Dwarkesh Patel and Dylan Patel of SemiAnalysis discuss how AI lab economics are shaping the compute market. They examine how Anthropic and OpenAI are moving from venture-funded losses to profits, how their revenue per megawatt compares with the cost of compute, and why they may come to control a large share of the world's usable compute by 2028.
The conversation covers fab capex, supply chain bottlenecks, China's share of new compute, and the shift of lab compute from inference toward research and training. It also explores whether roughly $11 trillion of AI capex could push up interest rates and trigger sovereign debt stress, and what the concentration of AI labor could mean for future power.
Listeners will hear the guests' projections, which are presented as forecasts and debated rather than settled facts.
In this episode Dwarkesh Patel, Dylan Patel
This episode's transcript on the economics of AI compute and centralization was made with HappyScribe's automatic meeting notes technology, a tool for turning recorded speech into text.
Ads in this audio differ from the text. Tap the sentence you're hearing to follow along.
Synced: the text now follows the audio.
Okay, I'm back with Dylan Patel, founder of SemiAnalysis. Our version of a family Thanksgiving dinner is a regular yearly podcast, but we are not actually related.
Don't tell the people this.
It will destroy the myth. Um, walk me through—so basically where the world economy is headed is more and more becoming a function of where, like, lab economics are headed, where the compute market is headed, etc. So I want to understand where the crazy future ends up within a few years, but, uh, let's start with just where we are today. So walk me through lab compute and lab revenue right now, and maybe projecting out a year or two.
Yeah, so when we go back to last year, even at the end of the year, most of GDP growth in America was just AI infrastructure. And as we look towards this year, about a third of the compute coming online is for the labs, for OpenAI and Anthropic. Now, it may be built by others and then rented to them, but it's at the end customer, it's them. Um, as we go forward into the future, the numbers for computer ballooning, right, we're at, you know, you know, a little bit over a trillion dollars of CapEx this year.
As we go out into '28, it's going to be more than two trillion dollars. Um, the labs are also taking an increasing percentage of this. And so ultimately, you've got a very interesting situation where the labs are going from companies that spend, you know, tens of billions of dollars a year to hundreds of billions of dollars a year to forecasting to spend trillions of dollars a year even, um, towards the end of the decade. And this is at least some of the contracts they've begun signing with their partners. And so this requires a big reshaping of what happens with their, um, economics, right?
You know, so up until now, they have been companies that mostly lost money. Um, Anthropic started turning a profit in Q2. Um, it's believed at some point in Q3, OpenAI could potentially start turning a profit even, um, with the big rise of Codex and 5.6 and all this. Um, but if we go back a year ago, everything that they—all the money they had was venture-funded losses, right? If we go back to even the beginning of this year, it was venture-funded losses. Um, they've now turned the corner and are actually starting to profit.
Now, that doesn't mean they're not taking in new capital. The new capital is still coming in to accelerate the growth further, but ultimately, there's more and more of their businesses being funded off of their own revenue rather than capital injections into them. Over the last, you know, year and a half, their margins have really skyrocketed. You know, the base cost of compute tends to be around 10 or 13 or 15 million dollars per megawatt. The most interesting aspect about what's happening now is before, again, they were generating, if they served a model, right, GPT-4 being served on, um, you know, NVIDIA Hopper GPUs was generating negative gross margin for OpenAI.
But now, when OpenAI serves GPT-5.6 or Anthropic serves Opus 5 or Mythos, uh, Fable 5, their revenue generation has passed well beyond the sort of incremental 10, 15 million dollars per megawatt. In the case of Anthropic, the revenue has gone as high as 50 million dollars per megawatt. Um, and what that now enables them to do is, hey, if I spend 10 bucks on inference capacity, I actually generate 50 bucks of revenue, and then I can turn around and incrementally spend all of that profit on training.
One thing I'm very interested in understanding is how you see decentralization of compute happening in the labs or the relative ratio of compute that goes to the world versus goes to the labs. Um, where if you say right now, a third of marginal compute is going to the labs, by when is it over half of the incremental compute in the world is going to the labs? And by what point do the labs have basically a vast majority of the world's compute?
Yeah, so, so earlier this year, you know, the beginning of this year, Anthropic OpenAI started at two for OpenAI and less than two for Anthropic. Um, end of this year, they're both above five. Um, so they've 3.4x compute as a whole. Um, when you, when you look at the incremental compute added, that's about 30 percent of the compute added this year. And as we step forward to next year, given what's already been signed and penned and inked, you've got something even more dramatic, right? You've got Anthropic OpenAI taking, uh, as much as 40 to 50 percent of compute, uh, next year.
And the centralization doesn't look like it's slowing down or stopping. In fact, it looks like it's only accelerating.
Right.
Now, who's building that compute for them will change. Um, you know, next year, a big new entrant is, for example, SpaceX is building a ton of compute, and they're actively going to lease quite a bit of it to Anthropic and OpenAI, most likely, because they're the ones who can, who have the marginal capability to pay the highest price. In addition, OpenAI and Anthropic are also starting to build their own compute. OpenAI with their own chips, Anthropic with TPUs that they're purchasing from Google and deploying with Fluidstack. And so when you ask, hey, when does half of the world's incremental new compute go to just OpenAI and Anthropic?
I mean, it's, it's really by the end of next year, it's already half of the incremental compute is going to Anthropic and OpenAI.
Because compute is growing so fast, incremental compute is going to be basically most of compute. So it's very soon, you're saying maybe within a year and a half or two years, most of the world's compute is owned by two labs or at least is serving the demand from two labs. Um, how long do you think, so there's this trend where maybe world compute in gigawatts doubles every year, but the compute at the frontier labs triples every single year. But if you keep the current trend going, it goes from like two at the beginning of this year to close to like six at the end of this year, just multiplying out by three, 18 by the end of 2027, 54 by the end of 2028.
Are you like, okay, at that point, they simply can't continue tripling given the amount of world compute or how, how do you see the world compute situation over the next few years?
Yeah, so if the incremental compute adds this year 30 gigawatts, next year 50 gigawatts, and the year after that 70, roughly, you end up with this really interesting phenomenon, which is, okay, well, a new watt deployed this year is significantly more efficient than the watts deployed two years ago. So actually, you know, a humongous percentage of the world's compute was deployed this year, even though it didn't double the number of watts deployed. I'm deploying GB300s and TPU v7s and Trainium 3s, which are way, way, way more efficient, you know, 3x, 5x more performance per watt than the prior generation chips.
And so ultimately, you've got a huge, um, ladder here. So if Anthropic and OpenAI take on, you know, 45 percent of compute next year, you've, you've got them in, let's say, December 27, they have taken on half of the world's incremental new compute, but that half of the world's new incremental compute is actually at a higher performance than everything else before it.
Right.
So you've got another multiplier on that. So by the time you're in, like, towards the end of 2028, if this trend continues, which I see nothing that's stopping it, um, you've got them just controlling most of the usable, you know, flops in the world on their own.
The thing I'm confused about is why you think we only add 80 gigawatts in 2028 if we enter into a world in which the price, the value of compute increases so much.
That's the upper bound, by the way. That's the, that's the, like, I'm so fucking bullish.
Right. Okay, so let's, let's do some chain of thought here. So when I interviewed you a few months ago, you said, in order to make a gigawatt of, I think, Vera Rubins, you need, one sec, you need 55,000 N3 wafers, 6K N5 wafers, and 170K DRAM wafers. Um, I don't know those numbers might have changed.
I'm going to troll you, but the way you said wafers was so fucking idiotic. Wafers.
By the way, when we first, uh, when we first moved to the U.S., I had the VEW thing pretty bad, and I was a vegetarian.
Vegetarian? I remember you told me about this.
In Northe Dakota, I was in elementary school, and I'd be like.
Can I get a wedgie?
Can I get some wedgies? Anyways, so that's for one gigawatt, right?
Yeah.
Now, I had an LLM run your wafer fab equipment model and figure out how much, um, tooling, how much the tooling cost to produce a gigawatt of compute basically every single year. And it was said like three to four billion dollars. Now, suppose you add in, you know, clean rooms and, uh, shell and everything else at the fab. So six billion dollars of like fab CapEx produces every single year a gigawatt. And a gigawatt produces right now a hundred billion dollars of revenue.
But also that six billion in CapEx is producing a gigawatt every single year. And that gigawatt is producing a hundred billion dollars every single year. So even over the course of five years, so, you know, the first gigawatt is generated, five years of profits, the second gigawatt that the fab is produced is generated, four years of profits, and so on. Um, six billion of CapEx at the fab level will have generated over a trillion dollars of end revenue, end AI revenue.
Yeah. There's a lot of Opex along the way. There's a lot of, um, other CapEx like the data center, the power.
And you have to pay like a, you know, OpenAI for the R&D.
