The SpaceX IPO, Fable 5, AI Capex Update & Market Check w/ Gavin Baker, Andrew Fox & Clark Tang | BG2

“In 30 days, we went from not being an AI hyperscaler to being number 4. And we passed a lot of companies, including Oracle.”

“I just think it's an unprecedented situation. And the right answer is, I don't know what's going to happen in the short term. And the right answer that I would just encourage every investor making their own decision is to just think exactly the way you articulated it. We have these different levers. We have these different variables. Think about each one of them from first principles. Make your own decision. Do your own due diligence. Be thoughtful. But there are a lot of variables here. And then it is a little funny to me that, you know, it was 100 times trailing TTM revenue. Well, after the deals they signed, I think it's at 39 times.”

“100,000 GPUs, that's, you know, easily the fastest supercomputer on the planet as one cluster. A supercomputer that you would build would take normally 3 years to plan... And then they deliver the equipment and it takes 1 year to get it all working. Yes. We're talking about 19 days.”

“Absolutely. So that's kind of crazy in 30 days. That's just extraordinary. What I would say is that I think there is a belief that these data centers are commodities, and I do not share that belief. I don't think anybody around this table shares that belief. And in the same way that Elon was able to re-engineer a rocket from first principles and make it reusable, he engineered an electric car from first principles. You know, everyone else was trying to, you know, make an electric car like an internal combustion engine car, and he thought about it differently.. And I think he looked at data center design from first principles, and he designed something fundamentally different. And I did actually ask the team, I said, hey guys, maybe it'd be a little less public about things that are very obvious to you, right, how to design a data center, but are revelations to other people, because I think what you're doing is maybe, um, more differentiated than you perhaps realize, because what you're doing is so logical to you, but maybe not logical to everyone else. And that's how he was able to do it in 122 days.”

“I mean, the irony is like, You and I have been doing this long enough to know, I mean, that's why Bezos built AWS, right? He had to build capacity for Black Friday. Yeah. Right, but then the rest of the year he sat on all this capacity they had to build and he figured out a really incredible way to monetize this. And by the way, investors at the time, 2009, 2010, when he was building out the capability around AWS, hated it. Of course. Because he was consuming all that free cash flow. Meanwhile, he was digging the biggest gold mine in the history of the world.”

“I think we're all pretty AI-pilled. And if you're AI-pilled, that means we got to build a lot more compute than the world thinks, and that these models are going to be a lot more valuable than people think. You combine that with their core business, I don't know another entrepreneur or another business that's a better bet on the future, right, than SpaceX. And so I think for most institutional investors, it's a must-buy, a must-own. It's set it and forget it, right, in order to have a real bet on both the space and the AI future. All right, here we go. Early morning Silicon Valley, BG2 is back. We're chopping it up on all things tech and markets. To do that, I have none other than GB in the house, Gavin Baker from Atreides. He's brought his main guy, Andrew Fox. And of course, I had to draft Clark Tang into the mix, my partner. Um, to talk, to talk about some of the big questions of the day. You know, how should we be thinking about the SpaceX IPO? You know, what are the big levers? There are big numbers out there for what's going to happen over the course of the next few years. So let's break that down a bit, help simplify it for folks. Mythos launched yesterday. I want to talk a little bit about like who's up, who's down in the race for super intelligence. Where are we? What did we learn with the Mythos launch? And Clark was in, uh, in, in Taiwan last week, um, with Jensen at Computex and GTC. So what was our takeaway there? What's going on with GPUs, memory? Where are the bottlenecks and where do we go from here? To start everything off, you know, maybe just kick it over to you, Gavin, talking about the SpaceX IPO. The IPO is in 2 days. You're a big shareholder. Congratulations. We're also a shareholder. You know, we also expect to be buying in the IPO. The Wall Street Journal is reporting, you know, the Goldman Sachs are both saying $160 billion in revenue in 2028. We know that the IPO is $135 a share, $1.77 trillion. So when we think about kind of what the big levers are, there's so many moving parts in this IPO. Nobody's better than you at just breaking it down, simplifying it. What are the key levers that we ought to be thinking about, that you're thinking about over the course of the next few years?”

“50%? 50% of the company. And by the way, he's locked up for 365 days or 366 days. So we know he's not selling.”

