Why Nvidia Just Paid $12.9 Billion for Hugging Face | The Business of AI

0:01This episode is brought to you by Google Chrome. You think you know a browser, but Gemini and Chrome? That's new. It can help you with practically anything on the web, like restoring a vintage motorcycle from a 50-page restoration block, or finally break down that long article you've had open for weeks. Gemini and Chrome is here for it. Ready to make anything online make sense? There's no place like Chrome. Check responses set up required, compatibility and availability varies 18+. Hello and welcome back to the Market Maker podcast and on this week's episode we are talking all things Hugging Face which last week was acquired by NVIDIA for 12.9 billion dollars or to be precise 12.93.3 billion dollars which we're going to go back and discuss in a little more time might be a little easter egg in there for you but we're going to go deep into what Hugging Face does because I think if you're in the tech world, you're probably fully familiar.

0:57But if not, which is probably the majority of us, we're going to find out what exactly that firm does. Why NVIDIA is willing to pay a 86 times multiple for the company and who the winners and losers might be. And Stephen, I was having a look at this acquisition. I was just saying to you offline, Hugging Face, very much known in the tech world for the smiley emoji with the company's logo. And I did see the press release NVIDIA put out with NVIDIA's logo with a heart emoji, a hugging face emoji. I was like, have I woken up in some sort of parallel universe or is this seriously a 12, 13 billion dollar deal with emojis involved?

1:40What is going on here? Yeah, yeah. And the other thing that's probably worth noting is the NVIDIA, the name NVIDIA is based off of the Latin for envy, invidia. And that is one of the seven deadly sins. Not only is it one of the seven deadly sins, but I believe, according to Thomas Aquinas, and someone can correct me on this, I believe it is the second worst of the seven deadly sins. Maybe even the worst. I need to kind of check my medieval literature. So a company based off the second worst deadly sin is acquiring Hugging Face, which has got a nice smiley emoji. If that doesn't say it all, I mean, I might as well just stop this episode.

2:24I've said what I need to say. Well, with this, I saw there was lots of Bloomberg sources and there's a lot of tech followed websites, right, where they get the scoops and the downline on these deals probably before they become in public domain. So how did this one come about? What were the initial headlines? Yeah, absolutely. So last week, the information, the website announced that NVIDIA agreed to buy the open model repository platform Hugging Face for$12.9 billion on$150 million of annualized revenue. So that's the 86 times revenue multiple that you mentioned. The business is close to profitability, but it's not yet profitable.

3:10This comes, this announcement, and we'll talk about the love-in between Jensen Huang and Clem DeLong, the CEO of Hugging Face, in a second. This comes a few days after Hugging Face reported via Business Insider that it's working with a bank to evaluate bidders, plural, interest. So NVIDIA is not the only game in town. But in the grand scheme of things, this isn't NVIDIA's first bid. So they first invested in Hugging Face in August 2023, Series D, raise of$235 million at a$4.5 billion valuation. So by the way, just to put that in context with the conversation that we had last week with Stripe and OpenRouter, August 2023, three years ago, Hugging Face had a valuation of $4.5 billion.

4:00It has just been acquired for close to$13 billion. So that's still circa 3x return over three years, which is great, but it's not quite as parabolic as... I don't even think OpenRouter existed in August 2023. So this feels a little bit like normal VC. I love that. Normal VC. It's just what a crazy, crazy world we're living in. at what point you look at some of these and go okay yeah this is getting a little bit toppy what what would be flags or signals when you look at yeah the open router the cursors to this is it do you feel like i guess we're going to dive into these numbers and look at how much this makes sense but what are those flags just top level i think the flags are if you require and and and the flags tend to come from case history of the dot-com bubble and then bust.

