Leopold Aschenbrenner — 2027 AGI, China/US super-intelligence race, & the return of history

1. 0:00 – 21:20

   The trillion-dollar cluster and unhobbling

   1. LALeopold Aschenbrenner0:00

      What will be at stake will not just be cool products, but whether liberal democracy survives, whether the CCP survives, what the world order for the next century will be. The CCP is gonna have an all-out effort to, like, infiltrate American AI labs, billions of dollars, thousands of people. CCP is gonna try to out-build us. People don't realize, like, how intense state-level espionage can be. And we have, like, literal super intelligence on our cluster, making, like, Stuxnet at the Chinese data centers. You really think they'll be, like, a private company? And the government would be like, "Oh my God, what is going on?" I do think it is incredibly important that these clusters are in the United States. I mean, would you do the Manhattan Project in the UAE, right? 2023 was the sort of moment for me where it went from kind of AGI as a sort of theoretical abstract thing, you'd make the models to like, I see it, I feel it. I can see the cluster where it's trained on, like, the rough combination of algorithms, the people, like, how it's happening. And I think, you know, most of the world is not, you know, most of the people who feel it are, like, right here, you know?

   2. DPDwarkesh Patel0:47

      (laughs)

   3. LALeopold Aschenbrenner0:48

      Right? Uh...

   4. DPDwarkesh Patel0:49

      Okay. Today I'm chatting with my friend Leopold Aschenbrenner. He grew up in Germany, graduated valedictorian of Columbia when he was 19, and then he had a very interesting gap year, which we'll talk about, and then he was on the OpenAI Superalignment team. May it rest in peace. And now, he, uh, with some anchor investments from Patrick and John Collison and Daniel Gross and Nat Friedman is launching an investment firm. So Leopold, I know you're off to a s- slow start, but life is long and I wouldn't worry about it too much. You'll make up for it in due time. Um, but thanks for coming on the podcast.

   5. LALeopold Aschenbrenner1:26

      Thank you. You know, I, um, I first discovered your podcast when your best episode had, you know, like, a couple hundred views.

   6. DPDwarkesh Patel1:32

      (laughs)

   7. LALeopold Aschenbrenner1:33

      Uh, and so it's just been, it's been amazing to follow your trajectory, and it's a delight to be on.

   8. DPDwarkesh Patel1:36

      Yeah, yeah. Well, I, uh, I think, uh, in the Sholto and Trenton episode, I t- mentioned that a lot of the things I've learned about AI, I've learned from talking with them. And the third part of this triumvirate, probably the most significant in terms of the things that I've learned about AI, has been you. We've got all this stuff on the record now.

   9. LALeopold Aschenbrenner1:53

      (laughs) Great, great.

   10. DPDwarkesh Patel1:54

       Uh, okay, first thing I have to get on record. Tell me about the trillion-dollar cluster. But, but by the way, I should mention, so the context of this podcast is today there's, you're releasing a series called Situational Awareness. We're gonna get into it. First question about that is, tell me about the trillion-dollar cluster.

   11. LALeopold Aschenbrenner2:09

       Yeah. So, um, you know, unlike basically most things that have come out of Silicon Valley recently, you know, AI is kind of this industrial process. Um, you know, the next model doesn't just require, you know, some code. It's, it's, it's building a giant new cluster, you know? Now it's building giant new power plants. You know, pretty soon it's gonna be building giant new fabs. Um, and, you know, since ChatGPT, this kind of extraordinary sort of techno-capital acceleration has been set into motion. I mean, basically, you know, exactly a year ago today, uh, you know, NVIDIA had their first kind of blockbuster earnings call, right? Where it, like, went up 25% after hours and everyone was like, "Oh my God, AI, it's a thing." You know, I mean, I think, um, within a year, you know, you know, NVIDIA, NVIDIA data center revenue has gone from like, you know, a few billion a quarter to like, you know, 20, 25 billion a quarter now and, you know, continuing to go up, um, like, you know, big tech CapEx is skyrocketing. And, um, you know, it's funny because it's, it's both there's this sort of, this kind of crazy scramble going on, but in some sense, it's just the sort of continuation of straight lines on a graph, right? There's this kind of, like, long run trend, basically almost a decade of sort of training compute of the sort of largest AI systems growing by about, you know, half an order of magnitude, you know, 0.5 booms a year. Um, and you just kind of play that forward, right? So, you know, GPT-4, you know, rumored or reported to have finished pre-training in 2022. You know, the, the sort of cluster size there was rumored to be about, you know, 25,000 H100s, you know, s- sorry, A100s on semi-analysis. Um, you know, that's, that's roughly, you know, if you do the math on that, it's maybe like a $500 million cluster. You know, it's very roughly 10 megawatts. Um, and, um, you know, just play that forward, half a boom a year, right? So then 2024, that's a, you know, that's a cluster that's a, you know, 100 megawatts. That's like 100,000 H100 equivalents. You know, that's, um, um, uh, you know, costs in the s- billions. You know, play it forward, you know, two more years, 2026, that's a cluster that's a gigawatt. You know, that's a, that's a, you know, sort of a large nuclear reactor size. It's like the power of the Hoover Dam. You know, that costs tens of billions of dollars. That's like a million H100 equivalents. You know, 2028, that's a cluster that's 10 gigawatts, right? That's, that's more power than kind of like most US states. Um, that's, you know, like 10 million H100s equivalents, you know, costs hundreds of billions of dollars. And then 2030, um, trillion-dollar cluster, uh, 100 gigawatts, over 20% of US electricity production, you know, 100 million H100 equivalents. And, um, and that's just the training cluster, right? That's like the one largest training cluster, and then there's more inference GPUs as well, right? Most of, you know, once there's products, most of them are going to be inference GPUs. And so, you know, US power production has barely grown for, like, you know, decades, and, uh, now we're really in for a ride.