“Right? Listen, when I look at the bull-bear case on the IPO, right, the bears are looking at last year's revenue, say it was $18 billion, and they're looking at the forecast from the banks of $160 billion, you know, 3 years from now. And they're saying, listen, not many companies in the history of the world have basically 8x their revenue over 3 to 4 years, right? So that's where, you know, I think people get nervous about the valuation. When I look at this again, when you break it down as an analyst, first principles, part by part, which is what I tried to do here, right? When you look at Starlink, it looks totally doable. When I look at what they're building in AI compute, terrestrially looks totally doable over the course of the next 3 years. When I look at the model itself after the acquisition of Cursor, you know, combining those things around the compute they have, that looks to me like it could be an upside surprise. So I would say that I think that in the IPO, but I think when you look back 3 years from now, there's a decent chance that everybody's like, oh my God, that was super obvious, right? Even though today, all of these things have risk associated. Back to where we started, I'm not, none of us are here to pump the IPO at $1.77 trillion. It's really to just break it down as we do inside our shop and to say, what is that distribution of future probabilities? What's the probability that it's higher from here? And I think we're all pretty AI-pilled. And if you're AI-pilled, that means we got to build a lot more compute than the world thinks, and that these models are going to be a lot more valuable than people think. You combine that with their core business, I don't know another entrepreneur or another business that's a better bet on the future, right, than SpaceX. And so I think for most institutional investors, it's a must-buy, a must-own. It's set it and forget it. In order to have a real bet on both the space and the AI future.”

“First, agree with absolutely everything you said. And I actually think about it the same way, set it and forget it. You've talked about you have ballast, you move around. And you move the ballast to one side of the ship when you want the ship to lean into the wind to go faster. And you move it to the other side. Exactly. Well, you don't want the ship to tip over. I think that's a great analogy. Think about all important companies in the portfolio the same way. So 100% agree. I mean, this chart is a bummer. What I would say is, you know, this data on IPOs, but what I would just say is this is a really unprecedented situation. We've never had an IPO this big. We've never had an IPO that's going to go into an index this quickly. We simply do not know how much selling there will be from investors. I would hazard a guess. I mean, I'm— I don't know, but Elon, I don't think he needs liquidity. And I think he owns— what does he own, Foxy?”

“Yeah, no, it's a great point. We've in fact called these companies quasi-public. You and I both know that SpaceX, and I put Anthropic in this category as well, Databricks in this category, these things in many ways have been more liquid over the course of the past 3 years than some public biotech companies we know. Absolutely. Right? And so there's a continuum of liquidity here. We treat it as a binary, private versus public, but it's really about this continuum. You know, let's keep going on models. You know, Anthropic launched Claude 4, which you referenced yesterday, which is basically Claude with some classifiers and safeguards around cyber and biology, chemistry, and distillation. When those things get triggered, it fails back to Opus 4.8. You know, there was a Karpathy tweet about this yesterday. He said, you know, it's so down all the benchmarks, but what really makes it special is long-running tasks. Okay. You retweeted our good friend, you know, Noam Brown. You know, ChatGPT 5.5 also exhibited these capabilities. You know, it led Noam, right, to suggest that it's not very relevant to do these snapshot benchmarks anymore. Like the x-axis has to be, time or tokens or compute because we can solve most problems now if we just let these frontier models for a very long point in time. So Gavin, what is this new class of model, right? Claude 4, ChatGPT 5.5. What does it mean for the race in superintelligence? Who's up? Who's down? Who's still on the frontier? Give us your thoughts.”

“There's 100 metric tons in one of those Starships. So you can back into the math of how much will it cost per gigawatt to launch these satellites into space. Right. Launch this compute into space. And the math that you get to before you account for things like bad GPUs, bad satellites, right? These will all be things that happen. But the math you get to is it's about $5 billion per gigawatt of capex to put these in space. Right. For comparison, terrestrially, talk about the switchgears, the generators, the transformers, the shell, getting the power, that today is about $20 to $25 billion per gigawatt. So we're talking about a 5x reduction in cost on half of your bill of materials.”

