulators objected, on similar grounds about controlling shared infrastructure[5]. Nvidia's own defense is that this case is different: Arm designed chips that Nvidia's rivals sold directly, making that a deal between direct competitors. Hugging Face sells no chips at all, which Nvidia would frame as a purchase of a different layer of the business rather than a rival.
That distinction will matter because, unlike many of Nvidia's recent deals, this one can't dodge a formal review. Nvidia's other partnership-style investments have often avoided a full antitrust filing. But because this is a straight purchase, it triggers something called a Hart-Scott-Rodino filing — a mandatory premerger notice that gives the FTC and Department of Justice a formal window to examine the deal before it can close[14]. Any neutrality promise only becomes an enforceable rule if regulators attach it as a condition of approval.
A French Company, Sold to an American Buyer
There's a second dispute running alongside the antitrust one, and it's about geography rather than markets. Hugging Face was founded by three Frenchmen and got its start in Paris, even though it's now headquartered in the U.S. with mostly American investors[8]. French Economy Minister Roland Lescure called the sale a "wake-up call," saying: "If we don't have enough capital in Europe to bring our champions to the next stage, they're going to go and find the capital elsewhere[8]."
That's not a complaint about Nvidia specifically. It's a complaint about a pattern — European companies build something valuable, then need late-stage money that isn't available at home, so they end up sold to American buyers. Europe can write AI rules, but this deal is a reminder that writing rules is different from owning the infrastructure those rules apply to.
There's also a quieter risk sitting underneath all of this, tied to who actually uses Hugging Face. Chinese AI labs are a huge part of what makes the platform valuable — DeepSeek is its single most-followed publisher, and as of July 2025, eight of the top ten models on its open leaderboard came from China[18]. Huang himself has praised DeepSeek, Alibaba, Tencent and other Chinese labs' models as "world-class[18]." Hugging Face's own risk filings warn that U.S. government restrictions on Chinese-origin AI models could hurt its business — a risk that existed before this sale, but one that gets sharper now that the owner is a major American chip company[8][18].
How the Story Got Told
None of this has moved Nvidia's stock in a worried direction. Shares rose 2.53% to $234.22 after the announcement[12], and Raymond James kept its "Strong Buy" rating, calling the price "largely immaterial" to Nvidia's finances but "strategically valuable" for controlling a key entry point into the AI model world[11].
How each outlet told the story split largely along which fact they put first. Fox Business and The Daily Caller led with the price tag and Nvidia's own promise of neutrality[7][15]. CNN led instead with a security incident — headlining Hugging Face as "the AI startup that was hacked by OpenAI," a real episode where Hugging Face says it had to use an open-source Chinese model to defend itself because closed-model license terms blocked other options[4]. AFP centered its coverage on Hugging Face's French roots and Lescure's capital-flight warning[8], while The Register supplied the sharpest technical detail on both the ROCm and data-access concerns[5].
Nothing about the deal changes what Hugging Face does technically, at least not yet. What it changes is who controls the defaults, the rankings, and the usage data behind a platform that millions of developers already treat as neutral ground[1][5]. The one thing that would settle the argument — a regulator turning Nvidia's open-platform pledge into a written, enforceable condition of approval — hasn't happened. Until it does, every side is still arguing about intent, not obligation, and the deal isn't expected to close until sometime in the first half of 2027[1].
Summary
Nvidia said on September 3, 2026 that it will buy Hugging Face, the website where most of the world's open AI models are shared and downloaded[1][3]. The total price is about $12.93 billion. Roughly $11.9 billion goes to Hugging Face shareholders. Another $1 billion is Nvidia stock set aside to keep Hugging Face employees from leaving after the deal[2]. Nvidia expects the deal to close in the first half of 2027, after regulators review it[1][3]. It would be Nvidia's second-largest purchase ever, behind the roughly $20 billion it paid for assets of chipmaker Groq in December[2].
Hugging Face is often called the "GitHub of AI." It hosts more than 3 million models, 500,000 datasets and 1 million apps, used by more than 18 million developers and over 200,000 companies[1]. Crucially, it has been vendor-neutral ground. Nvidia's rivals — AMD, Intel, Google, Amazon — publish there, and so do Chinese labs like DeepSeek[5][18]. That neutrality is the heart of the argument.