Installation. There's a lot of different people who need money here.
So, but take away half, take away half of it for all this, all these middlemen. That still means there's a hundred X discrepancy between fab CapEx and end revenue generated. More than that actually really, but, uh, we're just being very conservative. And as a result, this is capitalism, right? Like you would imagine that people are going to figure out, like, we're going to be, you're, you're, you have this huge discrepancy where you can turn one dollar into a hundred dollars and you're, they're not going to figure out a way to make more mirrors.
I mean, they are.
Right.
It's just these mirrors take some time to make, right?
But the emergency are so big where like, Anthropic and OpenAI are like, we could make a trillion dollars right now, but we're just bottlenecked on the mirrors that go into the ASML machines. Like they'd spent, okay, how can we make more mirrors if we spend a hundred billion dollars on this? Right. That's the situation we're going to be in pretty soon. And I'm just like, we're not going to be able to solve that supply constraint. That just seems quite hard to imagine.
No, there's definitely, um, you, you, you've seen people do funny arbitrages here where they buy like turbines and then they try and resell them.
Right.
Because the value of a turbine is way more because it's the thing bottlenecking a data center. You know, I think like if anyone had like 400 million dollars and the ability to convince ASML to sell them an EUV tool, they should totally just go buy one and, and wait, wait, wait, and then sell it for north of a billion dollars, right? But, but ultimately, like, yes, capitalism will cause these things to expand, but it's, it's a whip, right? It takes a long time for the whip signal to get to the tail end, um, of, of that. And so the supply chain doesn't react immediately.
In fact, you go to talk to someone at Carl Zeiss, they're like, yeah, yeah, yeah, we need to make a hundred EUV tools by the end of the decade. I think when we first had our, when we had our episode earlier this year, they didn't even think they needed to make that many enough mirrors to make a hundred EUV tools a year. Um, and, and so now they're, they've like sort of, they're like, okay, we need to do that. But in reality, you know, because of all the economics of what's going on, it should be even more. Um, but it takes so long to, to.
Suppose, suppose that every single company in the firm, sorry, in the stack got private equity. Like somebody came in who was super AGI pilled and was like, we're going to maximize production. How fast is it? What do you think the physical constraints on making more things would be? Because the reason it asks is we're pretty soon going to be in a world where the lab revenue or just AI cash flows, because obviously the accelerators also have these huge cash flows, will be so big that you can just fund extreme expansion of all this production from cash flows themselves.
Yeah, I, I, I do agree generally. There's obviously some physical constraints. Um, the way the supply chain is expanding currently, the hundred is roughly still the right number.
For 2030?
A hundred ASML tools for 2030. But, you know, if you said, Carl Zeiss, here's 10 billion dollars, please fucking just expand production, that would change things. And you would have to do this with every company in the supply chain. I don't think it'll happen this year. I don't think it'll happen next year. I don't have it. I don't think it'll happen the year after because the world is capital constrained.
But in a world where, say, the top labs are generating, let's say even combined a trillion dollars in revenue next year, they're not able to say 10.
I don't think they're going to do that, but.
Yeah, well, or hundreds of billions at least, right? It just seems like they realize where the world is headed. I feel like they could just make.
So, so the thing is the labs can spend hundreds of billion. They're going to generate hundreds of billions of revenue next year, but ultimately CapEx next year is like two trillion dollars. So you've got this big mismatch, right? Um, you know, the, the wafer fabrication equipment supply chain will do, you know, two, you know, something on the order of 200 billion dollars. The data center market supply chain will do even more. The accelerator supply chain will do even more. You know, the energy supply chain will do a number. You know, you, you sum all this up, it's going to be, you know, well north of two trillion dollars of CapEx.
So the labs have not yet gotten to the point where their cash flows can fund this stuff.
Of course. Yeah, yeah, yeah. I mean, obviously they will like never get to that point, right? Because they want to keep reinvest, you want to keep, make your CapEx higher than your returns. But the key question I really want to understand is if, um, yeah, if the current content continues to be like north of 50 gigawatts per lab by the end of 2028. So between them, they'd have a hundred gigawatts. Um, those gigawatts, as you're saying, drive many fold more throughput or more performance by 2028 than they are now, right? Because the hardware has gotten better. So not only have like flops for what increase, but also the hardware gets better at, um, working with AI workloads.
Okay. So a hundred gigawatts for the labs end of 2028. How much is like world compute?
I think that may be a little difficult given 2028. You start to have they've taken 70, 80 percent of incremental compute and I'm not sure what happens to markets then, right? You know, how much does the price of compute skyrocket for them to actually be able to buy 70, 80 percent of compute is, you know, Google or Meta or Amazon willing to sell even that much. Um, also one caveat when we're sort of talking about these gigawatt numbers is, you know, when Amazon is serving Bedrock Anthropic models, that counts as Anthropic compute in sort of our worldview because it is effectively at the end of the day counted as revenue for Anthropic.
Um, even though like there's a revenue share and credit back and all that. Um, but ultimately in 2028, it's, it's, you know, if they get to a hundred gigawatts combined, they have done really disruptive things to the market because anyone can make money off of 10 to 15 million dollars per megawatt compute today. You literally, like I kid you not, it's not that hard. Go get a GB300 rack, go download the Kimi weights, go download VLM or SGLang, set it up. You know, Codex and Fable can actually help you do this.
It's pretty simple. I mean, it's not like it's, you know, it's not trivial, but it's not like rocket science and go put it on OpenRouter. It's very simple. Um, and, and you'll start generating more revenue than you're paying for the compute. Um, and so this is, this is sort of already led to this compute pricing 10 to 15 million dollars per megawatt start to inflect up. Um, and to get to that a hundred gigawatts in 2028, you have to believe that the labs can outpay for compute because anyone can make money at 10 to 15.
You, you know, does compute now get to 25 million dollars a megawatt? Does it get to 40 million dollars a megawatt?
But as you're saying, it's already the case that the labs are generating way more revenue per megawatt than everybody else. If they stay as far ahead as they are currently, you would expect that to be the continuing case. If there's like some kind of recursive self-improvement where the AI labs are like relatively uplifted where they have models internally, they're not releasing externally, they're helping them make the next model better, you'd expect that to be even more the case. And I don't think you're already seeing this where like SpaceX or whoever's like slightly further behind will just sell compute to the highest bidder if they can't internally monetize it as well as the labs. So I feel like you continue expecting them to be able to gobble up like bid for larger and larger shares of the compute.
I think that is my worldview that they will continue to gobble up more of the compute, but ultimately they can't do it at, at current pricing or anywhere close to it.
Sure, sure.
They, they do have to start paying 25, 30, 50 million dollars a megawatt to really gobble up 70 percent of the world's compute in 2028 to get to that a hundred gigawatts by 2028, which is a very sort of aggressive goal. The other aspect of this that's really challenging is we've already seen a huge slowdown for the AI labs, right? This, this regulation that they advocate for is actually slowing down the labs a lot more than it slows down, you know, sort of the open source Chinese language models. Um, you know, OpenAI not releasing, uh, Astra, OpenAI stopping training for two weeks, Anthropic not releasing what their safety assessment said is model two, which is widely believed to be the next version of Mythos.
They're clearly not releasing their best models, and in which case the revenue per megawatt stalls or even can start to decline again because other models are competitive again. Um, so it's not that they're falling behind, it's just that they're not releasing their best stuff. What if there is some regulatory impact that prevents them from releasing their best models? Now their revenue per megawatt does not climb as fast. Then their ability to buy that incremental compute for a higher price than everyone else starts to diminish. And then maybe they can't get to that a hundred gigawatts is, is sort of, in a world where safety doesn't matter.
I, I do believe that's exactly what happens, right? They can start generating a hundred million dollars per megawatt or more, and they can pay 50 million dollars a megawatt, and no one else has any logical reason to do anything with their compute besides say, please Dario, take everything off of my hands. Um, but there are, you know, forces at play, uh, that, which we cannot describe, um, that, that would potentially slow this down.
Yeah, yeah, yeah. I mean, I think a good intuition pump is just how, what if the AI models were literally as good as a fully automated software engineer? They're not currently there yet, right? Like I think they're far from just being able to fully automate the job of like a full white collar worker, but white collar workers earn, you know, six figures or north of that a year. Um, and if you have a gigawatt that can sustain a population of like, say, a million of white collar workers, just like, let's say roughly, right? Um, that's like, you could then off the back of that, that would be a hundred billion.
That's actually surprisingly low.
Yeah. A hundred K per person, million population. Yeah.
Yeah, yeah, yeah. Um, I don't know, but it would be many hundreds of billions of dollars if you get like full AGI, uh, per gigawatt.