“You're talking about putting a gigawatt into space for $30 billion. And having lower operating costs. Now, the dynamic— first, $60 billion, that's inflationary. And that $30 billion, that 5, may be deflationary over time. But what we need to consider is, you know, the reliability and the maintenance. And so as long as, you know, everybody can do the math, but as long as these satellites in space aren't failing at an astronomical rate, The math, maths, as soon— by the way, we know GPUs melt and lasers fail. We know this happens in data centers, particularly during big training runs. And yeah, I mean, GPUs melt. So as long as the reliability maintenance is not dramatically lower, the math is there once we have reusability and then rapid reusability for Starship V3.”

“Right. This is the kind of, crown jewel of SpaceX. It's something that no one else really has, notably reusability. Right. And soon rapid reusability. Right. This is, I think, what you need to believe in to get to the economics in AI that make orbital compute something that's very economically attractive. Right. Outside of the idea that we are in shortage for power, shortage for chips. Right. So I think rapid reusability is the main thing that we're watching for and I think most people should watch for. Elon talks about it a lot, but getting these rockets to fly at a cadence that's comparable to an airline, right? And Gavin has used this analogy before, but the old rocket industry was kind of like, imagine boarding a plane, flying to California, getting off the plane, the plane explodes. So I think what SpaceX are ultimately trying to achieve is have a Starship fly both stages, not just the booster, 30, 40, 50 times before you have to retrofit that ship. And when you do that, you're amortizing the cost of the vehicle over many flights, right? And that's what brings the cost down significantly.”

“The founder of Replit. Yeah. He called it bitter lesson adjacent that coding may be the fastest path to AGI and ASI. Because if you're really good at coding, you can write code, if a model's good at coding, to do anything. Correct. So I think that's a profound point, and I think coding is going to continue to be very important. So I think if you think about that variable, if you think about Starlink Direct-to-Cell enabled by Starlink V3, and you think about how quickly they can or cannot bring on terrestrial compute, I don't think orbital compute is necessary for the IPO valuation, but it's certainly important and it's—”

“Right. When I see the models that the banks are putting out there, right? And Wall Street Journal, everybody's reported on these things. Things have been widely leaked. They largely have the revenue on connectivity. So let's call it Starlink, direct-to-cell, etc., going from, you know, let's call it $10 billion to $50 billion by 2028. And so I'm not asking you guys to react to, you know, to tell me your specific numbers. But when I'm talking to Clark, all I'm trying to size up is order of magnitude. Do we think we can 5x the business over the course of the next 3 years? Is there enough TAM both in terms of broadband and direct-to-consumer? And I think the answer to that is yes.”

“When you look at this, okay, so we went through Starlink. And we said, okay, like it just stands to reason that we're going to have direct-to-cell on Starlink. Like the assumptions there are, you know, again, seem like you can get your head around. Then when it comes to building terrestrial data centers, again, not a hard one to think that based on these couple of deals that Elon's going to build a much bigger— Starlink is going to build— or SpaceX is going to build a much bigger business there. And then you have this call option on space that would drop the price even further. The one thing we haven't talked about is their model, right? And I find this surprising, right? 6 months ago, xAI was competing. They were doing pretty well, but they've done something dramatic over the course of the past couple, couple months, which is they bought Cursor, right? Cursor is 700, 800 people, was already doing incredibly well from a revenue perspective. Our own projections were that they could exit this year at up to $10 billion of revenue. So they were growing very fast. One of the leading coding agents, but they also had this incredible team with the potential, right, to really build a frontier-level model, but they were compute constrained. So all of a sudden they get bought by X. X has massive compute that they can now train on. And when I think about the revenue in AI, that like, if I look at that line item in the models, having it go from $10 billion to $150 billion, yes, a lot of that will be the CoreWeave-type business that they have, But the question is how much of that is going to be the core xAI business that's really powered by the new team from Cursor? So any thoughts on that, Kevin?”