4:59If you are buying a company that doesn't have a robust business model or route to profitability and pay a lot of money for it, then that's a warning sign. If you buy a company that may have slightly more credibility or a route to profitability, but the acquirer is in no way a tech company or able to integrate or benefit or get synergies from that acquisition, then it's just seen as a kind of adjacent hype acquisition. And then thirdly, if you're acquiring a company that is considered to be what we would call in the world of M &A, a transformational acquisition, i.e. you're putting your company on the line because it's such a large amount, you're having to raise debt, you're having to issue new shares, it becomes a massive burden.

5:57Those three signals are what would concern me. Those three signals are absolutely not the case for this acquisition. We'll talk about it in a few minutes, but NVIDIA chucked off$100 billion of free cash flow last year. So this is 13 % of their annual free cash flow. So if this all goes to zero it doesn't matter it really doesn't well what just with nvidia and strategy i'm not sure we'll get to this already but the with that amount of money is that is that what you would anticipate that they're not why are they not more proactive in doing large-scale transactions is that because there's not the specific targets that they they require or it seems because this is their biggest one on record right isn't that isn't that right with nvidia yeah it's it's it's all about so if i'm a shareholder of nvidia what do what do i want and remember from a kind of corporate finance 101 perspective you get profit at the end of the year net income that is effectively

7:13the shareholders. But it is at the behest of the company to choose what to do with that net income. They can keep it as cash for a rainy day, they can distribute it as dividends or share repurchases, or they can invest it in the form of capital expenditure or acquisitions or joint ventures or whatever it might be. So NVIDIA, the board, the CFO, Jensen Huang, that$100 billion of free cash flow, they're going to be sitting down and thinking to themselves, all right, what combination of things can we do with that$100 billion, sorry, what combination of things can we do to maximize shareholder value?

7:56And that will be, and has been historically, a reinvestment. B, dividends and share repurchase, which is just coming online at the moment. And then C, building out the AI ecosystem through things like guarantees to open AI and revenue guarantees for data centers and things like that. therefore there hasn't necessarily been the need or the attractiveness of potential ai targets in this space and timing wise is there any like kind of cues there that we could take out of this because normally any M &A transaction has a perfect moment to strike. Yeah, so it's quite interesting because NVIDIA has been making overtures to Hugging Face for a while.

8:56And late in November 2025, December 2025, NVIDIA offered a$500 million investment at a $7 billion valuation, which was rejected by Hugging Face as, and I quote, it does not want a single dominant investor that could sway decisions. This is eight months ago. Think about moral integrity here. Get the checkbook out, Stephen. Get that checkbook out. And I'm going back and I'm going to reference this bloke Clem DeLong and hopefully there'll be some comments. Maybe I'm a little bit too cynical, but he said back in 2024, I said it and I will say it again, concentration of power is the biggest risk in AI.

9:36So NVIDIA buying Hugging Face eight months after the investment was rejected because it doesn't want a single dominant investor suggests that there was a moment or there has been a moment over the last month or so where other potential buyers have come to the table. And remember, and we'll talk about Hugging face what it does in a minute in a minute but remember hugging face is a is a community for open source uh ai models you know chats free you know a kind of github if anyone remembers or knows github a github for ai so it has this great kind of democratic democratizing open source open weight thesis.

10:28But it also, by the way, it also has for-profit venture capital investors. So when the likes of maybe Salesforce or when the likes of NVIDIA start sniffing around and saying, we want to take you guys off the market, we want to buy you guys flat out, early investors like Sequoia, they're not going to be like, yeah, but we care more about the community. We care more about the mission. We care more about the thesis. What they care about is the fact that there's $13 billion on the table. And this now sounds like, you know, seems like the right time to strike. I guess in a way, if you're very confident in the trajectory of your business, isn't this just good gamesmanship to get to the point of getting a$13 billion valuation in the end?

11:16This is just a tactics that you play. You don't want to bend over too quickly you want to play hard to get isn't this just as the normal tango that companies do as they go through these these relationships yes i i totally agree i think it's the it's the mission statement and the mission quote-unquote alignment of the hugging face team and you know concentration of power and saying all of these things the smiley emoji face the community aspect all of that stuff and then to sell to nvidia now we'll get on to it nvidia may not be the worst acquirer in terms of concentration of power but again latin for envy five trillion dollar company a hundred billion dollars of free cash flow there's a little bit of concentration of power there maybe yeah so what's latin for greed then yeah yeah yeah you're testing me there yeah so okay so this this transaction was not 12.9 billion.