I think the other aspect of this is, and we've continued to see this, the most of the value capture is not happening, right? Like most of the value that these models generate does not get given to OpenAI and Anthropic. Um, thankfully so far it is mostly just being given to the users.
Yeah, yeah.
Right. Jane Street with their exclusive contract with OpenAI for GPT 5.6 ultra fast mode or Jane Street where they're like one of Anthropic's biggest customers is generating way, way, way, way more value out of the tokens they're paying for than Anthropic is, uh, generating in terms of profit, right? Because they get to, you know, make money off of the market. Um, or Meta, who at one point was, you know, rumored to be, you know, as much as 10 percent of Anthropic's business. Um, you know, they're generating way more efficiencies by optimizing their ad algorithms or what have you and, and getting engagement time five percent longer and, you know, all these things.
They're, they're making way more money off of using these models than, than Anthropic. And so the ultimately, you know, and that's, that's what's required. So sure, if you had a million new software engineers, the cost per software engineer would also fall.
One thing I'm confused about is does the market come into equilibrium? And if it comes into equilibrium, would you just expect the price of compute to equal whatever Anthropic and OpenAI can generate from it or be very close to it with like a small amount of markup for Anthropic and OpenAI? Like right now it's really weird that there is a four X or more difference between what compute sells for and how much money Anthropic can make from it. And in the world where the revenue per gigawatt continues to increase, if Anthropic's ability to monetize a gigawatt doubles or triples or something, it'd be weird if then the gap continued to increase.
And so Anthropic just by like having some software, having some weights can take something that costs them $10 and then turn it, turn it into a hundred dollars.
Yeah. So there's, there's a bit of, um, there's always a fun question, right? Which is where does the value go in AI? AI is generating all this value. You've got, you know, the end user, which I think we all agree is generating more value than anyone else. Hence they're paying a lot for these models. But then you have, you know, the app layer. Well, so far the app layer has generated very little value. Um, and you've got the model layer, which again, up until, up until a year ago was generating negative gross margins and is now generating massive positive gross margins. Um, and looks like it's on the path to generating, you know, a hundred million dollars per megawatt.
Um, so turning, you know, 10, 15 dollars into a hundred dollars, as you said. Um, but if we go back again a year ago, the hardware supply chain was generating all this gross margin while literally everyone else was losing money on it. Um, OpenAI and Anthropic were just plowing VC money in. Um, and, and as were many other startups and, and many of these hyperscalers are building infrastructure without knowing if there was going to be a payoff. Um, so ultimately you had this like, you know, negative value being created on the model layer almost, if you will, um, because they were selling the tokens for less than it cost them on the infra side.
And all the values being created used at the chip, the fab initially in 2023, the memory guys were making no money off of, you know, HBM or memory for AI, even though theoretically their value they were delivering was humongous. Now you've got, well, actually TSMC makes way less value than the memory guys. Um, is that actually how much, you know, they're capturing less value, you know? So, so the, the value capture has shifted around a lot, which is very fun for, um, people tracking the market or participating in the market like, like Jane Street as an example.
There's sponsor.
You don't have to plug them that hard.
You don't have to plug them that hard.
Um, so, you know, what happens, you know, going forward? Does Anthropic and OpenAI, you know, they've, they've, they've slowly started to balloon in value capture. Do they balloon and take all the value capture? Well, that was the thought. And then, and then Elon showed, actually, no, I can sell my compute for 25 million dollars a megawatt or 40 million dollars a megawatt to Anthropic and Google. Um, even if it's a short-term thing, I've sold it for this price and I'll recoup my entire CapEx in a year.
So what's your prediction of how much the relevant tranche of compute, like B300s or whatever, that sold for 40 B a gigawatt, that SpaceX sold for 40 B a gigawatt to Google, what does that sell for at the end of next year?
I think most compute will still continue to transact at sub 20 billion dollars a gigawatt.
Even at the end of next year?
Because all of it has to be financed for compute that you can build without financing, right? If, if Meta can build compute, Microsoft, Amazon, SpaceX can build compute without finding a customer just saying, fuck it, I'm going to build this compute, and then turn around and wait till it's already built, they now control what's going on. So, so most compute is contracted well before it's built.
Yeah, yeah.
And so this is sort of what Elon took advantage of in the market is he actually had all this compute and he was like, hey, Anthropic, I know you're making like 60 plus billion dollars per gigawatt. Why don't you just buy my stuff for a crazy amount of money? And obviously, you know, it's not like Elon decided this or Anthropic decided this. It's sort of markets figured itself out. Um, other people, you know, you go to a random cloud, they're like, okay, I'm going to build a gigawatt of compute or a hundred megawatts of compute. I'm going to spend the CapEx. I need to turn around and find a customer. If I want to find a customer, I need to find the capital.
Who's going to give me the capital and the customer? The customer has to sign a deal. Then I take the customer's commitment to the credit markets and I raise the capital. And so there's this sort of like completely different power structure where Meta, who is effectively hoarding compute, them and SpaceX are plausibly the like number three. And the only plausible number three is because they're hoarding all this compute. They're using their balance sheets and capabilities to build compute, to build compute without an end customer that's monetizing it at a huge degree. And they have an actual balance sheet. So they can go to the credit market and being like, hey guys, I have, you know, you, you build a gigawatt, you can make, you know, your margin, not a crazy margin, but you can make a good margin.
And I now have all this compute. And now Meta and SpaceX have this optionality of looking around and being like, is my internal use case going to make me more money or should I go out there and sell it to Anthropic, OpenAI at crazy margins?
Yeah.
So now we've sort of entered a regime where, um, SpaceX and Meta are saying, actually, I'm going to build the compute and I can start to rent it out for not 13. I can sell it for 25, 50 and more.
If you want to try Grokbot yourself, go to x.ai/bot. What do you think the revenue per gigawatt is by the end of 2027? Like for Anthropic or OpenAI by the end of 2027?
I think, I think it's highly dependent on who has the best model, if they're allowed to keep releasing their best models, but I don't see why it wouldn't be 50 plus million dollars a megawatt.
By the end of '27.
Oh, by the end of '27.
Yeah.
Um, that's where it gets more challenging, but I think, I think it could get to, you know, higher than that. It's like 70, 80 million dollars a megawatt, uh, blended across a company, if not higher.
Yeah, yeah.
Yeah. And so I think if, if that's the case, right, then what happens to the price of compute? Well, if I'm Anthropic, incremental compute is worth it. Maybe I spend 40 million dollars a megawatt on SpaceX compute. Um, and if I'm SpaceX, you know, I, I look to the supply chain. I'm like, well, you know, I've struck this deal with Jensen where he's now all of a sudden using Twitter. Um, and, and, you know, there's Elon's saying they're exclusive to Nvidia, but why doesn't Jensen raise his prices? And then, you know, SK Hynix and Micron and Samsung looked at Nvidia and were like, well, why don't they raise their prices?
So I think, I think the value capture, there's a bull whip effect here, right? Where, uh, just because someone has risen the prices doesn't mean the entire supply chain rebalances immediately.
Yeah.
But over time, the supply chain will rebalance and things will cost more and more. And, you know, to get that incremental capacity, you sort of have to, right? So TSMC raising prices very slowly, but memory companies raising prices very quickly. You know, substrate companies raising prices very quickly. Different parts of the supply chain raise, raise, you know, Elon wouldn't have sold if it was 15, but he's selling because it's 25 plus. So obviously he rose his prices really quickly.
When, when does RSI happen? When does takeoff? Right. I think, I think.
Even if RSI doesn't happen, the current rate of progress continues. If you just look at how much progress have you made in, let's say, the last year and a half, like what was the model from a year and a half ago? Let's take like log 3.5.
I mean, my problem with this is the best model that exists in the world was trained in February.
Okay. You're saying maybe we just won't be allowed to release the next model.
Like OpenAI says they're not training models for two weeks, man. What the hell?
Yeah, yeah, yeah. I, I mean, there's another, there's one thing like internally, are they getting enough use for it as to like bid up the price of compute? Or another is like, does AI progress as a whole slow down because of regulation?
Yeah, but they're not even allowed to use this like new model internally. Like Astro's not widely deployed internally even.
Right, right, right. Yeah, yeah. But still, I don't know. It's just like the, the, yeah, if you go, if you have like a model that is, what was the model released? Like, let's say the beginning of last year, like GPT.
4.0? Is that 4.0?
Yeah. That, that's like, you're talking about a 4.0 to Fable size or from Mythos 2 size leap by this point. Uh, again, by the end of 2027.
Yeah, but Mythos 2 is not out.
Yeah. Or like even Mythos, right? Like that leap again.
Even Mythos is not allowed to be out, right? They've neutered it.
Yeah, yeah, yeah.
Like we can't, we can't use it to optimize inference performance or we can't use it to optimize all sorts of things.