“Yeah. So Clark, who I've known for many years, made a— did a great analysis here. And he shows that xAI's deal with Google for cloud computing generates more operating profit per gigawatt than Anthropic, than Meta, than Google, than OpenAI. Their deal actually with Anthropic also generates probably more operating profit than anyone but Anthropic. And so, you know, your colleague at Altimeter, Freda, also, she calculated a 55% IRR on Colossus-1. You know, if you can borrow money at 6%, 7%, 8% and invest in something with a 55% IRR, I'm not the most sophisticated thinker, but that math maths. Right. And so I think the most important variable, one of the two most important, is how quickly they bring on terrestrial data centers. We do know from Jensen that Elon brings data centers up faster than anyone, 122 days. Speed is literally cost. Right. Because every day you're paying electricians and plumbers, that's cost. And they're now monetizing them at arguably the highest rate. And so I think everybody should run their own math on that, but that is a massive variable. Yes. Truly massive variable. The second thing is we have a chart and it's wildly out of date now. It's kind of freaking amazing. This chart is, I think, is this chart from 10 days ago? But in like the 10 or 12 days since this chart, since we made this chart, which shows the Pareto curves for Opus 4.7 for coding, for Codex from OpenAI. And now we've had Opus 4.8, it was already out of date. And now we have Fable. Totally. And Mythos, which is freaking wild. In 10 days, like we would've had to update the chart twice. Right. But what the Pareto curve shows is how much intelligence you can get for a given amount of cost. And I do think being all revenue will accrue to the Pareto curve. All at least kind of frontier model revenue will accrue to the Pareto curve. And this is Pareto curve for coding. And what I think is so impressive is that you could see in the chart that Composer 2 was Pareto dominant, or, you know, at the lowest level of intelligence with very little training. This just reflects, and I know you know Cursor well, I think you know Cursor a shitload better than I do, a vast amount better than I do. But my understanding is that Cursor and Anthropic have more tokens of proprietary coding data than anyone else, and they each have more tokens of proprietary coding data that exist on the public internet. And so they fed, Cursor fed, used KimiK 0.25, used their own private data, did some RL, some supervised fine-tuning, and they got a really good model. And then they spent 3 weeks in the Colossus 2 cluster and they got a model that 12 days ago was Pareto dominant with Composer 2.5. Now it's on their benchmark, CursorBench. Maybe take it with a grain of salt. But I think this just suggests that the cursor data is very valuable for coding. And when it is trained, you know, chinchilla optimal or beyond chinchilla optimal with reinforcement learning, you know, I think it suggests that xAI and SpaceX AI has a shot of being a real player in coding.”

“Right now, so Composer 2.5 was Pareto dominant 12 days ago. It was trained on the KIMI K2.5 base model. Right. Now what's happening is the Grok 4.3, 1.5 trillion parameter model is training. One would hypothesize based on scaling laws that that might be a better base model. And then the Cursor data is being injected into the pretraining process, not just reinforcement learning. And we'll see. And I think that is gonna be a very important data point when that comes out. And I just think everyone should keep in mind that once you are at multiple places on that Pareto curve, if you have compute, you can scale really rapidly.”

“Yeah, Google Search might wanna have a discussion. By the way, I do think it is important. Grok 4.3, I think the cursor, if they acquire it, that may end up being very important. But Grok 4.3 was on the Pareto frontier. And as of 10 or 12 days ago, and these things move fast, most intelligent 500 billion parameter model in the world. And they were on the frontier and there are 4 companies on the frontier. xAI, SpaceX AI, Google, one with Gemini 3.1 Pro, and then the rest of it was dominated by Anthropic and OpenAI. But they were on the Pareto frontier and now we'll see what they do with Cursor. Yeah.”

“You know, that to me is, if I had to say what the one piece that's being lost in the story, Right. Like, it's easy for everybody to get excited about the deals with Anthropic because you can put your hands around that. You know how much revenue it is. I see debate about, you know, the 90-day termination and how long they last and what multiple do you put on those revenues. But I think the thing that's getting lost is I think they've dramatically advanced their capability when it comes to building a frontier model. People outside Silicon Valley may not know, you know, Michael and the team at Cursor as well. This is an extraordinary team. That he just downloaded, right, into SpaceX. SpaceX was already building good models. And what they have is they have this way to monetize compute that gives you this call option that you can pull all that compute in-house, right, to train a model and then to run the model. I suspect if there's an upside surprise, if we went around the table, I'd say this is the place that's getting the least amount of attention and could have the biggest upside surprise. Any thoughts, Clark, on what you think is being overlooked or areas that you think are misunderstood about business today?”

