All right, guys. Welcome. Let's talk about the Anthropic IPO as a once-in-a-lifetime event, biggest day of human history.
Certainly in the history of capitalism, right? No hyperbole there.
There's zero hyperbole. What an amazing IPO it'll be. How capital-efficient this growth curve was, how fast it happened. You can't ask for more.
They're going to come in at like the what, sixth most valuable company from the start of trading in the IPO in the whole world.
It'll be up there, I think. Yeah. Like, tokenomics, I think, covers like seven of the top 10.
Yeah. So this will be, this and OpenAI will be trillion plus.
Okay, what do you guys want to talk about for profitability?
I think that's a good place to start. You know, compute, you know, they kind of give the compute margin numbers, but I think even on the GAAP basis, you know, non-GAAP basis, you know, 10% plus margins already, two quarters profitable, really solid. I think what we're seeing on the gross margin line just the past three years, like the scaling of compute and inference efficiencies is amazing, and you know, what they're able then to spend and increase, you know, kind of the training splits, you're really impressive. So 85% inference gross margins, you know, 41% compute margins, and then extremely capital-efficient like sales and marketing and partnership thing too. I think a lot of like software and tech companies spent a ton to go acquire a customer, and you know, their net dollar retention number. So a customer last year at this time that spent a dollar is now spending nine. And in Q1 that was, you know, that was 500%. So a customer spending one was spending five. And that all falls just down to the bottom line. There's not much you have to do to upsell the usage base side of that. So really solid headline numbers all around the business model.
Yeah, I would say really, really interesting to me because if you talk to folks in the data center market, right, I'm the data center bro here. Folks have been looking forward to this IPO for, you know, one big reason: we chose that Anthropic has the biggest compute demand of anyone alongside OpenAI. They're on par with, you know, with hyperscalers, and they can't deal directly with the folks actually building the data centers. One, because maybe they have a team too small they're building, but two is mostly financial: they can't sign, like, you can't fund the data center on Anthropic credit. And so we saw the market kind of hating on Anthropic and OpenAI, like naked exposure, whether it be on the data center side or on the GPU side, over the last few weeks. And so I think it's going to be like very reassuring for folks to see the actual financials. So the first thing I wanted to say is like, talk to anyone on the data center market, everyone was just looking forward to this IPO. Like, dude, whenever these guys are going to IPO, I'm going to sell so much, so many megawatts to these guys, it's all going to be so easy. Obviously that's contingent on Anthropic being investment grade. And so when you look at some of these numbers, how much cash they have on the books, you know, like over a hundred billion dollars, how much they're going to have in addition with what they're raising and the promises they have from like Google and others, these guys are going to have over 200 billion. And as Joey said, that's a company that in the last three years, despite being like very unprofitable, barely burnt any cash. So I think extremely reassuring. Also, the compute commitments frankly aren't that large, which clearly shows you that they've stroked some good, like fairly flexible deals with folks like Amazon, Google, where instead of having to commit for like gigantic deals like Stargate, which is this huge, you know, $300 billion agreement for four and a half gigawatts, they've managed to do something like a bit more efficient on the commitment side, which I think is overall good news and, you know, good for AI infrastructure as a whole market. Jeremy, can you walk us through? I think a lot of people see like the headline, you know, 500 and 600 billion of compute commitments. Can you walk us through why that's not that large?
Yeah. No, it's a good, like people ask that, and it's when you break it down, it's not that much.