Nvidia says nothing about that changes. CEO Jensen Huang says Hugging Face will "remain an open platform for the entire AI ecosystem," and that Nvidia chips will not be required to build or deploy through it[1]. Nvidia goes further and calls the deal a "deconcentration platform" — its claim is that spreading open models widely weakens the handful of closed AI labs, rather than concentrating power[13]. Critics do not accept that. The Register argues a dominant GPU seller should not own the shared shelf its competitors stock, and warns Nvidia would gain a view into download and usage data for rival projects[5]. Nvidia's $40 billion attempt to buy chip designer Arm collapsed in 2022 on a very similar objection[5].
The genuine point of dispute is not the price. It is whether a promise of neutrality, made by the company with the most to gain from steering developers toward its own hardware, is enforceable. Investors were largely positive: Nvidia stock rose 2.53% to $234.22 after the news[12]. Regulators have not ruled. Because this is a direct purchase, not a partnership, it requires a Hart-Scott-Rodino premerger filing — which gives the FTC and DOJ a formal window to examine it[14].
The Event
On September 3, 2026, Nvidia confirmed an agreement to acquire Hugging Face, the open-model hosting platform, for about $12.93 billion[1][3]. The structure is roughly $11.9 billion to Hugging Face shareholders plus about $1 billion in Nvidia equity to retain arriving employees[2]. Nvidia said the deal is expected to close in the first half of 2027, subject to regulatory approval[1][3]. The Information had reported the agreement on August 26-27, before Nvidia's own confirmation[9][19].
Undisputed Facts
- Nvidia announced the agreement on September 3, 2026, and states the total transaction value at about $12.93 billion[1][3].
- About $11.9 billion is payable to Hugging Face shareholders; about $1 billion is Nvidia equity for employee retention[2].
- Nvidia says the deal is expected to close in the first half of 2027 and is subject to regulatory approval[1][3].
- Hugging Face hosts more than 3 million models, 500,000 datasets and 1 million applications, and reports more than 18 million developers and over 200,000 companies using it[1].
- Hugging Face was valued at $4.5 billion in a 2023 funding round — roughly a third of this deal's value[3].
- Nvidia's largest prior deal was a roughly $20 billion arrangement in December 2025 to license chipmaker Groq's technology and hire its staff — CEO Jensen Huang said at the time "we are not acquiring Groq as a company," making the Hugging Face purchase Nvidia's largest actual corporate acquisition[2].
- Jensen Huang stated publicly that Hugging Face will remain an open platform and that Nvidia compute will not be required to build on or deploy through it[1].
- Nvidia's proposed $40 billion purchase of Arm was abandoned in 2022 after opposition from the U.S. FTC and other regulators[5].
- Because it is a direct acquisition, the deal requires a Hart-Scott-Rodino premerger filing with U.S. antitrust agencies[14].
- Nvidia shares rose 2.53% to $234.22 following the announcement[12].
The Pressure
Strip away the moralizing and blame. What structural realities persist regardless of which narrative wins?
- Nvidia's customers are becoming its competitors
- OpenAI, Google, Amazon and Anthropic are all developing their own AI chips to cut reliance on Nvidia[2]. Every dollar of demand that shifts from those closed labs to open models is a dollar that still tends to run on Nvidia's CUDA software. Funding the open ecosystem is not charity — it is demand insurance[10].
- Software lock-in, not silicon, is the moat
- CUDA is Nvidia's programming layer. Developers write for it, and their code does not move easily to AMD's ROCm equivalent. Hugging Face is where developers first pick a model and a runtime. Whoever shapes that first choice shapes years of downstream spending[5].
- Open hosting is a cost center
- Storing 3 million models and 500,000 datasets and serving 18 million developers costs real money with weak direct revenue[1]. Hugging Face's independence was always financially fragile, which is what made a strategic buyer plausible in the first place.
- Merger review is the only binding lever
- Nvidia's recent expansion has often used investments and partnerships that avoid formal premerger review. This is a straight purchase, so it triggers an HSR filing and gives the FTC and DOJ a formal window[14]. Any neutrality commitment only becomes enforceable if a regulator makes it a condition.
- European capital gap
- France's economy minister framed the sale as evidence that European companies must raise late-stage money abroad and then get bought there[8]. That constraint operates regardless of anyone's view of Nvidia.
Material realityHugging Face is the default distribution point for open AI models, and its value comes from being neutral — AMD, Intel, Google, Amazon and Chinese labs all publish there[5][18]. After this deal, that neutral ground would be owned by the company that sells most of the hardware those models run on. Nothing in the announcement changes what the platform does technically; the change is who controls its defaults, its rankings, and its usage data[1][5]. The price is small relative to Nvidia — analysts called it "largely immaterial" to the balance sheet[11] — which means the deal is about position, not revenue. Chinese labs remain central to the platform's value: DeepSeek is its most-followed organization, and eight of the top ten open models as of July 2025 came from China[18]. That is also the platform's largest political risk, since U.S. restrictions on Chinese-origin models would hit it directly[8]. Regulators have not yet acted, and the deal is not expected to close until the first half of 2027[1].