“Yeah. So Clark, who I've known for many years, made a— did a great analysis here. And he shows that xAI's deal with Google for cloud computing generates more operating profit per gigawatt than Anthropic, than Meta, than Google, than OpenAI. Their deal actually with Anthropic also generates probably more operating profit than anyone but Anthropic. And so, you know, your colleague at Altimeter, Freda, also, she calculated a 55% IRR on Colossus-1. You know, if you can borrow money at 6%, 7%, 8% and invest in something with a 55% IRR, I'm not the most sophisticated thinker, but that math maths. Right. And so I think the most important variable, one of the two most important, is how quickly they bring on terrestrial data centers. We do know from Jensen that Elon brings data centers up faster than anyone, 122 days. Speed is literally cost. Right. Because every day you're paying electricians and plumbers, that's cost. And they're now monetizing them at arguably the highest rate. And so I think everybody should run their own math on that, but that is a massive variable. Yes. Truly massive variable. The second thing is we have a chart and it's wildly out of date now. It's kind of freaking amazing. This chart is, I think, is this chart from 10 days ago? But in like the 10 or 12 days since this chart, since we made this chart, which shows the Pareto curves for Opus 4.7 for coding, for Codex from OpenAI. And now we've had Opus 4.8, it was already out of date. And now we have Fable. Totally. And Mythos, which is freaking wild. In 10 days, like we would've had to update the chart twice. Right. But what the Pareto curve shows is how much intelligence you can get for a given amount of cost. And I do think being all revenue will accrue to the Pareto curve. All at least kind of frontier model revenue will accrue to the Pareto curve. And this is Pareto curve for coding. And what I think is so impressive is that you could see in the chart that Composer 2 was Pareto dominant, or, you know, at the lowest level of intelligence with very little training. This just reflects, and I know you know Cursor well, I think you know Cursor a shitload better than I do, a vast amount better than I do. But my understanding is that Cursor and Anthropic have more tokens of proprietary coding data than anyone else, and they each have more tokens of proprietary coding data that exist on the public internet. And so they fed, Cursor fed, used KimiK 0.25, used their own private data, did some RL, some supervised fine-tuning, and they got a really good model. And then they spent 3 weeks in the Colossus 2 cluster and they got a model that 12 days ago was Pareto dominant with Composer 2.5. Now it's on their benchmark, CursorBench. Maybe take it with a grain of salt. But I think this just suggests that the cursor data is very valuable for coding. And when it is trained, you know, chinchilla optimal or beyond chinchilla optimal with reinforcement learning, you know, I think it suggests that xAI and SpaceX AI has a shot of being a real player in coding.”

“I think this debate, like this same debate has existed since the beginning of— since we started training these models to begin with, which was, hey, we're always kind of 3, 6 months behind the frontier. But empirically, like, you can just see all of the revenue has actually accrued at the frontier. And I think that's because every time we release the frontier, a whole new, like, slew of use cases, right, that previously we could have never tackled before, like coding. But also just, you know, we've just been locked at our desks for the last day just, you know, hammering Claude because, you know, it's just fascinating the things that now we can do with Claude 4 that we just couldn't do with Opus 4.8 just a day before.”

“That has been decisively wrong. Probably more than 90%, and it may continue to be decisively wrong. Frontier might be 90% of the economic value. Right. Open source might be 80% of tokens. Something that I think is very important on open source, is that, you know, I think there's this belief that it's bearish for AI. It's actually, it may be very bearish for the frontier models. There's that bear case you talked about. It's actually really bullish for compute and hardware because if the frontier models are capturing less of the margin, then you're gonna spend more on compute. So the better open source does, the better it is for compute providers.”

“We hear the same thing. We did an enterprise survey that we'll post of 300 companies, which ones were optimizing. So these are folks who are kind of looking at model routing and saying, we're going to send certain tokens over here, which ones are thinking about optimizing, which ones aren't optimizing Yeah. And then what is their expected use of frontier model tokens? Right. And they're all expecting to consume a lot more even though they're already in the process of optimizing. Think of it in the context of JPMorgan. If they're doing some back-of-the-house stuff right on customer service or whatever, they may very well use an open-source model. Now, I think they're loath to use Chinese open-source models. So they're waiting on kind of US open-source models to, you know, be able to really deliver the bang that they need. But my hunch is for these enterprises, a lot of that back of the house stuff will get routed there. That will probably be a majority of the tokens. But I think the really high value stuff, you know, coding as an example, they don't want to write second tier code. I think the vast majority of that will continue to be on the frontier.”