Yeah, for sure. I mean, like if you look at recent deals that have been done, such as Anthropic-Endcale, let's take this one. The compute per gigawatt commitment is $100 billion. And so there's a world where you would have to, for like these half a trillion dollars, would be just five gigawatts, which is really not enough for Anthropic to serve their roadmap, right? Like what they have in terms of gigawatts contracted is actually materially higher than that. And so their effective ratio of gigawatts contracted relative to commitments is really low. And I think like, you know, this gives us like right now the compute they have is like under five gigawatts, that's going to go up materially as we get into the end of next year. Plenty of compute to, you know, accelerate revenue growth, cut cost of models, to keep finding efficiencies. And so it's pretty clear all of that is going to trigger an acceleration of revenue. So these guys are going to be in the multiple hundreds of billions of annual revenue next year. And so yeah, it doesn't feel like that complicated for them to pay half a trillion dollars given the revenue they're making, free cash flow profile, strong balance sheets. So I think overall like very reassuring, especially when you compare it to like, you know, what was the first freak out about compute commitments that was last year, that was OpenAI, and people were saying, dude, you're making, you know, 20 billion annualized and you have like 1.4 trillion, right? So I think we're in much more reasonable ratios here. Like, you know, 80 billion, half a trillion commitments, it's a lot, but there's 200 billion on the books post-IPO. So overall, I think, you know, it's going to be fairly easy for them. And as I was saying, like they've done a mix of like long-term deals like Endcale, but also more flexible short-term deals, hyperscalers, and folks like SpaceX, which they can cancel effectively anytime.
Can I... greatest day in the history of capitalism? I actually, I just made that up in terms of history of capitalism, but it felt right.
I mean, like, look, it's a big business, dude. It is a big business. You know, like I remember once upon a time, you know, one billion run rate was like the grown-up number, and they obviously did a lot bigger than that. I think the thing I thought was best on that is like incremental margins are 90% gross margin and 40% EBIT. And that's like, that's a best-in-class business. Like 40% EBIT is as good as it gets. That's like as good if not better than the hyperscalers in terms of like if they get there, right, gross margins at 90% incremental. Like it's very clear. It's on a very large base. I think it's as clean as you possibly could want it to be. And I think people, we've been saying this for a long time that there's obviously really great incremental, like there's real revenues, right? Like when you sell it, you're not selling a dollar for a dollar ten cents. You're selling a dollar for ten cents. Okay, that's margin, right? I think the biggest bears have always believed that there's a heavy subsidiation. Sorry, I can't say that word today. And it's very clear there isn't. So like, I don't know what else you want to say. I also think the fact that Ed Zitron hasn't tweeted is like probably the most interesting fact of all. Like, you know, him calling Jensen a crook last week was like pretty based in terms of like, you know, he's kind of like you're watching the speedrun of a short seller in real time. So yeah, we'll see. I hope he's mentally okay. You know, might be a hard day to be Ed.
Yeah. So inference is a good business model, I guess that's... Yeah, it's a great business model. Yeah. Controversial... controversial, people actually want these tokens. How about the customer concentration? I mean there was a comment in there saying that top five customers represent 25% of revenue now, and then the CSP commitments, the marketplaces are representing another 50%. So what's the leftover 25%? I mean it's just a really big base. So the leftover 25% is just still many billions.
I'm confused about what you mean. I think you think that it's either-or, mutually exclusive. 58% is the total, and it's like the total distribution, and then 25% can look through that if that makes sense. They don't add together. They could, like, you know, 5% of that customer, the entire top 10 customer could be 100% on the CSP.
You think that some of the top five or top 10 customers are actually going through the CSP marketplaces? It implies that the numbers show that, right? Because to be clear, I really want to make sure, Jordan, we get this straight on the podcast, that 58% of distribution is like their sources, right? But I don't think when they're talking about their net customer, they are not talking about the CSP as the net customer. They're talking about the, look, you know, I'm assuming there is some kind of pass-through, meaning that when Anthropic gets billed, they get the customer who purchases through the distribution platform, aka Bedrock. Right. So...
Okay, let's just cut this part out where I'm being stupid then and not talk about customer concentration.
Well, no, we can talk about the 5%.
No, no, we should talk like Jordan.
No, I mean, dude, the fact that you didn't, you like, it's important to know. People don't know that.
It's just the top five doing 25, which is pretty good. Honestly, that's most, the companies are actually more concentrated than I know. The top five are doing the 25, and then there's an additional 48% that goes through the CSP marketplaces, which does not include the top five customers, because elsewhere in the S-1 they also talk about how they have a direct relationship with people who are making purchase commitments to get a discount long-term, like any other.
Yeah, Doug, I think the point he's making is that like the 48% of revenue, none of the top five customers are going through Bedrock or Vertex.