Narrative as a weaponNvidia is the most active narrative shaper here, and its chosen word is doing the heavy lifting: calling Hugging Face a "deconcentration platform" reframes an acquisition by the dominant hardware vendor as an anti-monopoly act[13]. It wants you to believe the relevant monopoly threat is closed AI labs, not Nvidia. Hugging Face's leadership reinforces this by stressing that it approached Nvidia and turned down other bidders — a story of choice rather than capture[2][8]. Critics, led on specifics by The Register, want you to believe the Arm precedent is the right analogy, and that a promise of neutrality from an interested party is worth little without an enforceable condition[5]. U.S. business press largely runs the company's framing with a stock-reaction lens attached[7][12]. French officials and European commentators are pushing a third frame entirely — that the real story is Europe's inability to keep its own companies[8]. Watch for one thing that would cut through all of it: whether any regulator converts Nvidia's open-platform pledge into a written condition of approval. Until then, every side is arguing about intent, not obligation.
How Each Side Sees It
Each major actor’s view — how it frames things, its underlying incentive, and how it’s materially affected. Tap a side to read it.
Frames it asNvidia's case is that open AI is the thing keeping the field from collapsing into a few closed labs, and it needs a funded, durable home[1]. Nvidia calls the platform a "deconcentration platform" — its argument is that if anyone can download a strong open model, no single lab can gate access to intelligence[13]. On neutrality, Nvidia makes a specific, checkable promise rather than a vague one: developers keep choosing their own models, frameworks, clouds and chips, and Nvidia hardware is not required[1]. Nvidia would also note the structural difference from the failed Arm deal. Arm designed the chips its rivals sold; Hugging Face sells no chips at all. In antitrust terms that makes this a vertical deal — a buyer purchasing a different layer of the stack — not a horizontal one that removes a direct competitor.
WhyNvidia's chip business faces a real threat: its biggest customers — OpenAI, Google, Amazon, Anthropic — are building their own AI chips to depend on Nvidia less[2]. A large, healthy open-model ecosystem gives buyers an alternative to those closed labs, and open models overwhelmingly run on Nvidia's CUDA software today. CNBC characterized the purchase as a "defensive move"[10]. Owning the front door to open AI protects demand for what is behind it.
Impact on themThe price is small for Nvidia. Raymond James kept a 'Strong Buy' and called the deal "largely immaterial" to Nvidia's balance sheet but "strategically valuable," describing it as "less about buying revenue and more about controlling a critical entry point in the AI model ecosystem"[11]. The real exposure is regulatory: an HSR review reopens the same fight Nvidia lost over Arm[5][14].
Frames it asDelangue's stated case is that open-source AI hit a "turning point" and needed "more resources, more scale, more visibility" than an independent startup could supply[4]. He says he approached Huang over the summer, rather than being hunted[2]. Hugging Face also says it turned down several earlier stake purchases and takeover offers, and chose Nvidia specifically for its commitment to the open ecosystem[8]. The company's implicit argument is that neutrality costs money — hosting 3 million models and 500,000 datasets for free is expensive, and a buyer that profits from open AI succeeding is a better owner than one that profits from it failing.
WhyTurning a free, infrastructure-heavy platform into a sustainable business is hard. A sale at roughly triple the 2023 valuation returns capital to investors and founders and puts a much larger balance sheet behind the hosting bill[3]. The $1 billion retention pool signals that keeping the staff is a condition of the deal working[2].
Impact on themHugging Face's leverage after closing depends on unwritten commitments. Its own risk disclosures flag that government restrictions on AI models originating in China could materially harm its business — and Chinese labs are among its most-followed publishers[8][18].
Frames it asThe core argument is a principle, not a prediction: a platform everyone depends on should not be owned by one of the competitors who depends on it. That is the exact reasoning that killed the Arm deal in 2022[5]. Critics point to two specific mechanisms. First, ranking and defaults — Hugging Face is where a developer decides which model to run, and small changes to what surfaces first can steer traffic toward Nvidia-optimized builds and away from AMD's ROCm software stack, the open competitor to Nvidia's CUDA[5]. Second, information — Nvidia would see download and usage patterns for rival projects, a view no competitor gets today[5]. The Register's framing is blunt: Hugging Face is "too important to fall into Nvidia's hands"[5]. Note the crux differs by side. Nvidia asks whether it will abuse the platform. Critics say the question is whether it could, and whether a promise is enforceable.