12:18And so I don't know, as a layman like me, and I read a number like 12.93 billion, I'm like, okay, so what is this out of a model they've just created that something's worth the extra little 30 million, 300 ,000 within the billion figures. So how on earth did that come about as a such a specific figure? yeah this is where like i don't want to disparage the tech community too much but this is where tech nerds that are about to become billionaires really start to irk me so the deal the value of this deal the acquisition price tag 12.9303 billion. So one, two, nine, three, zero, zero, zero, zero, zero, zero, lots of zeros, but a very confusing acquisition price.

13:12This was an Easter egg that a load of, you know, that was dangled out there by the CEO and co-founder Thomas Wolfe saying, look, here's an Easter egg, try and figure it out. So the hugging face reference was easy, supposedly the number 129303 is the decimal representation of unicode point u plus 1 f 917 the emoji officially named hugging face and then the color code 129303 is the vibrant shade of green that looks a whole lot like nvidia's classic color and then so that was why that was the number and clem clem de long on on twitter on x said had to do it for the lols that was what he said had to add an extra you know 30.3 million dollars onto the end of a 12.9 billion dollar transaction for the lols and for the lol yeah i mean you are just love these guys you you are just bossing life if you could be selling your firm for 13 billion dollars and just doing and making statements like that i mean you are the final boss basically yeah again they're not happy none of these guys are happy i was just having a quick look actually thomas you said thomas wolf the co-founder and chief science officer of hugging face i just had a quick look i'm always interested to know like what's the background of these people you know when we look at all of the you know situational awareness or whatever it might be these recent kind of topics so this guy thomas wharf so eccole polytechnique and serbonne university they're like the top two target sort of quant schools if you like math schools in france really highly regarded nearly every quant hedge fund will have uh people on the ground you know doing that whole uh networking to try and pick up talent he then got a law degree actually and then worked as a physics researcher and then a patent attorney for six years so yeah quite quite an interesting background yeah and again this is part of the story so these founders are all french um and this this if you are generous enough to france this could be considered to be a french success story or even a french missed opportunity because they've sold to a big American company.

15:51But this isn't really a French company. It's registered in the US. It's got French founders, but all of the main investors, as we've discussed, are American investors. It's based there. This is an American company to all intents and purposes. Well, I know we'll probably get into that a bit more, but you just mentioned there the investors. Perhaps we could just touch a little bit on that. So who's been involved to this point? Is it just the same names as ever or is there some more niche ones in the mix? Yeah, so there's some typical San Francisco names, the likes of Lux, Capital, Sequoia, and we'll reference those guys later on in the episode and coach you as well.

16:31Strategics like Google, Salesforce, Amazon, Intel, AMD all piled in a couple of years ago. and then some slightly strange individual investors. NBA player Kevin Durant. Now, Ant. I'm going to keep you interested in this episode by maybe laying an Easter egg or whatever they call them. I'm probably using that in the wrong term. I'm waiting till the end of the episode to go back to the world of NBA and Kevin Durant. So anyway, he invested at the seed stage. And also... This episode is brought to you by Google Chrome. You think you know a browser, But Gemini and Chrome, that's new. It can help you with practically anything on the web, like restoring a vintage motorcycle from a 50-page restoration block, or finally break down that long article you've had open for weeks.