Okay, great. Surprise. So then there's... which makes sense, like... Yeah, that kind of makes sense. Yeah. Why would you... those customers would buy through that channel?
So then you have that, you have the CSP marketplace, which we know is like going towards your traditional buyers, you know, just your large AWS accounts, your large Google GCP accounts that are buying through that, you know, a little more Fortune 500. And then you have another, so that's 48. Then you have another 25, and then there's 27 left over. And I think even the point there, and there are some really interesting stats on like some of these customers, is, you know, there's actually, I think that's still rather concentrated. I think if we look, we still look at any, you know, alternative data set, the top 1% of customers make up, you know, 70, 80% of the panel. And they said this, you know, the top 5% of cloud enterprise customers, users I should say, at these end companies are spending 18,000 per user per year. Your non-cloud code user spends less than 75 a month on...
You're talking to three, maybe four of the top 5%. Come on.
Well, I was wondering, there's 251 million customers on a trailing 12-month basis. We, I know we're in the million-plus group, the 1900, one of 1900. We've got to be getting close at some point.
Yeah. At some point we'll be in that 250, I would think.
So that's not a, if Dylan's listening to this, that this is not a KPI, but... not a result, it's a result of some other KPIs being enforced. Not that we have KPIs. Okay, but talk about the actual customers. They named customer A, B, C, D. Somebody started at 43% of their revenue, went below 10%, is now back up over 10%. We've got our estimates for who those people are, but let's just say...
Like you can't look at 2023 numbers, I don't think. Like there were customers at OpenAI that were over 10% in 2023 that are nothing today.
So I don't know if the 2023 numbers, I mean we know some of those large customers. You know, today that are, you know, there was a one, I think 10% customer, and you know...
Who do you think, who do you think, like let's do some speculation. My meme was that it's Kimi, Moonshot AI, that was five to seven percent.
Yeah. Yeah. Yeah. Five-star percent customer.
That's, well, okay, to be clear, they're, I mean, even I don't know if their KYC is good enough to completely cut them off, right? I'm sure someone at Moonshot is using Claude Code. Like if you saw on the bench, I forget which benchmark, I think on DeepSeek they're using Claude Code to be the final verifier of, like, you know, to verify all the results. So they're using it obviously, but maybe not distilling.
So who do you think top five is, what I'm saying, like let's speculate.
We know Meta was the one that, you know, rationalized spend in Q3.
Or not rationalized spend, they moved from an external provider to internal. The tokens are still there from an industry perspective.
I think Microsoft is one, Cursor we talked about as being another one.
Nvidia, hedge funds like Jane Street, HRT.
Do you think Nvidia is that big of a spender?
Yeah, definitely. I think they're probably the 43% that went lower. Who else is using this much stuff in 2023?
I mean, my understanding is they were really late to getting access to Claude. They were a cloud adopter in '24.
I don't think they were using it. I don't think Nvidia was using it in '23. That's my impression. I might be wrong, but I feel like Jane Street probably feels like the most rational, or maybe you can argue Google or something. I don't know.
I mean, that would be crazy. I know Jane Street is producing a ton of cash flow right now, but they have such big commitments with OpenAI already. It'd be crazy if they were also a top five and so big on Anthropic as well. But I mean, it's totally possible. There's only so many people that can be spending tens of billions of dollars on these tokens.
Oh, well, actually, no. I was going to say, I feel like, yeah, I was going to say, I don't know. It feels, we're going to figure out who it is over time, and I bet you it's going to be like someone really obvious that we just didn't think about, essentially. I mean, the point of clarification for me is that like these top customers are not the CSP marketplaces, because they have this direct relationship with people through the CSP marketplace if they're using Bedrock or Foundry or Vertex. And I think that's also pretty interesting because if people end up having this direct relationship with Anthropic and they make these purchase commitments ahead of time, the way they do with like an AWS MSA agreement, that's a big lock-in for using those models. Made another chart. They said that in Q3, 20% of the revenue was from models released before April. So that's Opus 4.6, 4.5, which they kind of said, okay, 80% is the latest model, so people move to the latest models. But I'm actually shocked that 20% of revenue is still stuck on models that are three or four months old. Like, how long does it take somebody to get off of the old models and onto the new ones? And I would think that somebody buying through Bedrock, like a Fortune 500 who has to certify something, they're going to have that lag time, whereas a startup's going to be a little more nimble to move around. And so both of these things just point to me as being like really sticky.