“Yes. And I think that this current state is likely to persist. Harvey had a great blog post that they put out on X. And they used, and it's just amazing how everything gets out, out of date like in 5 days, you know? But they use their own proprietary legal data to do reinforcement learning and supervised fine-tuning with Fireworks on an open-source model. And then they used a router, and a router being something that picks which model you send which query to and which model you use to check which model. And they got better outcomes than Opus for either 4.7 or 4.8 at a lower cost. And I think That is the future. And the reality is they were still consuming a lot of Opus, but a majority of the tokens they were processing probably were in their own open source model.”

“They made a great chip. I think the question there and the question for everybody is going to be, is that the highest and best use of your time? I tend to think that the frontier companies, there's this belief that they got to be vertically integrated. But if you believe like I do that the race to superintelligence, particularly as we get these recursive loops working, may be over in the next 2 to 3 years, then I think focus, focus, focus, focus. You exist to build the best intelligence in the world. And to deliver the best intelligence in the world. And that means you have to have all the revenue because if you want to build out the compute that's going to be required to continue to push the frontier, you have to have the revenue in order to support it. So I think that, you know, subject to the focus question, I think they certainly did. This all brings me back to kind of a reality check, though. You know, we just got done talking about test-time compute, inference time compute, long-running agents. This is really the thing. That's unlocked the revenue this year. It all pushes us in the direction of more capex. Google just raised $80 billion, right? We've now taken the Mag 5 or Mag 7 free cash flow, you know, down dramatically, 80% from just a few years ago. And Morgan Stanley, you've got this chart in front of you, upped their 2027 capex forecast from $950 billion to $1.1 trillion. I mean, we were talking about this with Jensen. That was his forecast 2 years ago. You know, obviously this doesn't even include SpaceX, CoreWeave, et cetera. So I think the number on 2027 is likely closer to $1.5 trillion. And if we compare this to the total incremental inference revenue, so the thing that the market gets worried about, you know, back to my Sam Altman podcast, you know, in October of last year, can we really afford to spend $1.5 trillion of capex a year? If we're only generating X amount in inference revenue. The thing I think that lit the fuse this year was Anthropic showed up in a major way with revenue, right? And so we have, you know, the AI lab revenue, everybody combined at around $300 billion next year, right? So can't, you know, and roll that out to 2027 or that is 2027, $300 billion. So we're spending $1.5 trillion of capex on $300 billion of inference revenue. Does that math math for you? And what would cause you to get more nervous again about our ability to continue to make these investments? Because the second we get nervous about it, the entire semi complex is going to come down a lot.”

“I would guess they're probably a little bit higher than that. I might say 60% or 70%. But I mean, that math starts to math. And what I would just say— is I think that $300 billion is low, man. I just think it's low.”

“For sure, for sure. And listen, I would say consistently Elon's been taking the over, Sundar's been taking the over, Sam, Dario, you know, Dario did the podcast with Dwarkesh when he was talking about country geniuses in the data center. He said that will be here by 2028. He said revenues will go into the low hundreds of billions by 2028. So let's call that, you know, $300, $400 billion of revenue by 2028. And he said that a while ago now, so he may even be revising up his number. And he said, it's hard for me to see that there won't be trillions of dollars in revenue before 2030. And if you're on that revenue trajectory, if we're on a trajectory to $200 by the end of this year, let's call it $400 or $500 by next year. And a path to trillion plus by 2029, then the math maths.”