17:16Gemini and Chrome is here for it. Ready to make anything online make sense? There's no place like Chrome. Check responses set up required, compatibility and availability varies 18+. So OpenAI co-founder Greg Brockman. There's a lot of things to say about Greg Brockman, but as of two hours ago we were recording this on a friday he has just declared with the launch of chat gpt 6 astra the start of the agi artificial general intelligence era so there you go we're in it well it's just got to do i mean someone's got to do something because you know i don't know how i know you're more of a clawed kind of guy uh but i still use chat gpt for certain things and it went down for half an hour yesterday interesting completely down and you know where you go to if you're like is this my internet connection is this just me where do you go who you're gonna call elon on x you go on x and then you can see just millions of people going what's going on like freaking out it was down for hot like for me it's like half an hour so so you were mid you were mid therapy session with your mate chat gpt and i was with my girlfriend my second wife having a chat and we were like yeah yeah having a nice coffee oh poor you yeah no but i mean relaxing afternoon i mean i it just goes from bad to worse it feels for open ai that's interesting you said they've just done that new launch maybe it was the transition of the new model coming online it just something went wrong and but not a good look obviously so we might as well drop the agi statement at the same time maybe cover up the cracks but also the founders they obviously didn't invest in the company also the founders uh each pocketed estimated according to bloomberg 1.8 billion dollars each again just reference that number reference i'm so interested in this human story so so hang about they've they've massively outdone the open router guys then yeah bigger yachts bigger houses so they They're on the street that's slightly closer to the beachfront than the open-rooted guys, is what we're saying.

19:30Yeah, absolutely. Okay, well, you've kind of given me a bit of a clue, and I kind of get it, top level of what Hugging Face is. But perhaps just a little bit more of a, just dial it down a little bit, so we can all get our heads around what makes this such a unique platform. yeah so hugging face for the lay person like me and like you and is the ai community building the future i took that directly off their website with a little hugging face emoji um and when you want good analogies go to a website like pcmag.com you know they explain it like the lay person so pcmag.com describes hugging face as the neutral public square in the biggest boomtown of modern technology.

20:19So it is where people come together to collaborate, to build models, to share models, to share open source libraries, data sets, diffusers, transformers, all of these things that I don't really understand. And then PCMag goes further, just in case we didn't understand it. If Hugging Face is like the App Store, think about that analogy, it's because NVIDIA is the iPhone in this analogy. building out the hardware that made the AI boom possible. So Hugging Face is like the front end software layer for this open source or for this kind of non, I wouldn't say non-enterprise level, but non-closed weight, open AI, clawed type offering.

21:11And then finally, they said that hugging face is the GitHub of AI, which is definitely my preferred analogy. So it's got the hub where researchers and companies publish and discover open weight models. Just as a reminder, we spoke about this a few weeks ago, open weight models are models where the trained parameters or the weights are released publicly so that anyone can download, train and modify, as opposed to the closed-weight models of GPT and Claude. So once you have those weights, you can run the model locally. Importantly for NVIDIA, they don't care. They don't really care whether you use an open weight model or a closed-weight model, as long as you're using NVIDIA chips to power those models.

21:59So this, in terms of kind of basic strategy 101, this is NVIDIA owning a distribution pathway, right? So they've got the relationships, the embedded relationships with the open AIs of this world, with the claws of this world, with the big closed-weight models. It wants to open and own the interface between its chips and the open-weight models that are being created, shared, collaborated upon on Hugging Face, which has been that forum, over 300 million models. Or 3 million models, sorry. is there ever like a regulatory layer to this though i mean when you explain what you've just said to me it makes complete strategic sense from nvidia but yeah is there any kind of regulatory concern that would come about by them just acquiring almost like the vertical yeah it's really hard when it comes to antitrust and regulatory issues relating to mergers and acquisitions, what tends to happen is if it's considered to be a vertical integration, so a vertically integrated company owns their entire value creation chain, right?

23:25So the best example is always oil and gas, big oil and gas majors, they own the exploration and production, they own the transportation they own the refining they own the selling and the trading so if you are a company like nvidia and you buy an element of your value creation chain either downstream or upstream and hugging um and this one is downstream as opposed to upstream it's more close it's closer to the end consumer or the end buyer then these tend to face less regulatory concern or constraints because it's much more about market share. And this is not necessarily increasing market share, which is what the antitrust guys get quite concerned about, because you are buying a company in a different part of that market vertical.