I mean, if we're saying Microsoft is a top five customer, then they're probably the one spending on all the Opus 4.5. Yeah, I think via GitHub Copilot also, but you know, Copilot 365 or something like that, like the Cursor, Perplexity equivalent where they don't pass through the relationship to the end user. It's kind of their API key that's doing all the spending. Microsoft has both the Foundry business where they do pass through the relationship, and then also the GitHub Copilot stuff where they consume so many tokens. And you can probably see, if they had more granular data, you could probably see Microsoft spend be high at the beginning, dip down when they have all these OpenAI options, and then come back up now that they're actually more motivated to serve Anthropic and not tied so closely to OpenAI. We'll see.
Okay. What else is hot? What else is hot about this S-1?
Compute commitments. Jeremy, talk about the $70 billion MPA.
Oh, sure. I mean, they're just essentially going to get even more money than they already have, which is mostly good news. A bunch of them are already known if you look through the filings of like Google and Amazon, but they've signed a bunch of like large fundraisers. For example, the 40 billion with Google, which have like multiple milestones, right? So like Google has a labor 20 billion, there was a remaining 20 billion which is just going to be more cash for Anthropic based on milestones such as, you know, going public. So effectively like all of that is just additional funding, which again, like as I was saying earlier, just means these guys are going to have more money on the books, going to be easier to pay for stuff. They got the Broadcom one as well, which is $42 billion, which is going to enable them to buy a whole lot more TPUs. And so yeah, overall like the 2027 compute roadmap for Anthropic's risks, which kind of goes back to what we said like in an article on a podcast a few weeks ago about like the data center build-out, like yeah, this is most of what they're going to get next year is de-risked and it is going to be huge. So the implication is, okay, they're going to have all of that compute next year. Then can they actually accelerate the revenue enough, right? And I think that's still the question. But maybe one thing that folks weren't aware of is like how much do they have on the books right now? And now we know, and it tells you at least they're not going to go under.
Yeah. And if rates go up, they'll even make more money, baby. You know, hell yeah. Make, you know, rates are five per, let's say rates are 5% a year, they're making 5 billion ARR, you know.
That gets us to one trillion almost, Doug.
Yeah, dude. Just a little bit of the way. I honestly, the thing that I really thought was most interesting in the whole S-1 is the crap that they put in, and like honestly, I think they've been pretty consistent about this. Dario's been saying this for a long time. It's like we never want to overextend ourselves, right? And then essentially just being like we'll never miss your expectations is an insane thing to put in the S-1. Like that is insane. I think this is the most nutty thing they could possibly do, and then everyone is like really worried and barfed up about Q4 because what they've been telling the bankers for ARR, and you're like, they just told you they're never going to miss your expectations. They're like, we've talked to all these, oh, I think Joey's frozen. We've talked to all these people and all these stuff. You should never be worth one more dollar than what you're actually worth. And so we'll underpromise and overdeliver. I was like, oh my god, dude.
Yeah, that's a core tenet of the business. Underpromise and overdeliver, Doug.
Yeah, I guess that maybe that's in their fit interview, dude. I mean, it's crazy.
All of this talk about, okay, wait a second. So this is a core principle of the business. So all of this talk about cybersecurity ending the world and drugs and all of this stuff, this is them underpromising for the AGI that they're forecasting 10% chance of death of humanity. This is underpromising.
Yeah. Apparently, according to Dario, it can get worse.
It can get way worse. It's the worst. Did you guys see the head of cybersecurity? He's like, if you think it's bad, it's gonna get way worse.
I'm laughing. Scary, man.
It is actually scary. This is going to be our last secure podcast. I was freaking out last week with all of the cybersecurity stuff, and then interesting stuff over the weekend as Elon and Chimoth are calling for the death of all Neoclouds. And then we have this counterweight to all of that, which is like we'll save it with capitalism. I guess that's the thesis of today.
