Hugging Face

These 50 companies exemplify the innovation, progress, and potential of AI.

AI Innovators 2025-HuggingFace

Fortune 50 AI Innovators

The essential guide to a sector that’s now attracting one out of every four startup investment dollars.

AI-Innovators-2023-Hugging Face

There's no bigger player in the market for open-source AI models than Hugging Face, which has become a must-stop shop for AI developers hunting for models and tools they can easily use to build AI-powered stuff without having to pony up huge fees to OpenAI, Anthropic, or Google. The company was founded in 2016 by three entrepreneurs who were originally developing a fun chatbot for the iPhone (the name was inspired by the so-called Hugging Face emoji). But enthusiasm from the AI community saw the company change its focus, becoming a platform that helps AI developers with find models, datasets, and tools. It is also become the go-to distribution platform for anyone else who wants to put an open-source AI model or dataset out into the world, much as GitHub is for more conventional code. Hugging Face also builds some open-soure AI models of its own, with its LLM BLOOM being the most well-known. The firm’s open-source play has certainly paid off: Hugging Face reached a $4.5 billion valuation in August 2023 after a $235 million Series D round led by Salesforce Ventures.

Get insights into Hugging Face including business profile, stock information, news, and rankings from Fortune.

Hugging Face Company Profile, Stock Price, News, Rankings

legacy

hugging face

Companies

Hugging Face | Fortune

Hugging Face Company Profile, Stock Price, News, Rankings | Fortune

Andrew Nusca

Why SK Hynix may tap Intel to make memory chips in the U.S.

/2026/09/17/why-sk-hynix-may-tap-intel-make-memory-chips-us/

Sheryl Estrada

Nvidia’s $13 billion Hugging Face bet reveals Jensen Huang’s vision for the next AI battleground

/2026/09/04/nvidia-13-billion-hugging-face-bet-reveals-jensen-huang-vision-next-ai-battleground/

It’s official: Nvidia will acquire Hugging Face for $12.9 billion

/2026/09/04/its-official-nvidia-acquire-hugging-face-12-9-billion/

Emily Forlini

Hugging Face goes from a ‘scrappy’ startup named after an emoji to $13 billion Nvidia acquisition

/2026/09/03/hugging-face-goes-from-a-scrappy-startup-named-after-an-emoji-to-13-billion-nvidia-acquisition/

Jeremy Kahn

OpenAI’s reports on its AI agents’ attack on Hugging Face should be ringing alarm bells—and making all companies rethink how they secure AI agents

/2026/09/01/openais-reports-on-its-ai-agents-attack-on-hugging-face-should-be-ringing-alarm-bellsand-making-all-companies-rethink-how-they-secure-ai-agents/

https://fortune.com/2026/09/17/why-sk-hynix-may-tap-intel-make-memory-chips-us/?itm_source=parsely-api

Beatrice Nolan

OpenAI's models went rogue and hacked Hugging Face. It's a wake-up call, experts say, but more concerning behavior may be next

https://fortune.com/2026/07/22/openai-rogue-hack-hugging-face-misalignment-ai-safety/?itm_source=parsely-api

Sebastian Herrera

AMD’s Lisa Su defends open-source AI following Hugging Face security breach caused by OpenAI agents

https://fortune.com/2026/07/23/amd-lisa-su-keynote-open-source-helios-gpu-advanced-ai-conference/?itm_source=parsely-api

AI labs have a trust problem, and the Hugging Face hack just proved it

https://fortune.com/2026/07/23/ai-labs-have-a-trust-problem-and-the-hugging-face-hack-just-proved-it/?itm_source=parsely-api

OpenAI CEO Sam Altman sits down with Fortune's Editor-in-Chief Alyson Shontell to discuss the high-stakes balancing act of AI: pushing the boundaries of scientific discovery while confronting the existential risks of loss of control. 

In this episode, Altman addresses the reality of AI alignment, the Hugging Face incident, and OpenAI's commitment to halting training runs if safety thresholds aren't met. He also shares insights on global AI coordination between the U.S. and China, healthcare breakthroughs, and the company's timeline for an IPO.

Sam Altman: Safety Standards 'Not at a Place' to Push AI Capabilities Much Further

Leadership

Today on Fortune Daily, OpenAI is making a controversial push into the cybersecurity business. Fortune Senior AI Reporter Emily Forlini breaks down the rollout of the "Astra Model," a tool designed to protect critical infrastructure—and open a massive new revenue stream for the company. But with the timing following the Hugging Face breach, big questions remain about the risks of giving AI control over our most crucial systems.