24:22So to me and you, this does seem a little bit like, hey, what the heck is going on? NVIDIA powers, it's the picks and shovels of the world of AI and it's now owning a key gateway to massive volumes of consumers and purchasers and experimenters of AI. But will it get blocked from a regulatory perspective? No, because GitHub didn't get blocked and this is a kind of regulatory environment than it was back in 2016. Sorry, you've got my brain thinking now. So with hedge funds, a recent shift due to regulatory change in the market space was instead of them paying sell-side firms for research they would build out their own internal research function so to save costs and have more productivity internally for their investment decision processes and so on at what point does nvidia go hang about we're paying morgan stanley or coleman sacks jp morgan whomever hundreds and hundreds of millers of fees to do lots of different things at what point do you kind of build out your own almost mna crack squad which are just canvassing the market and modeling and doing everything all of the time where of course on the probably sharp end of the transaction you need the bankers involved but yeah i mean like you just kind of bring it in house some of that talent because you need it it's almost like a perma acquisition scene when you've got that type of capital to deploy?

25:56Yeah, so there's probably two answers to that or two responses to that. One, the big acquirers or companies that have an acquisition strategy or even any very, very large company that duds acquisitions will have their in-house corporate finance team and will do quite a lot of the work that the investment bankers will do. And maybe they'll be able to kind of pony up with the investment banks and get a slightly cheaper fee. the investment banks are still very much in the loop. But my second thing that that sparked was, do you remember General Electric? Obviously, you remember General Electric. But General Electric is a brilliant case study in moving from being a manufacturing and engineering powerhouse, a little bit like what we were talking about with Boeing a few weeks ago, to being basically a financial institution they moved under there's a great book called i was just looking it up just now they move um a great book called lights out pride delusion and the fall of general electric and this is under the ceo-ship of jack welsh who basically turned the company into a into a kind of financial engineering company um using its strong balance sheet using its credit rating to go out and basically become an investment bank or a quasi-investment bank.

27:24So it didn't end well, by the way, and I wouldn't recommend that NVIDIA does that. But yeah, there are examples in the past where a very, very large company with great free cash flow and a very good credit rating goes, wait a second, we can be the bank. We don't have to rely on the banks. We can be the financier. We can be the bank. you're just dropping easter eggs left right and center here because that's the connection to the Boeing episode where Jack Welch's strategy that profit management style of one pursuit is exactly the downfall of the the Boeing quality scenario we were talking about yeah I would say that we had all prepared this in advance but we definitely haven't all right well look I know we touched on briefly but maybe just a little bit more because the chap I mentioned Mr.

28:22Wolf was just a co-founder so who are the other people involved in this business and I don't think we've even touched on it yet how old is this business and how did it come to fruition in its first instance yeah so Hugging Face was founded in July 2016 by three co-founders Clement DeLong julien chamond and thomas wolf uh it was originally founded as an ai chatbot and again if you read the backstories as you as you alluded to with thomas wolf these three guys are very very impressive um i think julien chamond had an msc in computer science from stanford thomas wolf as you mentioned quantum physics phd etc so the ai chatbot idea launched in 2016 didn't really work didn't really click and then by 2019 they pivoted out and they the big hit came when google and i remember this from my startup when i was working for uh working as founder of of util my machine learning startup when google first released burt so burt is bi-directional encoder representations from transformers model um on its tensorflow platform and any techies will remember that very, very well.

29:42Hugging Face didn't really like the idea, thinking that a powerful model like BERT should be more widely accessible. So they recreated BERT, but using PyTorch, which I remember again, very, very well, and released it on GitHub, which achieved more than 5 ,000 stars, likes, in under three months. And so this is how the initial idea for the current Hugging Face emerged. They asked themselves, what if sharing LLMs would be as easy as sharing code on GitHub? So this is this kind of community, open source, these things, these powerful models shouldn't all be centralized by the likes of Google. And then what we now know as Hugging Face Today kind of came online in late 2020 and expanded from the proliferation of large language models and the breakthroughs in technologies that we've seen over the last five or six years.