I don't know. Yeah. I mean, the money looks good to me. I think Elon is just, Elon's smart and he knows when he has an advantage, and he's trying really hard to lower his cost of capital at SpaceX. And I think he's, I mean, he is a Neocloud dude, so he's saying that while being the largest Neocloud in the world, so yeah, it's kind of book.
I would also say like our view, and I think it's starting to, like folks are starting to be really worried about that, is hey, there may be an inversion of like right now it's data centers that are more constrained than chips, but maybe by the end of next year it's the other way around. And there you have this guy that says, hey, I'm single-handedly going to build up to 20 gigawatts, maybe just 10, you know, so he's going to make sure he has the chips, right? And so that's the game of Elon, dude. One of the ways to make sure others don't buy the chips is just better make them go broke. Yeah, just break markets. They can't fund them all of them for you.
I mean, that's happening like right now. There's essentially a liquidity run on all the Neocloud debt. And so anyone who wants to place new Neocloud debt isn't able to. And so if he presses, you know, if he presses his advantage, like he presses his foot at their throat and they essentially never get funding, that's all his opportunity, right? Because then everyone's like, dude, I want to invest, I want to invest, I want to do this SPV. And SpaceX is the only game in town. And he's like, great, that's a better cost of capital for me, right? Elon is a competitor.
Yeah, exactly. And then as he said, he wants to do half a trillion dollars to a trillion dollars of ARR next year. The less these guys can fund their stuff, the more he can just build it himself and score deals at 50 million bucks a megawatt a year, and you know, just make a whole lot of money.
Yeah. I mean, Jeremy, how real do you think it is that $14 an hour GB300 contracts that are on demand and can be canceled any month just continue to go because Anthropic just continues to grow and they can't bring enough compute online, and so they just keep it?
I mean, look, like there's many things that go into like understanding how sustainable this is. But one of them is if the market, first of all, it's the basic supply versus demand. If demand continues to outperform supply growth, we've said very publicly we think supply growth is going to be huge. But hey, what if ARR and overall demand also outperforms that? Then that's perhaps even going to augment the premium, like that's a realistic scenario because who else has as much capacity available in 2027, right? Now if the equilibrium stays the same, then I think there's still probably availability to sell, and obviously like if demand starts to fade, then of course things are more complicated. Now there's a few other things to look at. One, I mean we talked about compute commitments, kind of the beauty of SpaceX contracts is it makes your, it's like much more balance sheet efficient or like compute commitment efficient than signing a six-year deal, right? So you saw at scale at like $100 billion per gigawatt, that's likely going up, so perhaps going to 120 or more. And this guy actually 50 a year, but 90 days, so the commitment actually is extremely small, right? It's like $15 billion a gigawatt. So yeah, it's more expensive, margins go down, but on the other hand for your books, for what people are looking at, it's actually quite efficient. And the last thing is obviously like as the drivers of this market remain like these small companies that don't have yet full trust of creditors, of lenders, it's going to be hard for them to fund an insane amount of supply. I mean even for hyperscalers it's not that easy, right? Like we saw the tiny tiny cracks on like Meta for example when spreads went up. So obviously like this environment is very positive to companies that can take a lot of risk, and these guys are basically taking more risk than anyone else by just equity funding all of that stuff, some as well, but and you know building stuff without a customer.
Yeah. Let me throw this chart on screen. Joey, what's your take when you saw this chart that they put up about compute efficiency?
I think it's great. I think even looking at just inference compute efficiency, the revenue per megawatt just in inference going from, you know, like 16 million a megawatt last year to 60 plus, you know, currently. It tells even more the story of how great a business model just inference is. And this is the entire fleet. So what you need to do to continue to invest in new models. This is great. Obviously in Q3 it went down a little bit with some of the new SpaceX compute coming online, and then a little bit less monetization via Meta distillation, and then some additional safety compute. But I think this really tells the story of hey, AI model businesses are really solid and they scale extremely extremely well. To Doug's point, incremental margins are amazing.