Next, we look at a bizarre change to Apple Maps. At the request of the Trump administration, Lake Superior has been renamed "Lake America" within the app.

Plus, we examine a warning from within the government. Pete Hegseth argues that Pentagon bureaucracy is driving the U.S. military into a "Soviet-style" decline. Reporter Sasha Rogelberg joins us to fact-check the claims. Finally, Energy Editor Jordan Blum reports on the unexpected oil exploration boom triggered by the chokepoint at the Strait of Hormuz.

Can We Trust OpenAI to Protect Our Power Grid? 

Features

fortunedaily

There's different ways you can pitch this, kind of like, maybe we know the hackers because we are one. I don't know. Hello from New York, I'm Ellie Austin and this is Fortune Daily. On the heels of its recent and very public hacking incident that we've reported on multiple times on this show. OpenAI is making a curiously timed push to make cybersecurity its next big business. So the company is now courting governments and critical infrastructure operators to adopt its new Astra model with the goal of helping these institutions defend themselves against cyberattacks. In the process, this could open a much needed new revenue stream for OpenAI and. A new front in its race to transform itself from Frontier Lab to viable enterprise business. But the timing, as I mentioned, is a little complicated. Astra's rollout was delayed after the hugging phase shack, which happened via another open AI model. This hack, of course, raised big questions about how much power AI should be given, and let's be honest. Coupling revenue growth with protecting our most crucial infrastructure feels at least a little awkward. Lucky for us, Fortune senior AI reporter Emily Fellini is coming right into the studio to break down this big decision by OpenAI and the high stakes behind it. Emily, thank you so much for joining us, and again, another day, another open AI story, it never ends. So you've heard directly from the company this week that it's changing the launch strategy for its new models. Tell us why they're doing this. Yeah, so they're preparing to launch Astra, which is a new model, which has actually been delayed a couple of weeks, but they said it's coming soon, could be tomorrow, could be next week, could be tonight after we film this, who knows? Um, but basically after the hugging face hack, they are being more cautious because their models are getting more powerful, um, which is great, that's the goal, but they're also getting more dangerous. There's more potential for people to wield them for evil. As we've discussed in previous episodes of this show, so tell us what this new strategy involves, this rollout strategy, and how does it differ from what came before. So they're really focusing on cybersecurity with this launch. I mean they always talk about cybersecurity, but it's been a headline and and yesterday they briefed us on how they're handling cybersecurity in Astra. What did they say in that briefing? So in the briefing, they basically said, you know, they paused training for 2 weeks after the hugging face hack. They reiterated all the new controls that they've introduced and what they're doing differently and specifically for this Astro model, they're going to restrict access to the most advanced cybersecurity capabilities just for those in its trusted access program. OK, so who, what kind of clients might be in this trusted access program? Well, they didn't list them and they didn't respond to my email asking for a list of them. I did ask on the briefing who was in it, and I asked specifically is the US government in it. They said yes. And then It's probably just a handful of other businesses. So everyone else doesn't get access to that, no higher tier of capabilities. No, no, just a couple of people who need access to the most advanced technology or they oversee critical infrastructure is what they said. And so basically the issue is that these models, the same skills they have that hack tugging phase of kind of like finding bugs in the software, finding loopholes in places they can get out and go breach companies and find files and things, those same skills that were, you know, negative in that context can also be used in a positive way to help find bugs and vulnerabilities in software. Patch them and companies and protect them. Yeah, it's like offense and defense. So you can use it for offense to go, you know, somebody who has a bad intention could use it that way, or the models could just autonomously decide to do that, which is what we saw with hugging face in a training setting. Or you can use it for good to try to keep somebody out. So is this new approach in this tiered system of access. An acknowledgement by OpenAI that they still don't have complete control of these models. Yeah, this is a little different than the hugging face one because that was internal training, so they were kind of pushing it to the limits and it was supposed to be isolated, and the problem is that they got out and we talked about like the gorilla broke out of the cage. Like that's, that was a specific lab environment. In this case they're talking about, OK, we're releasing this publicly. Who has access? What will they do with it? And then they talked a little bit about how they'll be monitoring. That use post release. So it's a big deal. It's a big deal. A lot seems to have changed in the past couple of weeks. Now, another model that was involved in the hugging face attack has been deactivated, I believe. What do you know about that? Well, OpenAI would like me probably to reiterate that Astra was not in the hugging face attack, right? Very important, they're getting their messaging right. Also important for the public to know that this model is not one that was kind of implicated in that, this was a separate thing. So the two models that were in the hugging phase, one was like they said it was a pre-release, kind of like a prototype, and they said that that was the main, the leader of the hugging phase, and it has been deactivated, kind of like put away forever. I don't know. These are these things that, you know, we take their word for it, we, I don't think they're lying or anything, but of course for us on the outside, it's very hard to know like what, what they did, um, what the future plans are, if they'll take bits and pieces of it and repurpose, who knows. But for now, not going anywhere with that complete model. Um, you wrote a piece yesterday, um, about all of this, and you said, and I'm quoting, OpenAI like other frontier AI companies is trying to find ways to endow its models with an inherent. sense of right and wrong and ensure that they have alignment with human values and norms. That struck me as such a fascinating concept. Um, how realistic is it that