Right. Yeah. Maybe there's like some slowdown in AI progress or the deployment of AI that means that the revenue per gigawatt can be lower, but I, that, that's the only way I could see it being only a hundred per megawatt by the end of next year.
Yeah. I mean, as, as long as the, the model gets better, the value generated out of it gets better. Obviously, who captures the value is still up for debate, but ultimately everyone's going to raise their prices.
Yeah, yeah, yeah.
Because they can. And it's super inflationary, especially if the method of regulation is right now, so far it's just don't release the models, but more and more the method of regulation is New York's banning data centers. Texas is holding memoratoriums. Ohio's saying you have to, or at least trying to say you have to like pay everyone's property tax in a certain radius. Um, these sorts of things are going to decrease supply and increase cost, and that's going to get passed on as well. So you start to end up in a spot where, um, progress does slow, at least in the external sense.
Yeah.
Um, even if the models internally keep getting better and better. I, I, I see no reason why, like, you know, again, like in a, in a takeoff scenario, why would Anthropic not have their best model six months ahead of what is externally available because of safety and regulation, but also, you know, the competitive advantage. Um, and then that six-month difference, if progress accelerates, is actually a bigger, um, differential. So, so that's the thing that would cap revenue per megawatt gains to a much lower growth than we've seen in the first half this year.
Here's something I'm very interested in. As these companies go public and they're accountable to investors and they, let's say at the end of next year, they have, I don't know, close to 20 gigawatts. So like 10% of the compute, two gigawatts. Let's say they want to go from 60% compute to, um, training to 70% compute to training. And their investors are like, well, if you're able to generate a hundred billion dollars per gigawatt, you're basically saying no to like 200 billion dollars of revenue in order to increase your training compute. As investors are like, what the fuck?
You're already spending so much on training. Why are you spending even more on training? As a public company, do you think, yeah, what, what do you think would happen if they're just like, no, we will keep increasing the share of compute we spent on training to offset the increase in revenue that each gigawatt of compute is giving us?
Yeah. So, so this is sort of what I, I personally believe that the labs are going to allocate less and less compute to inference over time, which I think is very non-consensus, right? Everyone's sort of, the standard belief of most people is, oh, most compute will go to inference. Um, most of it will go to forward passes for training, not maybe necessarily revenue generating inference, but ultimately you end up with, if they're generating, you know, 30, 40 million dollars per megawatt today, you allocate 40% to inference. If you now get to generating 60, 70 million dollars per megawatt, do you still allocate 40% to inference and generate all this profit and then do dividends and share buybacks, or do you go build AGI?
Yeah, yeah.
And I think the obvious answer from Anthropic and OpenAI, and not just at the executive level, but also their board is go build AGI because it's way more profitable.
Right.
Um, and, and so ultimately you're going to see them ratchet up their percentage of compute dedicated to training.
While each increment of compute is getting more and more profit generating if they had dedicated to inference.
Right. And, and so the, the whole point is, well, okay, if I'm generating, if I'm selling tokens, is Anthropic, is OpenAI releasing ultra-fast mode for just external or are they doing it internally too? And it turns out, no, actually I'm going to allocate it to internal and external because my internal, you know, value that I'm generating from super-fast AI or the best AI model is way more than what someone externally is.
Yeah, yeah.
And so ultimately, sure, I could generate a hundred million dollars per megawatt, but if I turn that towards AI research, what is the incremental progress that I get? And then what does that do towards my future earnings potential, the discounted cash flows of whatever the hell I've done, right?
Yeah.
Um, and so, you know, they're not going through that calculation, but ultimately it's, it, it makes more sense to dedicate more and more compute internally. And the only reason to, you know, have inference compute be so large is so you can grow your training fleet.
Right, right, right. I think this is an interesting economics question that I feel like we can have the models digest of what is the, at what, what would have to be true about a world where they reduce fraction of compute spent on inference?
I think they have been over the last three months already.
Interesting.
I think, I think at parts of this year they were increasing fraction of compute. So let's, let's just take month by month. You would agree that every month Anthropic has added more compute than the prior month. There might be some noise when they like sign a SpaceX deal or whatever, but in general, the amount of compute is, is a curve up. And so in January they added less compute than December. And yet their revenue ads, you know, skyrocketed and then they've flat, you know, sort of plateaued. They're only adding, you know, they're not adding 25 billion dollars of ARR every month now. And so that means the marginal megawatt they're getting is going higher percentage to R&D than it is inference.
Interesting. Yeah.
And so they are factually increasing their compute towards R&D today.
Yeah, yeah, yeah.
Yeah. I think this, this is like self-evident. Um, if you, if you like look at what they're doing enough.
Yeah. So I, if I look at the numbers you said of like how fast world compute grows, here's some things I want to understand. So it seems like if I added the numbers you just said, it would be over 200 gigawatts of world compute by the end of 2028, right?
Yeah, globally.
Okay. And, uh, how fast can that continue growing? Like global, global AI compute after 2028?
Yeah. So 30 this year, 50 next year, 70 in '28. Um, 29 should be like on the order of 90 to a hundred.
Like then just a hundred more every single year or something.
I think, I think the, the slope could continue to go upwards. I mean, it's hard to predict anything more than four years out given, uh, who knows what, what's, you know, are we in RSI regime or, you know, when is the world economy growing at 10% a year?
Yeah.
Uh, 'cause if you're at a hundred plus gigawatts a year, you're, you're at absurd revenue, uh, uh, GDP growth.
Right. If you think there's 200 gigawatts globally in 2028, how much is in China by that point? And how does like, yeah, how does Chinese compute continue increasing through this whole trend? Because if, if the RSI stuff kicks off in the West before China has a large amount of compute, maybe we're living in a different world than when it doesn't.
Yeah. So China today, um, so if we, if we sort of level set back to 2022, the US was adding about 45 to 50% of the world's compute. China was adding about 30 to 35% of the world's compute and the rest being taken up by the rest of the world. Since 2022, we've had big regulations against China, um, and a dramatic increase in America. So today, 70% of watts are being deployed in America. And, and, and, you know, China is, is really, uh, a very small number.
It's sub 10% of watts being deployed for data center AI compute is in China. Um, and as we step forward, they're still at a very small number. Um, their domestic production is quite small. Their purchasing from Nvidia is still quite small. Um, and a lot of that ends up in, in other places as well, right? You know, Malaysia or what have you. Um, so ultimately China domestically still continues to have 10%, sub 10% of incremental new compute. So in 2028 might start to inflect up, I think.
Um, but it's pretty easy to say China will have like 30 gigawatts of AI compute or less.
By 2028?
Yeah, in 2028.
Okay. And then how fast does their hockey stick go up?
I, I do think in 2028 they have a big uplift in what compute they're able to deploy. 2026, they're still mostly relying on a lot of the smuggled chips. Um, you know, you know, a lot of the chips that, uh, TSMC made, uh, for companies that they thought weren't Huawei, but ended up being Huawei or a lot of HBM that Samsung is shipping, you know, sort of. But in '27, fabs start to go up. In '28, especially fabs start to go up from SMIC and CXMT and such where domestic production is actually reaching many millions of units a year.
And now they're incrementally adding, you know, five, 10 gigawatts in just 2028 of domestically produced chips. Those chips are definitely worse than the chips that Nvidia will have in '28 or Google will have in '28 or OpenAI will have in 2028.
So even, even the gigawatt number overstates things you're saying. Like it's like 30 gigawatts, but it's really much worse chips. But then how, yeah, how does it, if you think the world is going to add a hundred gigawatts the following year or something, you know, I know you said you can't really say that far out. How much is China able to add the subsequent year? Basically, I want to know, did they just hockey stick at the point at which they are able to start shipping large amounts of compute or is it still going to be less than US plus allies?
Um, there's a lot left to, you know, whether or not the US passes the MATCH Act, um, whether or not tools continue to get export controlled, how fast China can build their new equipment that they're starting to be able to domestic, uh, produce, uh, domestically. Um, but ultimately, you know, China, China is definitely going to hockey stick if there's anything, uh, China's really good at is, is scaling manufacturing really, really quickly. Um, and, you know, I imagine, you know, China's, China will start to be able to extract more and more purchasing of even foreign chips, um, into, into domestic China or at least, uh, close the gap in what the US is allowing, uh, you know, Nvidia to sell them or what have you.
But do you think China could do adding 50 gigawatts by 2029? Marginal incremental gigawatts in 2029.
I think that's, I think that's completely reasonable.
Yeah.
Um, and part of that could also be purchased from foreign.
Yeah.
Um, but yeah, I think it's completely reasonable that China in 2029 could do 50 gigs. Um, but if most of those are domestic chips, there is some factor there where that 50 gigawatts is really worth as much as 20 gigawatts in America or in, in from, from American chips.