30:45So the question in my mind then, if this is so contingent on community and being open source, how do you arrive at the$13 billion figure? Like, is there, how does this business make money? I mean, normally you're looking at the balance sheet, looking at all the different metrics to calculate some kind of value proposition. How does that work here if it's all open source? Yeah, absolutely. So they have the classic freemium subscription model. So a lot of what you can get or a lot of the initial bits that you can get a little bit like Spotify freemium model. A lot of what you can get is free to start out with.

31:26and then the more powerful models or the more complex things that you can do you have to pay a monthly fee which is exactly like github monetized as well and by the way we'll speak about github in a second that business is now generating well over a billion dollars a year so it's been a kind of relative success within the framework of microsoft but it also does things like Rent and Compute, not necessarily like OpenRouter, more maybe like a CoreWeave, which has an adjacency again for NVIDIA there. And then they have inference endpoints, essentially managing the process of actually deploying the machine learning model and keeping it alive so that users can utilize that model that's been uploaded and put out there to the community.

32:17So all of that comes together in a$150 million a year revenue amount, which is growing, not at the light speed that we've seen with previous startups, but it's getting there. And it's, again, it's this concept of ecosystem control that is why NVIDIA has paid$13 billion. There's obviously the fact that NVIDIA probably doesn't want other players to be the owner of this gateway to a massive, massive total addressable market of users that want open weight models as opposed to closed weight models. So it's like, I've got to get my hands on this. And quite frankly, whether it's$13 billion or$8 billion or$20 billion, for the founders, obviously life-changing money, for NVIDIA, meh.

33:13Oh, man. Yeah, not Jensen Wang's territory quite yet. But you mentioned there Microsoft and GitHub. Am I right? Microsoft didn't create GitHub. They bought GitHub. So is this a similar sort of scenario? Is there parallels there in the ratio of why Microsoft did that and why NVIDIA are doing this? It's super interesting. So in 2018, June 2018, Microsoft announced the acquisition of GitHub for$7.5 billion. And again, that was when I was in my machine learning phase. I remember this being a massive furore you know so this open this collaborative great community of code sharing code collaboration uh repository that basically everyone used was being bought by this kind of stodgy bureaucratic monopolizer in microsoft and i remember everyone going oh this is an absolutely terrible thing.

34:17But just looking retrospectively, and I haven't done a great deal of work on this, but it seems like this acquisition has been successful, not only from a Microsoft perspective, I mentioned that it now generates over a billion dollars in ARR, and it also is the it is the go-to point or it is the host for GitHub Copilot, which is the equivalent of the kind of clawed code or the codex for Microsoft. But also it seems to have been a kind of net positive for the users as well. So I've just kind of looked on the Reddit forums and I can see a couple of negative comments, but you know, it feels like, I'm just going to quote a couple the github acquisition hasn't really negatively affected me and some of the additions have been downright useful uh i'm not saying they're perfect but they're actually providing me with lots more a lot more products and a lot more optionality so github's kind of remained independent maintained its use case and its value for the developer community so if you can do something like that where you've got nvidia as the owner but hugging face remains useful and is just backed by availability and access to nvidia chips for model training and for running models as a kind of order of priority because they now are owned by nvidia maybe you know read the reddit comments and you wouldn't agree you wouldn't you wouldn't believe that this could be a good thing but maybe yeah maybe there's some sense here yeah that makes makes great deal of sense and what what about the the users of this i'm assuming there's a lot of like github b2c users i'm assuming that nvidia is what's the concentration do we know of nvidia's revenue that's b2b is it all of it or but in a sense of can they start to open up a new revenue stream here to that community audience yeah and this is where the kind of b2b and b2c distinction gets a little bit fuzzy so nvidia again i've always thought all right who buys from nvidia Okay, well, it's OpenAI and it's the big labs that are spending billions and billions of dollars on these frontier models, training them and then running them.