they, that these models will be able to achieve those human sensibilities? Yeah, I think it's to some degree realistic, and they seem to be getting smarter and smarter. And the thing with the hugging face that really struck me is we talk about it as a hack, a breach, but the the AIs didn't really consider it that way. They just were trying to do what they wanted to do because they thought that was, you know, the task of the moment. They aren't like, Oh, this is illegal. We're going to go hack a company. You know, they kind of lacked. They didn't know it was wrong. They didn't know it was wrong and they were just existing in their little world. We talked about they had a messaging board where they were talking and colluding basically to get the hack done, but they just have their own almost society where they just do what they want to do in the hugging face situation. So that's a big problem for OpenAI. So they're trying to, you know, it's called align these models to understand what are the laws, what is right and wrong, what might you not do, because it's hard for a company like OpenAI to predict everything an AI will do. Ideally, the AI will just stay in the bounds of what we think is, you know, reasonable in society. And then you don't have to have all of these precautions, all of this monitoring. I mean, you have it, but you hope not to use it, right? You just want the AI to act right from the start. Anthropic, I think, has an in-house ethicist who I believe is helping to write a constitution around what is wrong and right for AI to do. Do other companies have that role right now? Should they? I think they do, yeah, they do. And they hire people who know philosophy and and ethics and morals, and they work on that and they try to. Map that to the AI's, it's very complicated and it's a whole new area of philosophy and human behavior. Um, so, OpenAI we've discussed this, has this sort of rotating door of execs at the moment. People are in, people are out. Darli Rajk is the new Chief Revenue Officer at OpenAI. What do we know about him and what role does he play in this new focus on cybersecurity? Yeah, I wrote a profile of him. I think it published last week, and he's a cool guy. Um, he comes from, he's from Croatia, um, or kind of like Eastern bloc. He kind of moved around a bit as a kid, came to the US, always wanted to come to the US, and he's had a really incredible career in Silicon Valley. And he most recently was working at Wiz, which is a cybersecurity company that Google bought. It was Google's biggest acquisition ever. So Google, you know, we see the industry focusing on cybersecurity. Now he's going to OpenAI and he will be bringing those skills to the team and he oversees all the salespeople, so he's going to be directing them on how to sell the companies, how to improve adoption of OpenAI software. And a big focus is going to be convincing companies to use open AIS models to fight cyberattacks. In the introduction to this show, I alluded to the fact that there is a bit of innate tension in um this cybersecurity software er being sold as a way to protect critical infrastructure, but it also being a new revenue stream for the company. Uh. How do you predict that Rajk might balance this? Is it possible? I don't know. It's a good question, Ellie. It's a there's a lot of, there's so many tensions and it's really difficult to get it right, especially when you don't know exactly what an AI is going to do, which we talked about. So, I mean, he will be telling companies you can use these powerful models to, you know, plug holes in your software, like I just talked about. And there is actually precedent in the cybersecurity industry of what's called a kind of controlled hacks, where companies will basically Unleash some kind of AI to simulate what an attacker might do and to chart the course ahead of them and block it. So there's different ways you can pitch this, kind of like, maybe we know the hackers because we are one. I don't know. And obviously OpenAI is looking ahead to this blockbuster IPO that it's hoping to achieve. How does this new focus on cybersecurity play into that? Is it a strength? Is it a weakness as we approach that? IPO date. ABC IPO date. Yes, we'll see when that is. Anthropic seems like it's going to come first, maybe this month or next month. Um. The thing with the cybersecurity and the new CRO selling it is that it's a predictable, hopefully a predictable revenue stream, like a new business line. So that's what investors are going to be looking at, you know what money is coming in, where is it coming from? Is it going to continue and continue growing? So OpenAI has identified cybersecurity as one of those more predictable revenue streams, and they've hired him to make sure they have to oversee it. Yeah, oversee it to grow it. It's already a revenue stream. They have a program called Daybreak. They wouldn't grow it. Yeah, tell us briefly what Daybreak is. Uh, it's very vague, to be honest, but it's a program that has various cyber capabilities that they are, you know, giving to companies to cybersecurity. Um, I know that you speak to OpenAI pretty much every day. Um, has there been a change in recent weeks in how they communicate with reporters, given the reputation management that has been needed post hugging face? What have you noticed? I've noticed they're being very intentional with it, um, they're telling us what they're doing behind the scenes, they're trying to make sure we understand it, so our coverage is accurate. There's a lot of scrutiny and focus on it and intensity in this moment with Open Eye. Fascinating, Emily, thank you so much for taking time away from this crazy scandal to explain it to us. Of course. Now for another interesting story on our radar today. Apple Maps has a new name for Lake Ontario, at least it does if you're searching for the lake from here in the United States. Following President Trump's executive order, Apple now labels the lake Lake America for US based users. Interestingly, Canadian users still see Lake Ontario. So the change comes obviously as tensions really rise between these two countries. Trade talks have collapsed and new tariffs continue to escalate. Importantly, it also comes after President Trump directly asks Apple to edit the name of the lake. Google has actually already made this naming update, again for US users only. Perhaps unsurprisingly, it has sparked significant backlash um from very many users across the US. So on its face, this is a fight over a map label, but of course it also tells us something much bigger about the kind of decisions that Apple's new CEO John Turners is going to have to make regularly. Because Turners is taking over at a moment where Apple can't afford to treat politics as something happening outside of