Right, right, right. So yeah, you're, you're actually projecting a world where maybe the leading lab in 2028 has more compute that China will have and like all of China will have in '29 or even '30 if you, if you weighed gigawatts by their quality.
Implying that there's nothing done to slow down the US labs.
Yeah, that's right.
Like clearly the government is starting and politicians are starting to do that.
Yeah, yeah, yeah.
Whereas China's not going to slow down AI. In fact, the only thing they're going to do is accelerate it.
So honestly, I, when Andrew Jensen asked about export controls, I am a libertarian person and I'm like, I wasn't like genuinely sure what I thought about this issue. I was steel manning what I, what is like the opposite view that he has and because I think it's important to hash out ideas. Um, but I'm like, yeah, maybe there's a world where if you just cooperated with China, it would be better for us, especially since they control so much of the supply chain and the other things that will be needed for robotics and other things. Um, but I didn't realize the compute situation was as fucked as you're saying.
Like actually the export controls do seem to have like really, if they ship the amount that you're saying, that's a huge difference. By the time we have automated coder and even getting into like automated researcher, China is like way far behind on the compute stock. And so if, if that ends up being the case, that would have worked. I think that's actually a notable success.
Um, I would say the only caveat there is some of it is export controls, um, but some of it is also just, uh, financial systems, right? American financial systems are more willing to YOLO and to startups than Chinese financial systems. But once Chinese financial systems choose an industry to focus on, they'll subsidize it a hell of a lot more.
Yeah.
Um, and so the Chinese semiconductor industry has significantly more subsidies than the rest of the world's semiconductor industries combined. Um, which, which points to like, you know, if takeoff is not as fast as sort of you're implying, but actually it takes longer, then ultimately China will catch up drastically on the semiconductor side, which then is compute at some point. The other aspect of this that I would think is like, I think is like noteworthy is, um, Chinese companies today are not that far behind in AI models, at least perceptively by the public, uh, relative to the amount of compute they have, right?
The leading Chinese labs have a hundred, 200 megawatts total of compute at most. Um, ByteDance, uh, Seed, Seed being the one outlier where they have significantly more than that. Um, but, you know, Kimi is not running, you know, a gigawatt or anywhere close to it.
Yeah.
Um, whereas Anthropic is, you know, nearly five gigawatts by the end of the year, right? Or more, sorry. Um, and so, you know, the question is sort of, well, does it matter? And I think right now it doesn't matter that much, the, this difference in compute, because, you know, when we break down the compute ratio or budget of a lab, historically it's, um, been, you know, let's say 60 or so far it's been like 60% training, 40% inference, but that training gets broken down further. And that is actually like 50% of the compute is research, like 10% of the compute is development, and then 40% is inference.
And what I mean by research and development is, you know, researchers are generating ideas, testing new architectures, testing new data mixes, testing new hyperparameters, whatever it is they're doing, new attention techniques, blah, blah, blah. But ultimately, when they do the training run, when Anthropic trains Mithos, it's sub 200 megawatts, right?
The pretrain of the whole thing.
The pretrain.
Yeah.
Um, it's sub 200 megawatts for, call it two months. Um, and then the RL is even less.
But you think the RL is less the compute than the pretrain?
Uh, at least in terms of, uh, single side inference. I mean, single side of pretraining. Yeah.
But total compute was probably higher, right?
Total compute, but it's like sequential, right?
Yeah.
So at most, the most they ever used at one point in time was maybe 200 megawatts.
Mm-hmm.
And then in reality, they had multiple gigawatts. So most of their compute was going to the research, not the development of a model. And there's reasons for this, right? You can't, it's hard to coordinate all these clusters. It's hard to, um, colocate all of them. It's hard to do multi-site training. It's hard to do, um, RL, you know, generating even more rollouts during RL does not necessarily make it better. There's all sorts of reasons why you may not be able to leverage, you know, all two gigawatts that you have for training onto training, right?
Actually, I can only leverage 200 megawatts. As we get closer and as we get further and further down implementation, automated coding, automated researcher, I actually expect the percentage of the compute budget that goes to research versus training to start to, and, uh, to start to like become a lot more fuzzy or even higher for training. Um, also things like continual learning, right? All of these things start to mean that more and more is actually going to training the model.
Okay. So if you have, if you end up in a world where you're doing a hundred gigawatts a year, at current prices, that would be 5 trillion of CapEx every single year.
And then stack on the fact that you have to build the power plants way before then. Slash it's a 30-year asset. You stack on the fact that the data centers are a, you know, 15, 20-year asset and you have to build that then too. So the 5 trillion, you know, you have to account for future years' growth. So it's actually going to be more like seven or 10 trillion.
Huh.
Of CapEx.
I didn't understand because you're not including the fact that like, that doesn't include the fact that there's not the infrastructure for the power, uh, generation or whatever in the data center itself.
Right. Exactly.
Yeah, yeah.
Um, and, and the data center itself is, um, when, when you talk about AI CapEx, people are saying 40, 50 billion dollars, but that's really just the critical IT.
Yeah.
Right? The servers, the networking, the fiber, the transceivers, optical communications, all this sort of stuff. Um, it doesn't account for the data center itself or the power plants themselves.
Yeah.
Which are being built ahead of time.
Yeah, yeah, yeah.
Um, if I'm building a hundred gigawatts this year and 150 gigawatts next year, well then all of the buildings for that 150 gigawatts need to be built and CapEx this year.
I see.
And if I'm building 200 gigawatts the year after that, all those power plants need to be spent. You have to buy the turbines this year.
Yeah, yeah, yeah.
Right? And so you've got this like lat, this like actually it's much bigger than even 5 trillion dollars if you're building, uh, a hundred gigawatts.
Right. So very plausibly incremental CapEx every year is getting close to 10 trillion dollars.
By the end of the decade.
Right. Which is going to be like close to a tenth of the world economy and like a third of, if all of it's going up in the US, it's like, well, your US economy will have grown as well, but still at the current size of the US economy, it'll be like a third to a quarter of the US economy, which is going towards data centers. And as I say that out loud, I'm like, maybe, maybe you're right and we just won't allow it. And that's the reason this doesn't happen, right? Because like for this exponential continue, just like a quarter of the world, a quarter of America's economy is just building data centers.
Um, yeah. I mean, I believe in capitalism and reallocation of resources towards the most profitable thing, but at the same time, politics exist.
Yeah, yeah.
Um, and credit markets exist and capital markets exist. So to enable, let's say that a hundred gigawatts by 2030, or let's even like, let's even like pare it down to 2028 where it's like three or four trillion dollars of CapEx across all of these items. Um, you know, a couple, you know, over, you know, two and a half towards compute, uh, IT CapEx and then another one to two on data center and energy, um, and all the supply chain downstream, like semiconductors and all that stuff. So if you're at, if you're at three or four trillion dollars of CapEx, where does all this cash come from?
No one is generating that much cash from the business yet, right? Um, hyperscalers, they funded all of the growth up until now. Google, Microsoft, Amazon, uh, Meta, they funded a huge percentage of it. They were more than half of compute, but they now don't generate cash. They actually spend everything on CapEx. And in addition, they raise debt and spend everything on CapEx, right? You've seen Meta do it, even Amazon, even Google, um, you know, Microsoft will be there soon. Everyone is raising debt to pay for their CapEx.
So now who is the incremental person to pay for this? Um, that was not doing it before. In the case of like Google, it was pretty simple for them to stop doing buybacks or Meta stop doing buybacks and turn around and buy computer infrastructure. And that doesn't have a huge effect on the market, um, but it does have some effect. But as you, as you step forward to 2028 where the hyperscalers are now raising hundreds of billions of dollars of debt, um, and then all of their supply chain is raising hundreds of billions of dollars of debt, who pays for this? And so there's a few different ways. Um, you know, there's the semiconductor companies like Nvidia, uh, and Broadcom and the memory companies turning around and deciding to fund some of this CapEx.
There's the traditional infrastructure investors who are turning around and gathering capital and investing in infrastructure. And instead of bridges, it's data centers. Um, and then lastly, there's everyone in the economy who's realizing maybe I shouldn't buy a home or maybe I shouldn't invest in credit for a home that's helping people buy homes, or maybe I shouldn't buy government debt. I should just buy hyperscaler debt, or I should just buy this data center's debt, or I should buy Anthropic's debt because Anthropic's willing to pay 20%, uh, you know, rates for, you know, the incremental billion dollars to build their capacity because they know their revenue from it's going to be huge and they're going to pay 20% because it's still better than renting it from SpaceX for 50 billion dollars a gigawatt.
Yeah.
Um, so you, you've got all of this contention, but if you now do this, the whole world economy is like really shifted around.