WhyAMD, Intel and the large cloud providers all publish on Hugging Face and compete with Nvidia in chips[5]. Their interest is preserving a distribution channel they do not have to rent from a competitor.
Impact on themIf the deal closes without conditions, rivals rely on Nvidia's good behavior. If regulators impose conditions — data walls, ranking neutrality commitments, governance guarantees — those become enforceable terms rather than blog-post pledges.
Frames it asHugging Face was founded by three Frenchmen and incubated in Paris, even though it is headquartered in the U.S. with heavily American investors[8]. French Economy Minister Roland Lescure's argument is about capital, not xenophobia: "If we don't have enough capital in Europe to bring our champions to the next stage, they're going to go and find the capital elsewhere." He called the deal a "wake-up call"[8]. The steelman here is that Europe can write AI rules but cannot own AI infrastructure, and rules without ownership are weak.
WhyEuropean policymakers want domestic AI infrastructure they can regulate and rely on in a crisis. Each late-stage exit to a U.S. buyer weakens that case at home.
Impact on themThe EU is one of the jurisdictions that can review the deal; it was part of the coalition that sank Nvidia's Arm bid[5]. Some European observers argue the opposite way — that Nvidia's money could actually strengthen open European AI access[8].
Frames it asThis group's concern is practical rather than ideological: will the models stay downloadable, and will publishing stay free? Chinese labs are a large share of what makes the platform valuable. DeepSeek is the most-followed organization on Hugging Face, and as of July 2025 eight of the top ten models on its open leaderboard came from China[18]. Huang himself has praised DeepSeek, Alibaba, Tencent, MiniMax and Baidu models as "world-class"[18]. The strongest version of the worry is that an American owner becomes a chokepoint under U.S. export or security policy — that the platform stops being a global commons and becomes a jurisdiction.
WhyDevelopers want the cheapest, freest access to model weights. Chinese labs want global distribution and mindshare, which Hugging Face has provided.
Impact on themHugging Face's own filings warn that U.S. government restrictions on Chinese-origin models could materially hurt its business[8]. That risk existed before the deal, but a U.S. chip giant as owner raises the stakes.
Like this article?
The Bias Ledger average rating 4.7
The same story, as framed by outlets across the spectrum, ordered least to most biased. The bias score (1 = straight, 10 = heavily spun) is an AI assessment of that framing — click an outlet to see its track record. The tell is the word choice or omission that reveals the angle.
| Outlet | Vantage | Bias | How they frame it | The tell |
|---|---|---|---|---|
| CNBC | U.S. center, business press | 3 | "Hugging Face approached Nvidia's Huang weeks ahead of $12.9B acquisition, CEO tells CNBC" — and separately, "Why Nvidia's 'defensive move' to acquire Hugging Face is about much more than chips"[2][10]. | Leads with the CEO's own account of who approached whom, which is favorable to both companies and unverifiable outside their statements. The follow-up piece does supply the sharper analysis — that this protects Nvidia against customers building their own chips — but only after the access interview ran first. |
| Fox Business | U.S. right, business | 3 | "Nvidia to acquire Hugging Face for $12.9B to expand AI platform strategy"[7]. | Straight transactional framing built around Nvidia's stated strategy. The word "expand" adopts the company's own growth language, and the antitrust question is not the organizing frame. |
| Agence France-Presse | French wire service, state-adjacent funding but editorially independent | 4 | "Hugging Face, the French startup that became AI's warehouse"[8]. | Frames the story around national origin and European capital shortfall, and gives a French cabinet minister the interpretive quote. Accurate — the founders are French — but it reads as a loss for France in a way U.S. coverage does not, and the company is U.S.-headquartered with largely American investors, which the piece does note. |
| The Daily Caller | U.S. right | 5 | "Newly Bought Tech Startup Will Stay 'Open, Neutral' After $13,000,000,000 Takeover, New Owner Vows"[15]. | The headline is the acquirer's promise, in quotation marks, with the skeptical response absent. Writing the price as $13,000,000,000 in full digits is an attention device, not information. The verb "vows" flags it as a claim, but the claim is still the whole headline. |