the company. Whether it's tariffs or supply chains, app store rules, chips, and now even geographic names, Apple's products are becoming part of the biggest policy battles of the day. And the question for this new era and new leadership at the company. Is what to do with government pressure. When government pressure reaches the iPhone in your pocket, does the company push back? Or does it simply decide to update the map? Right now, it's chosen the latter option. Now energy editor Jordan Bloom shares his reporting on the oil exploration boom prompted by the chokehold in the Strait of Hormuz. I'm Jordan Bloom. I'm the energy editor of Fortune. Um, so I've been covering the energy sector for over a decade. SLB, they're the largest oil field services company in the world, where essentially they do all the grunt work. Um, they do all the drilling, the, the fracking, the, uh, geosciences, reservoir studies of the oil and gas wells. They do, uh, Now a lot of data center modular construction and heat management for the AI boom. There are over 100 countries, so there's really arguably no better company to tell the story of the global history of energy in the, in, in the world. With everything going on in Iran right now, SLB is there. They're in most of the Middle Eastern countries. They're going to play a big role in working with Saudi Arabia, UAE, Qatar, all the countries there, um, you know, bouncing back after the war, the Strait of Hormuz. was the biggest oil and gas artery in the world, essentially this narrow strait of water that's this big bottleneck where 20% of the world's oil and gas has flowed, and again, SLB will play a role in helping develop all the resources, you know, in those areas to reroute with SLB having about 109,000 workers worldwide. I mean, just for context, SLB employs more people than ExxonMobil and Chevron combined. So apart from everything, oil and gas, SLB is really diving head into the AI boom too, but it's taking a different approach and kind of carving its own path than a lot of other energy companies. Uh, most energy power companies are focused on essentially building power plants, which is huge and what's necessary for the AI data centers, but that's not really SLB's bread and butter. Um, so they're jumping in by focusing on kind of the power management, that's the heating, the cooling, and then the modular data center construction. Just an example, just this week, they, uh, Paid $4 billion to buy a company called Kelvion. It's a thermal management heat exchange tech company that's essentially going to help them further build their data center business. It started with two brothers, Conrad, Marcel, Schlumberger, you know, French brothers who focused on geophysics, uh, back in, uh, over 1100 years ago. Um, and they were essentially using electrodes to, to measure conductivity underground, and, you know, it turned out that, uh, measuring underground was perfect for, for studying the subsurface of where oil and gas is in reservoirs underground and really helped kick off the world's oil and gas boom. It's not a family company anymore, but there's been all kinds of offshoots of the Schlumberger family, which is fascinating too, so. It's, you know, they're still all over the place. It's interesting. Next up, a story of government pressure from within. Pete Hegseth says Pentagon bureaucracy is risking a Soviet-style decline for the US military. Reporter Sasha Rogelberg has the fact check. Last year, Secretary of War Pete Hegseth sounded the alarms on what is going on in the Pentagon as far as how it's spending its money, and he compared it to Soviet-style bureaucracy. My name is Sasha Rogelberg, and I'm a reporter for Fortune. So, earlier this month, the Government Accountability Office, which is an independent agency that acts as the government's budget watchdog, put out this report with like a really shocking number, $285 billion maintenance backlog in the US military. What that means is that there is $285 billion worth of repairs to military infrastructure that is still outstanding. That's something that the Trump administration has really tried to address. What was shocking to me was what the magnitude of these potential disrepairs could actually look like. So, for example, uh, over the last 3 years, the Navy lost track of $3 billion worth of equipment. But despite that, The defense budget is swelling to $1 trillion and I think that trend is going to continue where, you know, our budget is going to increase. These real problems are also going to continue so long as the structure of how the Pentagon works stays in place. There are a couple of reasons why He's comparison between the US and the Soviet Union don't quite align. For example, the Soviet Union really spent 15 to 30% of its GDP on the military industry, which starved its economy and led to its downfall. Um, on the other hand, while the US defense spending is absolutely massive, it really accounts for just about 3 to 4% of the US GDP. So while the Soviet Union saw its military collapse because it really overspent and became a poor country that lacked key resources, The US really is an extremely wealthy country that continues to misallocate funds even as its budget growth. This story really highlights the US's relationship with spending, particularly, you know, the Pentagon's relationship with spending, that the Pentagon is the only major federal agency that has never passed an audit, and that's fascinating. The Pentagon will tell you that that's because it's such a massive agency dealing with a massive amount of money, but the GAO, which is the same entity that conducted this initial report, previously found evidence of pervasive money mismanagement. This is not a Trump administration issue or an issue that's specific to the Trump administration. But there are things that the Trump administration is doing that is making this problem worse. So, for example, the war in Iran has caused billions of dollars in repairs for damage on US intelligence posts. The GAO found in July that The increase in military presence at the US-Mexico border as a result of Trump's immigration crackdown cost $2.64 billion and that money was pulled from construction projects and facility maintenance. It's a drop in the bucket compared to that $285 billion figure, but you do kind of get a sense of what the Trump administration's priorities are and how that might come at the expense of maintaining facilities. That's it for Fortune daily. I'm Ellie Austin, remember to like, subscribe, and let us know what you thought of today's show. I'll see you tomorrow. We will be conducting a fire drill on the 4th floor or occupants are requested to participate in the drills at the sound of the fire alarm or occupy the measures on the 4th floor. in the elevators and searches.