But with Antithesis, the original timeline is always reproducible, so you can test as many hypotheses as you need. And if you don't want to do all this time traveling yourself, you can just have your agents do it for you via the Antithesis API. Go to antithesis.com/dwarkesh to learn more. So you and I have been debating, uh, off air for the last few days whether there will be a sovereign debt crisis as a result of AI. And the logic is this, um, AI is, you, you have a situation where, as we were mentioning, very little investment turns into a lot of money, right?
So the rate of return.
What a fucking problem, dude.
Yeah.
Can't believe it.
No, it is a huge problem for everybody else who can't turn a little money into a lot of money, right? Um, so the rate of return is incredibly high. Even at the data center level, you know, if you build, if you like build a data center and you're like getting rented out to Anthropic or OpenAI for like 10X what it costs you on a depreciated basis to build it. It's fucking crazy. Um, and so you turn one dollar into like $2 or $10 or something at the end of the year. That raises the rate of interest higher. Now, if the rate of interest goes higher and if it does that for the entire economy and people are borrowing more and more money, they're competing against the other lending that the government would have done or that other companies would have done or that you as a consumer or a mortgage buyer would have done, uh, then that's just making it basically more expensive for everybody else to borrow.
This has huge implications for tons and tons of people. Sorry, I'm going to go on a bit of a monologue here. Um, but we, we, we, we, we've been thinking about this together. So I think the US will be fine at the end of the day because they can, if the data centers are built in America, you can fundamentally just like tax the data centers. But the way the current tax system is set up, you know, corporate income is like less than 10% of federal revenues and 80% plus is payroll taxes and income taxes, which as more and more automation happens, will shrink.
Um, at the same time, on the spending side, currently 20% of tax revenue spending goes towards paying, servicing the debt, basically paying interest payments on the debt. Um, now a lot of the debt is short duration, so refurbishes every five years it rolls over. Why are you fucking laughing?
Because, you know, it's like things we've learned, you've learned in the last.
Yeah. Like it's any different for you? Like you got a degree in fucking financial economics?
I did. I did. The internet thinks I'm a beekeeper. A few months, few months, few months.
Yeah. Um, this is our business, Dylan.
I know, I know, I know. Sorry, sorry.
Um, and, uh, so.
Now I'm self-conscious. Fuck.
No, it's good. You're doing good. I just think it's funny. Million people, listen to this guy who just learned about debt this month. Um, so you go from 20% of, is the supposed interest rates rise, uh, 1%, then the, um, over a five-year basis, the amount of the fraction of tax revenue that goes towards servicing the debt basically goes from 20% to 25%. If there were five percentages, that would go towards like north of 40%. But if you take into account the fact that the government is borrowing $2 trillion every single year, then that goes from like 40% to like north of 60%.
So 60% of tax revenue basically just goes towards paying interest payments on the debt. Now, I think the US is going to be fine because also the tax base will increase if we let data centers get built in America. Um, other countries are absolutely fucked in my opinion. I was just looking at which countries have a lot of debt, have very little tax revenue, and also a lot of their debt is serviced quite often. And those countries like Pakistan or Nigeria or something, I think are just going to be very fucked in this new interest rate regime.
So, so, so this, this, this crowding out effect is actually like the thing that I've like, is the reason it's not like YOLO one billion gigawatts.
Yeah, yeah.
Right? Um, you've got, you've got all these industries and countries that use a lot of debt, whether it's, you know, all these impoverished countries that you mentioned earlier that are just going to default. You've got like consumer packaged goods, right? Like all of these like companies that make things, you see at Trader Joe's or wherever, use a lot of debt. All these telecom companies use a lot of debt and banks use a lot of debt. And so if interest rates go up, um, in the market, not necessarily the government set interest rate, but in the market, the spread of interest rate, uh, between what the government says their federal rate is versus what everyone else is charging because Amazon wants to raise a hundred billion dollars of debt next year or whatever the hell the number is.
Um, you know, probably less, but, um, you end up with this like really challenging problem of where does the cash come from? Um, there is some level that is funded by cash flows and cash flows keep going up, but the logical thing to do is to invest way more than your cash flows because then the returns in the future years will be amazing. So you have this delta, um, and then what's pushing down on the delta is all of these other things, right? There's regulations against data centers, regulations, uh, you know, consumers getting mad, uh, politicians getting mad, regulations against AI, um, the AI labs not releasing their latest models because of safety reasons.
Um, all of these things and, and, and interest rates going up, um, are, are, are an influence on all of these things. So all of these things bend the curve from what does capitalism want in terms of just pure simple economics to what does the complex system that we have want and bends it lower and lower and lower to where not as many gigawatts as should be built will be built.
Well, the interest rate is part of capitalism, right?
Yeah, but like, you know, like in the simple economic model versus like the more complex what we have.
Yeah. Um, what is the rate at which you think Amazon or Anthropic or whatever will be releasing bonds for debt next year? They do hundreds of billions of dollars of debt. What is the rate? What is the average rate?
I, I don't think Amazon will do hundreds of billions of dollars of debt.
Total. Let's say the big.
The hyperscalers in total will raise and, and all the clouds.
Yeah, yeah.
In, in the modeling that we do, we have about $11 trillion of CapEx from 2024 to 2029.
Total.
Total. And if you do, you know, if you fund a lot of this with cash flows and as much as you can, you still end up with north of $5 trillion of credit that needs to be issued for this $11 trillion plus buildup.
You don't think the AI revenue continues even 3Xing year over year?
AI revenue does go up. I don't think it can go up forever. I don't, you know, like just like without like certain constraints being hit, I think certain labs will have certain incentives. Um, and, and labs are not the ones building all the compute in many cases, even though they're increasingly trying to go that way.
They'll have all these cash flows. Like if their revenue keeps increasing, whatever, that's fine. But if you, how much did you say their revenue will be? You think they'll not have that much revenue?
No, I'm just saying till 2029, there's, you know, something on the order of $11 trillion of CapEx and six of that is funded with cash and five of that is funded with debt. Um, and if that's the case, $5 trillion of debt being raised across the whole ecosystem does make interest rates go up. And then what prevents that? You know, there's a couple of things. One, do labs increase their revenue per megawatt more and keep inference allocations large, in which case they're taking all this profit, they're accumulating all the profit across the S&P 500 because everyone's paying to, you know, reduce their costs.
Um, of course their profits will also go up, but, um, you know, cash has to come from somewhere. Um, so there's an upper limit on how fast the revenue can grow versus the value they deliver into the world. And there's a diffusion aspect of the technology. Um, but ultimately labs revenue keep going up. They can't cash flow fund everything. The optimal scenario is you actually use credit as much as you can to fund, because even if cash flows from the labs fund a lot of stuff, you want to build more than that.
Of course.
Um, and so there is some amount of credit that gets built. Our current modeling has $5 trillion of credit and $6 trillion of cash funded infrastructure investments through '29. Um, and when you take that, you, you've sort of got, you know, this is not enough compute relative to what the demand growth is from the AI models. And so you've got the obvious answer, which is revenue per megawatt keeps going up.
Yeah, that, that, that makes sense. So how much do you think interest rates will increase by 2029 as a result of all this?
Dude, you know, I was just vibing a number, but if you're vibing a number out, you know, growth in the world and economy is growing up a lot. So why wouldn't interest rates.
Yep.
For Amazon go from, you know, from where they are today? Um, I think Meta paid, okay, let's like, so this is going to be extremely vibed out, but recently Meta's raised at like 5 to 6%. I don't see why they wouldn't pay 8% because they would happily pay 8% because the return from the compute that they're going to build is humongous.
Right.
Um, and the market won't want them to, but if they, they don't want to pay 8%. The flip side is if they pay 8% versus the five they do, five and a half, six they do today, you know, two 50 bips increase, that makes everyone else in the economy also pay two 50 bips more.
Yeah, yeah.
Which then causes a lot of things, right? Banks will scream because if their credit spread goes up, then their assets don't rep, their, their, uh, their debt themselves reprices faster than their assets reprice. Um, and you ultimately end up with they're losing tons of money if their credit spread blows up.
The other consequences of this are.
Yeah.
This is a point you made, but the, if interest rates rise, the discount rate increases, which means that the discount for cash flows of all equities.
Yeah.
Creator, which means that even though the stock market as a whole might be doing fine, like S&P 500 will be fine.
Yeah.
Any individual stock will probably have just like cratered in value, especially the, um, the Buffett, like Berkshire type, you know, pay good cash flows for 30 year type.
Yeah. It's like, it's like, why would I pay this much for, you know, um, you know, Johnson and Johnson.
Right.
Who, you know, like they're, they're seen as a stable stock, good cash flows, they'll return their cash flows over time, or a railway company, like why the fuck would I invest that much if my discount rate isn't 3% or 5%, it's now 8% or 10%.