| Bloomberg (Opinion) | U.S. center, market-oriented | 5 | "Nvidia's $13 Billion Hugging Face Deal Is a Bargain Hedge"[16]. | Labeled opinion, and it argues a verdict: the price is cheap for the insurance it buys Nvidia. That is a defensible read, but 'bargain' evaluates the deal from the acquirer's shareholder seat only, with no weight on the neutrality question. |
| CNN | U.S. center-left | 6 | "Nvidia inks $13 billion deal to buy the AI startup that was hacked by OpenAI"[4]. | The identifying fact chosen for Hugging Face is a security incident involving a competitor, not what the company does. That foregrounds a closed-AI-labs risk narrative and pushes the actual merger question down the page. The underlying detail is real and reported — Hugging Face says it used a Chinese open model to defend itself because closed-model licenses restricted that use — but as a headline it does editorial work. |
| The Register | UK tech press, adversarial toward large vendors | 7 | News side: "Nvidia buys Hugging Face for $12.9B, promises not to squeeze too hard." Analysis side: "Hugging Face is too important to fall into Nvidia's hands"[5][6]. | "Promises not to squeeze too hard" is a verdict embedded in a news headline. The analysis is openly argumentative and says so. It is also the outlet doing the most specific work — naming ROCm, the rival-usage-data problem, and the Arm precedent — so the spin comes with checkable substance attached. |
References
- NVIDIA to Acquire Hugging Face — NVIDIA · Primary source — the acquiring company's own announcement
- Hugging Face approached Nvidia's Huang weeks ahead of $12.9B acquisition, CEO tells CNBC — CNBC · U.S. business news, owned by Comcast/NBCUniversal; access-driven CEO interviews
- Nvidia confirms it will buy Hugging Face for $12.9 billion — TechCrunch · U.S. startup/tech trade press, owned by Regent LP
- Nvidia inks $13 billion deal to buy the AI startup that was hacked by OpenAI — CNN · U.S. center-left general news, owned by Warner Bros. Discovery
- Hugging Face is too important to fall into Nvidia's hands — The Register · UK enterprise-tech publication, openly skeptical of large vendors; this piece is analysis/commentary
- Nvidia buys Hugging Face for $12.9B, promises not to squeeze too hard — The Register · UK enterprise-tech publication, adversarial house style
- Nvidia to acquire Hugging Face for $12.9B to expand AI platform strategy — Fox Business · U.S. right-leaning business network, owned by Fox Corporation
- Hugging Face, the French startup that became AI's warehouse — Agence France-Presse · French wire service; state-subsidized but editorially independent
- Nvidia agrees to buy Hugging Face for $12.9 billion, report says — CNBC · U.S. business news; this piece relays reporting first published by The Information
- Why Nvidia's 'defensive move' to acquire Hugging Face is about much more than chips — CNBC · U.S. business news; analysis piece
- NVDA Stock On Track To Hit Over 2-Month High – Analyst Calls $12.9B Hugging Face Deal 'Strategically Valuable' — Stocktwits · U.S. retail-investor platform; relays sell-side analyst notes, which are issued by firms with banking relationships
- Nvidia stock climbs as Hugging Face acquisition deal announced — Traders Union · Commercial trading-education site; market-data reporting
- NVIDIA Insists Its $12.93 Billion Acquisition Of Hugging Face Will Escape Antitrust Scrutiny, Calling It A "Deconcentration Platform" — Wccftech · Hardware-enthusiast trade site; ad-supported, aggregation-heavy
- Nvidia's $12.9B Hugging Face Deal Must Pass Antitrust Review Its Quasi-Mergers Dodged — Tech Times · U.S. commercial tech news site; aggregation-heavy
- Newly Bought Tech Startup Will Stay 'Open, Neutral' After $13,000,000,000 Takeover, New Owner Vows — The Daily Caller · U.S. conservative news site co-founded by Tucker Carlson
- Nvidia's $13 Billion Hugging Face Deal Is a Bargain Hedge — Bloomberg · U.S. financial media owned by Michael Bloomberg; this item is signed opinion, not newsroom reporting
- China's AI firms roll out DeepSeek rivals in open-source drive — Xinhua · Chinese state news agency
- What's next for Chinese open-source AI — MIT Technology Review · U.S. tech magazine owned by MIT; institutional, generally pro-technology
- Nvidia closes in on Hugging Face acquisition — TechCrunch · U.S. startup/tech trade press
- Hugging Face goes from a 'scrappy' startup named after an emoji to $13 billion Nvidia acquisition — Fortune · U.S. business magazine, executive-audience framing