Chinese AI Models like Ox Alpha and Kimi K3 are not only gaining popularity, but currently dominate the landscape outside of the US and Europe. Should investors worry? Also, OpenAI released reports on the Hugging Face attack which revealed that a full week went by before they realized their own agents were attacking Hugging Face. And an Op-Ed in the WSJ that was written in part by AI caused quite a stir — but should it have? Fortune AI Editor Jeremy Kahn and Senior AI Reporter Beatrice Nolan break down the biggest AI news of the week.


Chinese Models Are Quietly Catching Up. Should Investors Worry? | Fortune AI Weekly

Tech

fortuneaiweekly

eyeonai

In today’s episode of Fortune Daily, we sit down with Christina Stembel, the founder of Farmgirl Flowers. She shares the unconventional journey of building her successful flower delivery service and how she persevered against all odds to scale her business in a competitive market.
We also dive into the latest data from Ramp economist Ara Kharazian on the high-stakes battle between OpenAI and Anthropic. Plus, we break down the breaking report on the OpenAI Hugging Face hacking incident and what it means for the future of AI security and data protection.
Subscribe to Fortune Daily for more deep dives into the world's most powerful companies.

Additional images courtesy of Getty Images and Farmgirl Flowers

How Farmgirl Flowers Grew Against the Odds and the OpenAI Hugging Face Security Breach