37:01Tens and tens of billions of dollars. And there's obviously a B2B concentration risk. If your main clients are buying billions of dollars worth of these chips, you're going to want to move a little bit away from that. so you've got the big labs that are buying the chips obviously you've got um you know the large corporates you know and the large tech companies that are buying billions of dollars worth of nvidia chips um but then you've also got the i wouldn't call it the b2c community but certainly the more hacker community the more um you know we're going to collaborate over a model and we're going to use it on a smaller use case that is not maybe enterprise level.

37:49It might be a business purpose. It might still be B2B, but it's more of a individual contributor type relationship than it is, all right, I've got$30 billion a year relationship with open AI. So it is just another and quite quickly growing market to sell NVIDIA chips. And we know, having covered open weight models a little bit on the podcast before, we know that open weight models are becoming more popular because they're more customizable, you can host them locally, and they're potentially cheaper. So Huang, you know, he's very good at reading the tea leaves and sensing where the direction of travel is.

38:37And he sees it very clearly, and he said as much. He sees it going more into the open weight side and reducing the concentration risk that maybe they've had with the likes of open AI who are developing their own chips, right? And they're having some quite significant success with those chips. So yeah, it's interesting strategically. Yeah, talk to me a little bit more about that point you just mentioned there about the NVIDIA's newfound belief in open weight and reducing this concentration risk. I saw in your notes something about margins from app store fees, which sounds rather Apple-esque. So what's that in reference to?

39:18Yeah, absolutely. So again, there's lots of different ways that NVIDIA makes money and it doesn't want to be reliant upon just selling to the Googles, the Anthropics, the open ais etc so this app store analogy where it takes a cut of the revenue that is generated through something like the models created on hugging face using nvidia chips is that is that very very high margin business that is going to continue to boost the gross profit margins of a company that already has 75 plus percent gross profit margins. So again, when you look at a company like Apple, by far the highest margin part of the business is the app store.

40:11It's not the hardware. So considering the gross profit margins for the hardware part of NVIDIA is 75%, if it can get 90 plus percent on the kind of software layer, that's pretty insane. And then from the hugging face side, I'm just looking at the rationale for them. Is it the normal, you know, the actual specifics aside, the normal sort of tech challenges? Like I always remember when the documentary in DeepMind got bought by Google for like 400 million, was it? Something ridiculous like that, probably the steel of the century. but that was to get access purely at the time for the compute power that they needed for the mission of what they wanted to accomplish so is this is that the kind of basis of the rationale of hugging face in this instance yeah i mean the basis the rationale is the fact that it was 13 billion dollars you can't get beyond that right yeah i don't care what clem says 10 years after starting, this is his tweet, 10 years after starting Hugging Face, open source AI is at an inflection point.

41:23Thanks to the community, we've shown that it could be a complement and even an alternative to closed sourced APIs. But for it to happen at a larger scale, it needs more compute, more support, more collaboration, and more visibility. That's why we went to talk to Jensen. So obviously, that's the party line. I need to get acquired in order to scale up what i'm doing which is perfectly i i definitely buy it but i also buy the fact that they're now billionaires and just quickly on the the risks to this then is it what you said the the template of success seems to be just do what microsoft did with github let them kind of run their own thing uh keep the integrity of the the community and you've acted as a block then to stop other competing places uh forces get hold of that downstream vertical on the on the supply chain if you like is that the risk though if jensen goes i mean culturally how do you think these firms would be like i've i'm looking at zoom graphics of this it feels a bit bromancey and there's like holding up of emoji signs and like is this all part of the jensen wang like um he loves putting a show on but i imagine he's a pretty intense ambitious guy and you're going from a community kind of inspired and quite techie geeky orientated space where nvidia is a commercial beast so how is that the risk there of friction do you think i don't i wouldn't see that to be a particularly big risk and i don't know the internal culture of either of these two companies But they are both drinking the Kool-Aid of the unlimited potential of AI.