And, and, and for developing countries, what's the, the, the Basel Hopperin, who's a good friend and he's an economist, he, um, made this point that we'll see a second Volcker shock. So in the eighties, to fight inflation, uh, Fed Chair Paul Volcker raised interest rates like more than 5%, um, or it was like something like 8%, uh, real interest rates 8%. And that caused some 40 different countries, mostly Latin America, to default in that decade. And I think that will probably happen again. In fact, okay, now we're getting into like singularity talk.
So we've been talking about what happens if interest rates.
I think this all happens before singularity.
Yeah, that's what I'm saying. That's what I'm saying. So we were talking about like, you know, before singularity, interest rates rise 2%, 3%, et cetera. At some point, I think it's very likely that the world economy will be doubling every single year. Okay, this is not happening in five years, but it'll happen eventually. It's just like there'll be, there's this, uh, there's a researcher, Damon Binder, who's done great work on this. But basically, if you look at like input output tables in a fully automated economy, just like what would it take to like double the entire stock of things in the economy ever.
Yeah, if economy grows at 3% a year, then it's like, you know, rule of 70, it's like 20 something years.
Right. But then, but he was like, okay, well, right now we're bottlenecked by the fact that there's people and you can't like double people every single year. But in a world where like you can also double labor force every single year, how fast can the economy grow? And I think it could double every single year or at the very least it would be like tens of percent every single year. Okay. The rate of interest should be pretty close to the growth rate. It won't be exactly that because of consumption, but it should, it should be pretty similar. So then we'll go into a world, I think in the 2030s, where the rate of interest is like tens of percent. Um, and like, I, I don't know, part of my brain is like it might be hundreds of percent, but like, okay, let's say it's at least tens of percent.
I'm just like, okay, every country that is not involved in the production of AI defaults, um, every stock that is not an AI stock is like worth basically zero because discounted cash flows are worth nothing. If the federal government can't figure out a way to tax AI, you know, the, the servicing the debt is more than the current, uh, tax revenue. Um, all these other effects that I'm sure we're not even pricing in, like you can't get a mortgage, et cetera, et cetera. Because fundamentally, what is happening in this world? Like this is all nerd speak, right? But like, let's step back. What's happening?
Just now it started the nerd speak.
We'd be entering a regime where just we're in a totally different growth regime, basically. And the economy's basically saying, hey, you like paying people, you borrow, the government borrowing money to pay people pensions, the opportunity cost of that is extremely high now because that money could be spent building a robot factory that builds a robot factory that builds a robot factory. And so the opportunity cost of capital is going to increase a ton. And that just like, that's fundamentally what the cause of all of these things we're talking about.
Yeah. So, so as interest rates go up, equity markets get pummeled.
Yeah.
Um, and even AI companies, right? People are like, you know, some people who really believe in AI are like, why does Micron or Hynix or Kioxia trade at two or three times earnings? And it's like, well, if you're really AI pilled, everything in the economy should trade at like two or three times earnings. Um, and if you're not AI pilled, then sure, they're, they're over earning.
Yeah.
Um, so it's sort of like a, an argument for why like I think memory is going to do great, but you know, memory, memory stocks shouldn't, you know, 10X or whatever again, because if they were, if we're in the market where there's that much demand for memory, which means AI's caused this drastic change in the economy, then everything should trade at like two or three X multiples and the stock market should fucking crash.
Right.
Right. And so in a sense, like Meta trading at, I don't know, I think Meta trades at like some, they're like $1.5 trillion company. It's like, what? Silly. Um, they're worth way more than that, at least in a, like a logical sense. You just look at their cash flows and, and like all the infrastructure they're hoarding and all the compute that they're going to be able to sell for crazy amounts of, uh, dollars per watt, either as tokens because their lab works or just Anthropic and OpenAI, um, ultimately becomes a question of like, you have to reallocate all the capital to the AGI and you do that by pricing everyone else out.
And, and, and so the limiter on AGI is not how fast can the research engineers, like, you know, our roommate Shoto can crank the gears. It's actually just like, how much does the rest of the world let that happen?
Right.
Because they're going to regulate, they're going to obviously increase interest rates. They're going to say no data centers. They're going to say stop building fabs. They're going to say, oh shit, every company's equity value is tanking. So how can I pay for AI, you know, to increase my business? Well, then, you know, like, okay, then Anthropic and OpenAI have to start like building their own stuff. And obviously they're going to eventually focus on, you know, they're building their own chips already or at least designing their own chips and it'll expand out their, you know, they're contracting their own data centers and building their own, uh, infra in the next couple of years.
Um, you know, there, there, there's sort of like, how does this reallocation of the economy happen? But there's a lot of downward pressure on it not being, you know, just straight takeoff. Um, even if the models were capable of it, uh, which, which I think you and I believe we're in a world where models are capable of that, but slow takeoff is, is, you know, at least my hope possible because everything in the economy and, and regulatory world, like governments saying don't release your models, governments saying actually you can't even use your models internally that much, because that's going to happen soon.
Um, they're already saying you can't release your models.
Which is actually the thing I'm most worried about is, you know, a singularity, which external deployment is, is actually helping, right? So the fact that we're preventing external deployment.
Well, does that, does that prevent singularity?
I mean, right now it would lead to more revenue because the models aren't capable of RSI.
Yeah.
But I'm worried about a world where it's 2030 and the government's like, we're going to wait six months before you can release your model.
US model to the public. Six months, 100X, let's go.
Yeah. And in that six months, they're just like, they do like recursive self-improvement internally. They just have all kinds of crazy shit happening in the company. Meanwhile, the rest of us are stuck with models that are like at current pace, years behind.
Yeah.
So here's my thought. Okay, suppose that the whole world gets in on this conspiracy to like try to slow down AI.
I don't think it's a conspiracy. It's like, it's outwardly written.
Yeah, yeah.
You know, from like.
Suppose you.
Every politician.
Suppose they basically prevent an entire, they slow down AI by a year. They, if compute is increasing two to three X every single year, they prevent a whole year of AI deployments such that you're a year behind where you would otherwise been. During, um, RSI, you're getting three to six years of AI progress in a single year.
Well, but, but they can't, they don't just limit compute, right? They also limit the lab's ability to release the model internally, right? We saw that Anthropic had to stop giving Mythos to foreign employees for a bit.
I didn't, I didn't know that was true. Like internally as well?
I mean, that's what they claimed. And they claimed.
I thought, I thought it was like just a different checkpoint that was not Mythos, but it was basically a Mythos.
Yeah, yeah, yeah. But I mean, like stuff like that is not going to be allowed either, right? Like the government is dumb, but they're not that dumb, right? Like, you know, I, I, I would hope at least, um, you know, governments are going to not want companies, at least the US government has the cards here. They're just not going to want Anthropic to use Mythos 4 internally. They're going to be like, hold the fuck on, right? Like slow down, you know, because, because all of these regulatory reasons, everyone who's elected is going to hate AI. Even the people who are elected already hate AI. Um, all the constituents, you're going to literally have like, I bet you at some point your parents are going to call you and be like, Duane Kishmita, you're doing a terrible job.
You're making every AI progress happen faster.
It's a podcast.
It's going to happen. It's going to fucking happen.
It's a podcast. I'm going to start hurting AI progress.
I mean, maybe you educate people, right? And, and maybe if they're smarter, they're progressing AI faster. Um, but anyways, like you, you're going to have real world constraints on the progress and development and deployment of AI, even though, you know, it will happen eventually. It's like we're, we could tear ourselves apart before we get there.
So if like compute at the frontier, you know, in flop terms is growing four or five X a year. And further, the compute required to achieve the likelihood is like decreasing three X a year. So the compute at the frontier, basically the effective AI population size at the frontier labs is increasing 10X year over year. And so that doesn't really matter that much right now because AIs are not good enough to do full jobs or be like as autonomous as people in their capacity to do work or pull off schemes or whatever.
But if the current trend continues, you have a world where OpenAI goes from having, say, 10 million basically AI laborers this year to 100 million the next year to a billion the year after that. And then pretty soon, even if compute scaling slows down, it doesn't take many more years before each company individually has more labor equivalence than there are people on earth. Um, and I think that's like a thing that is very plausible by the end of this decade, that there's more AI labor, more effective population within a single lab, um, than there are people on earth.
So we talk often about centralization of power because of nationalization or whatever, but we don't think enough about the fact that we're actually moving very fast into a regime where most AI labor, or sorry, most, most people, like in terms of like the work output or something, is just like concentrated within two labs who are consuming more and more of the world's compute. And so if, if these AIs are misaligned, then most of the world is misaligned basically, because like most of the world's minds are there. Um, but even if they're not, it just, very few companies have like a lot of influence or a lot of control.