“Anthropic? No way. No way did they think they were going to be anywhere close to breakeven. Yeah. Right? In this part of the curve. And the reason, like, I called it accidental profitability that, you know, people have been talking about that because they want to spend a lot more money on compute. They've just had a hard time doing it. Now maybe with SpaceX, you know, they could take some of those dollars and go spend them other places. But that to me is, you know, a fundamental change. The first argument against the Frontier Labs was they'll never generate revenue. OK? And then that got blown up. Then it was like, even if they generate revenue, it'll be really shitty gross margins and they'll never be able to make money. And then kind of that's blown up. And, you know, I think now, you know, people are falling back and they're saying, well, they're overcharging. This is token maxing. My good friend, you know, Chamath has said there's no ROI on any of this spend. It's all this token maxing. My best evidence for why— we all know, of course, when somebody puts on this much spend like at Altimeter, we're not optimally spending every single dollar. But the question is, why are millions of independent businesses, small, medium, and large, why are millions of consumers all choosing to do the same thing? They're not dumb. These are, you know, rational economic actors that are all simultaneously saying, I want to do this because it makes my life better, it makes my business better, etc. To me, that is the best evidence as to why I think this revenue can continue.”

“Yes, okay. So we'll call it 35% is spending that's not revenue generating that is gonna kind of make the next model. So I think the math maths. And there's still this prisoner's dilemma where if you opt out, that may be an existential decision.”

“Yeah, I was just shocked. I mean, I'm out here. I did a board meeting with one of our companies Earnings release and just, you know, their biggest one thing they emphasized is we thought the world would be, have, be consuming less NVIDIA than it is. And if anything, NVIDIA is accelerating and they just continue to out-execute their competitors. And I think a lot of people are indexing to this OpenAI gigawatt and, you know, NVIDIA has 10, Broadcom has 10. Who has 6? AMD. AMD has 6, and they have warrants. And then Cerebras, our shared portfolio company, has a gigawatt. And I just— that is what's on paper. What actually gets deployed, let's see. I will be very surprised if that 10 out of 27— what's that math? Let's see who's best at math. What percentage market share is that?”

“So I think one of the takeaways from spending time in Taiwan was there is certainly a lot of excitement around the next wave of ASICs. But I think it's like a very clear moment now where NVIDIA— it used to be an argument of NVIDIA versus ASICs, one or the other, and total domination, one or the other. Now, I think it increasingly every year, every— everyone assumed that Nvidia was going to lose share dramatically on a revenue scale, on a gigawatt scale, on a unit scale. And actually, if you actually look at the last few years, you know, they've actually maintained their share very, very handsomely. Actually, if you accounted for the fact that Anthropic was not really using Nvidia, they probably actually gained share against if not for in '25, '26. So I think what was very interesting though was a new class of accelerators or ASICs, MediaTek with their new V8T versus Broadcom's V8I for TPUs actually was a big topic of discussion. And I think for ASICs, The argument now is that more and more will look custom to the actual workload. And that is like one vector that people are moving in versus NVIDIA now has kind of shown itself as the predominant provider of compute to a lot of the world. And for internal workloads, perhaps they will go more and more custom and more and more down the stack. And I remember just one year ago when it was kind of a Broadcom or NVIDIA battle. It seems there's a lot more nuance now to what type of accelerators will fit which workloads and fit which customers and fit which business models. And yeah, I thought that was a new topic.”

“So I think one of the takeaways from spending time in Taiwan was there is certainly a lot of excitement around the next wave of ASICs. But I think it's like a very clear moment now where NVIDIA— it used to be an argument of NVIDIA versus ASICs, one or the other, and total domination, one or the other. Now, I think it increasingly every year, every— everyone assumed that Nvidia was going to lose share dramatically on a revenue scale, on a gigawatt scale, on a unit scale. And actually, if you actually look at the last few years, you know, they've actually maintained their share very, very handsomely. Actually, if you accounted for the fact that Anthropic was not really using Nvidia, they probably actually gained share against if not for in '25, '26. So I think what was very interesting though was a new class of accelerators or ASICs, MediaTek with their new V8T versus Broadcom's V8I for TPUs actually was a big topic of discussion. And I think for ASICs, The argument now is that more and more will look custom to the actual workload. And that is like one vector that people are moving in versus NVIDIA now has kind of shown itself as the predominant provider of compute to a lot of the world. And for internal workloads, perhaps they will go more and more custom and more and more down the stack. And I remember just one year ago when it was kind of a Broadcom or NVIDIA battle. It seems there's a lot more nuance now to what type of accelerators will fit which workloads and fit which customers and fit which business models. And yeah, I thought that was a new topic.”