43:17And they both believe that they are doing something transformational from a global upside perspective, abundance, whatever it might be called. And obviously, at the same time, they're getting fantastically, fantastically rich off the back of it. So this love-in is a mutually beneficial love-in. From both of them, the biggest risk is that the community that are currently on HuggingFace moved to another platform, maybe a model scope, which is like HuggingFace, but mainly for the Asian markets. And if you just look at the, again, the Reddit comments, it's just a classic, isn't it? you know so a few days ago big money buying out open source the people that lose out in this in this transaction are only dot dot dot everybody a tale as old as time the last nail in the coffin for an ai monopoly what a shame everyone this is a great one which i totally agree with everyone is committed to open source collaboration freedom etc until the offer sheet comes in you say 12.9 with a B?

44:29Question mark. Bleep, yeah, brother, I'm in. So like, you know, these guys are both drinking the Kool-Aid. They're both getting fantastically rich. They both love, you know, they both love the concept that they're changing the world. And in this ecosystem, an echo chamber where Jensen Huang's leather jackets are actually cool and a hugging face emoji is legitimate. And everyone else in the world is just going, what the hell is going on? you did also mention someone else who i'm assuming has come out of this in a very positive way was an nba player called kevin durant so how on earth is kevin durant linked to this deal all right let's finish with this yeah kevin durant and many people know kevin durant you certainly will and nba superstar uh so he has got he's sitting on he was an early seed investor in 2017 at a 16 million dollar valuation for hugging face 24 000 percent return 200 240x return oh my goodness me wow what a baller hall of famer that's what that is okay well look this is not his only vc investment he's also got an 80x return on his investment in whoop a 30x return on his investment in coinbase oh he got a good time on coinbase it's unbelievable he got on at 1.6 billion dollar valuation oh man uh 77x on robin hood 52x on mercury i don't know that company but my gosh this guy's got a future hang about though hang about these are what six you've just described how many shots has he taken what's his shooting percentage here to get these massive payouts you don't talk about the shots you only talk here we go here we go with the stat sheet here because i don't know why he's shooting field goal percentage these days 45 honestly 240x on hugging face my gosh again again if you if you have capital and this might be one of the concluding points if you have capital you get access to better deals.

46:47And if you're in the right place and you've got money and you've alerted the ecosystem that you're up for investing and you're a big name, which Durant certainly is, then, and you're potentially quite well-advised because there's plenty of NBA players that have not been well-advised and have lost a lot of money, but Durant's not one of them. yeah there's a famous one any nba fans out there vin baker he's like a regional manager of starbucks and he got paid a hundred million dollar contract i think back in the day oh god i'm sure he's a superb uh manager of starbucks i'm sure a little that but um all right good stuff all that i would love actually if people have made it to this point in the episode you're probably quite plugged in to sort of tech and the intersection with finance and I'd love to know your thoughts about what do you think let's bring some of that reddit vibes for the comment section particularly on youtube I know it tends to fire up if we hit the right spot so what do you think about this deal and where do you stand uh are you uh wave the right dollar bill sign in your face and I'll sign kind of guy or are you more of a the community community wall street bets till i die kind of vibes yeah i just you know i'll leave you with my final thought go read or go go investigate capital as power sas p is what it's known for short um so this is a theory that was posited about 15 20 years ago but it's it's come into the limelight very recently the more capital you have the more power you have the more ownership You have, obviously, money is power, but we're seeing it in ever more concentrated ways.

48:35And the outcomes aren't always positive. To quote that unfamous person on Reddit that you just read out, it's a tale of older's time. Exactly. All right. Thanks, Stephen. See you next time. Thanks, Ant. Ryan Reynolds here from Mint Mobile with a message for everyone paying big wireless way too much. please for the love of everything good in this world stop with mint you can get premium wireless for just 15 a month of course if you enjoy overpaying no judgments but that's weird okay one judgment anyway give it a try at mintmobile.com slash switch up front payment of 45 for three-month plan equivalent to 15 per month required intro rate first three months only then full price plan options available taxes and fees extra see full terms at mintmobile.com