Yeah, it's, it's sort of, um, there was the whole spat recently where it's like, um, I think Gavin Baker was like, Anthropic, Dario believes that there's only going to be one company in the world. And then, you know, Shoto and Dario came out and were like, no, no, no, we didn't say that. Um, but ultimately, you know, if you believe in RSI, you believe in the labs are the most effective, uh, user of compute and can generate the most, um, value from the compute, then the only thing that's going to happen is centralization of compute. And if you believe in, you know, sort of AI researchers, RSI, AGI, then all of this exists.
All of this is the base.
This is even true if there's no RSI. The current effective, like effective population at the frontier is currently increasing 10X year over year for a given level of capabilities, right? So if you get to the level of capabilities, which is a human, um, a very competent remote worker or like a very competent software engineer or very competent researcher, that population of those would like 10X year over year at the current rate of current rate of capabilities.
And without RSI, then once you have RSI, it's even crazier.
Then it's like maybe growing like a hundred X a year or a thousand X a year or they're like intelligence is increasing, but the population isn't increasing or some mixture of the two, right?
Um, yeah, I mean, I guess, I guess like what world do you see, Dwarkesh, where everything is not centralized? Um, because it seems to me that every force is screeching towards centralization.
Right.
And that's scary as hell.
Yeah.
Um, I don't, I don't, you know, I would love for it not to be centralized completely. Um, but maybe that's, that's the whole point of a machine, a machine that loves grace, right? Is, is, it is everything and it makes our lives great.
Yeah, it's so hard to think about the future. Um, but I agree with you that I think the fundamental problem is that lab AI training has huge economies of scale because any effort you spend into training an AI for a specific skill or a specific set of knowledge gets amortized across billions of sessions or billions of users. Um, the further, so that's like one effect. The other effect is if you're slightly ahead in the AI race and compute is in shortage, you can charge a much higher markup because you can better economize this scarce resource.
So there's like two effects which are give more and more to the person who's like ahead in the AI race. There may be more, right? So if there's models that are learning from deployment and one model is like deployed much more widely than another one and it's getting much more like real world data.
Yeah, your point, your point is taken that like whether it's user deployment and continual learning, um, whether it's, uh, training, having these economies of scales, um, whether it's the incremental progress that the best AI model helps you to make the next AI model, um, RSI, all of these things.
Oh, I didn't even mention RSI.
All of these things point to centralization.
So I think one of the big intellectual projects, honestly, um, that, yeah, we should spend some time thinking about, uh, async or at least I'll spend some time thinking about is what is a vision of like a decentralized, broadly empowered future after AGI that takes these economies of scale seriously? The alternative vision is that the government controls it. And maybe you think that you can trust the government more because it's not a private corporation.
I don't trust the government and I don't trust Dario and I don't trust Sam.
Yeah, yeah. That's the problem, right? Um, but there's no, at least obviously, obviously it's like very easy to be wrong about the future and you don't anticipate a key effect or something that changes everything. But ex ante, it's very hard to see a reason why they're, or like how we avoid a scenario where we have to choose one person.
I mean, it's why capitalism worked, right? It's the decentralized decision making and decentralized power. And why super centralized capitalistic economies actually grew slower than super decentralized capitalist economies to some extent. You have to have rule of law and all this, but then AI flips all this on its head. And ultimately you're like, actually private ownership is probably not the most efficient economy and therefore it grows slower than an AI economy, which is centralized.
It's no private ownership, but it's like how many firms are really involved in this, this like the share of the economy that's not, what, like 5% of the economy or something like that in the US? Sorry, 1 trillion divided by 30, less than that. Sorry. But, um, yeah, maybe two, 2% of the economy right now. It's like Nvidia is a huge share of it and Anthropic and OpenAI and these hyperscalers and obviously there's other firms involved, but like a large share of the AI stuff is just happening from very few companies. So it's like it could be private property, but like very few companies are involved.
I mean, this is what the structure of the market is doing. So, you know, what, what can prevent it? I don't know. I don't, unless AI progress slows down, unless governments regulate the fuck out of it, this is all that happens, in which case, you know, we're headed for a world where either we have super concentration of resources and we pray that that one company gets everything right, or we have governments slow everything down and people slow everything down and, and you have a slowdown of progress somehow, hopefully. And, and there is a more of a balance of power. And even as we go towards an AGI, ASI, RSI, um, everything along the way will still lead to someone's going to allocate, going to capture more resources.
So, so it's kind of hard for a framework in which AI doesn't lead to super concentration.
Yeah.
Now, the one positive thing here is that today Anthropic does not capture most of the value. So we can talk all we want about, oh, you know, they went from $20 million per megawatt to $100 million a megawatt, but they're still paying 13, um, for a lot of the compute they're buying. But at the end of the day, the reason they've gone to $100 million per megawatt is because Gene Street is capturing $300 million per megawatt or $500 million per megawatt. Um, where Dwarkesh from researching his podcast and learning about credit is capturing, you know, how many dollars per megawatt?
Now, how much can you use? Tough.
Yeah, yeah, yeah.
Um, but you know, I think, I think that's the like one saving grace is that the rest of the economy maybe profits so much more from Anthropic.
No, no, but the whole logic you were laying out earlier of them reallocating inference to AI R&D, the whole logic of that is that the, uh, returns to labor inside AI labs is much higher than the return.
This is my cope.
Returns outside. Yeah.
This is my cope. I agree. In all scenarios of the world, you know, there's 80,000 worlds and only one of them Anthropic doesn't own the whole world is, is that, is that, you know, again, power concentrates because I don't want to send the tokens outside. They're more valuable inside. And so it's the same thing, right? Why would I let Gene Street, you know, make all this money off of these degenerate options traders?
Hey, there's, there's some odds there. Come on.
Jesus Christ. No, I think it's great. I think it's great. It's a good value for the world to make it an efficient market.
Yeah, yeah, yeah.
You know, Gene Street making all this money off of getting the world view correctly, making money off of degenerate options traders, whatever it is, you know, why would Anthropic allocate compute to that? If, if, if the end, you know, monetization that Gene Street has per megawatt is 200, so they're willing to pay Anthropic 100. Well, what if Anthropic can just generate hundreds of millions of dollars per megawatt by using that compute internally? And that's, that's what's happening.
Yeah, yeah, yeah. Well, on that somber note, um, I guess, I guess, uh, I guess we'll meet again when the RSI is officially kicked off.
You're not going to, you're not going to have me on your podcast again for like two months?
All right. Cool. Thanks, dude.
Had a lot of fun chatting again with my twin brother Dylan Patel.
We went through lab economics over the next few years - the shift from inference to training as RSI draws near; and how Anthropic and OpenAI are on track to control most of the world’s usable FLOPs within the next few years (because they can monetize compute better and thus outbid everyone).
And then we discuss whether the >$10T of total AI capex we’ll see by the end of the decade will cause a sovereign debt crisis, where hyperscaler debt raises interest rates, drives non-AI exposed countries into bankruptcy, and crashes non-AI equities.
One question we weren’t able to resolve is whether there’s anything that can counter all the forces barrelling towards centralization in this industry - the economies of scale in training, the scarcity of compute, and eventually continual learning and RSI.
Watch on YouTube; read the transcript.
Sponsors
* Grok Bot has been quite helpful with my search for a new editor. I created a recruiter bot and described the type of editor I was looking for. That bot then spun up a handful of subagents that combed through my emails and X DMs, read the end credits of various documentaries I like, and figured out who edits for some of my favorite YouTubers. It took all of those results, and then delivered me a shortlist of candidates that matched my criteria. Try Grok Bot for yourself at x.ai/bot
* Antithesis lets you add time travel to your software testing toolkit. Since the Antithesis platform is fully deterministic, everything that happens inside of it is perfectly reproducible. So if your software crashes, you can rewind to the exact right moment, freeze time, and investigate. Or you can test different hypotheses by perturbing the system: kill a node or disable a feature, see what happens, then reset the trajectory and try something else. Learn more at antithesis.com/dwarkesh
* Jane Street is hiring for two separate ML internships right now, one focused primarily on research and one focused on engineering. In both cases, interns are expected to contribute to real work, not contrived exercises: one common project is adapting a frontier LLM paper to financial markets, which tend to come with a ton of different gnarly challenges. Importantly, you don’t need any finance background to apply. 2027 applications are open now at janestreet.com/dwarkesh
Timestamps
(00:00:00) – Two labs will soon control most of the world’s compute
(00:07:01) – $6 billion in fab capex enables $1t+ of end revenue
(00:13:08) – Compute prices will rise if the labs outbid everyone
(00:18:22) – Which layer will capture most of the surplus?
(00:25:40) – What could slow down progress?
(00:29:43) – Labs are shifting compute from inference to R&D
(00:33:27) – China gets less than 10% of new compute, but its labs need less
(00:48:48) – Will AI cause a sovereign debt crisis?
(01:07:52) – Will the world’s future workforce belong to a few companies?
This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com