“Well, you might not have the revenue to fund, to fund that, the revenue or the margins to fund that ASIC. And I do think Nvidia is highly likely to be the world's dominant provider of open source AI. And I do think Jensen will bring open source, you know, right now it's whatever, 6 months behind the frontier. Yeah, we might see it. Creep closer and closer and closer. And I do think Jensen has a big business decision. I see this chart here. So let's chop it up about Nvidia, as you say. But if all of his customers are going to compete with him, then why not compete with his customers? And we have all these Neo Clouds. So that's a cloud computing business that can compete with all these cloud computing businesses. He has his own models that are really, really good. NeMoTron 3 or 3.1 was actually really, really cool from a compute efficiency perspective. And he's always careful to release small models, right? So as to not tread on Anthropic, OpenAI, Google's toes. But I do think that is a choice he is making. And just, you know, if, if the economics change, right, I think Nvidia can join the frontier and become one of the world's largest cloud computing companies. Much faster than people think. Interesting.”

“Yes. That game is over. Yes. You've had a lot of stocks that forget climbing a mountain or hill. Yes. They've gone straight up a cliff.”

“Because nobody has run Claude for a year continuously. And we may never know how smart each generation of models actually is or was, but because we don't have time to appropriately evaluate their intelligence before the next model comes out. I mean, this is a profound statement. And just imagine, okay? So I always say like, when you think about FSD, just imagine a human being who never gets distracted, never gets tired, never talks on the phone in the car, never drinks and drives, never yells at their kids, never has to go to the back seat to give their baby a bottle. And like, of course you would think that over time that is superior to humans who are distracted. I don't know how long, how long can you think deeply about one topic, Brad?”

“It's unlocking all this. I mean, like they gave examples yesterday in the release Anthropic did, you know, 50 million line Ruby code base at Stripe that was refactored in a day versus many weeks with many people. You think about where this is impacting biology and life sciences just across the spectrum. And to me, it really gets back to this fundamental point. Number one, if you believe this to be true about long-running agents, then we're going to produce and consume more tokens in the future as far as the eye can see. So the world— this gets me back to TerraFab and Space Orbital and all this because we may in fact unlock real thresholds of intelligence, but we're going to have to let these horses run for a long time in order to get there.”

“For sure. And I think like coming into this year, going back to this kind of what narratives were violated, you know, I think into this year everyone expected token pricing, the price of compute, it's all deflationary and it will be kind of a smooth line, deflationary over time. But I think this year what we've seen is the opposite. And, you know, it all comes back to supply demand. The demand side of the equation seems to be far outstripping the supply. And I think you look at the deals signed by SpaceX and others, the monetization rates per watt are increasing. And look, that is on a pretty nascent small base of users, right? Like Alex at Whale Rock, he has this great way to frame it. Less than 0.2% of people on Earth are actually using AI. In an agentic way, right? Like, I'm not a technical person, but I'm consuming 500 CPU cores in a VM instance, 5 GPUs 24/7. I mean, if you draw that out to any meaningful percentage of the population, I mean, we're going to be in, you know, this kind of shortage environment maybe for some time. So I think that is all positive for this ROI question.”

“I'm sorry. It's a great point. Like, you thought you were getting— when you made these decisions in November of 2025, you thought you were getting a certain return. You may be getting triple that return today.”

“For sure. And I think like coming into this year, going back to this kind of what narratives were violated, you know, I think into this year everyone expected token pricing, the price of compute, it's all deflationary and it will be kind of a smooth line, deflationary over time. But I think this year what we've seen is the opposite. And, you know, it all comes back to supply demand. The demand side of the equation seems to be far outstripping the supply. And I think you look at the deals signed by SpaceX and others, the monetization rates per watt are increasing. And look, that is on a pretty nascent small base of users, right? Like Alex at Whale Rock, he has this great way to frame it. Less than 0.2% of people on Earth are actually using AI. In an agentic way, right? Like, I'm not a technical person, but I'm consuming 500 CPU cores in a VM instance, 5 GPUs 24/7. I mean, if you draw that out to any meaningful percentage of the population, I mean, we're going to be in, you know, this kind of shortage environment maybe for some time. So I think that is all positive for this ROI question.”