   12. DPDwarkesh Patel4:41

       So I mean, I, when I had Zuck on the podcast, he was claiming not a plateau per se, but that AI progress would be bottlenecked by specifically this constraint on energy, and specifically like, oh, gigawatt data centers are gonna build another Three Gorges Dam or something. I know that there's companies, according to public reports, who are planning things on the scale of a gigawatt data center. 10 gigawatt data center, who's gonna be able to build that? I mean, the 100 gigawatt center, like, w- uh, a state, where you're getting the, are you gonna pump that into one physical data center? How is this gonna be possible?

   13. LALeopold Aschenbrenner5:15

       Yeah. You know-

   14. DPDwarkesh Patel5:15

       What is Zuck missing? (laughs)

   15. NANarrator5:16

       (laughs)

   16. LALeopold Aschenbrenner5:16

       I mean, you know, I don't know. I think the 10 gigawatt, you know, like six months ago, you know, 10 gigawatts was the talk of the town. I mean, I think, I feel like now, you know, people have moved on. You know, 10 gigawatts is happening. I mean, I don't know. There was then the information report on, um, OpenAI and Microsoft planning a $100 billion cluster. So you, you gotta, you know, if you do the math-

   17. DPDwarkesh Patel5:31

       Is that a gigawatt or is that the 10 gigawatt?

   18. LALeopold Aschenbrenner5:33

       I mean, I don't know, but you know, if you try to, like, map out, you know-

   19. DPDwarkesh Patel5:36

       Sure.

   20. LALeopold Aschenbrenner5:36

       ... how expensive would the, would the 10 gigawatt cluster be? You know, that's maybe a couple hundred billion.

   21. DPDwarkesh Patel5:39

       Yeah.

   22. LALeopold Aschenbrenner5:39

       So it's sort of on that scale. Um, and they're planning it. They're working on it. You know, so, so, um, the, um, you know, it's not just sort of my crazy take. I mean, AMD, AMD, I think, forecasted a $400 billion AI accelerator market by '27. You know, I think, I think it's, you know, and AI accelerators are only part of the expenditures. Uh, it's sort of, you know-I think sort of a trillion dollars of sort of, like, total AI investment by 2027 is sort of, like... We're very much on track now. I think the trillion-dollar cluster is gonna take a bit more, sort of, acceleration. But, you know, um, we saw how much sort of ChatGPT unleashed, right? And so, like, every generation, you know, the models are gonna be kind of crazy, and people... it's gonna shift the Overton window. Um, and then, and then, you know, obviously the revenue comes in, right? So these are forward-looking investments. And the question is, do they pay off, right? And so if we sort of estimated the, um, you know, the, the GPT-4 cluster at around 500 million... Um, by the way, that's, that's sort of a common mistake people make, is they say, you know... People say, like, $100 million for GPT-4.

   23. DPDwarkesh Patel6:30

       Yeah.

   24. LALeopold Aschenbrenner6:30

       But that's just the rental price, right? They're like, "Ah, you rent the cluster for three months." But as you know, if you're building the biggest cluster, you got to, like, you got to build the whole cluster. You got to pay for the whole cluster. You can't just rent it for three months.

   25. DPDwarkesh Patel6:38

       But, but can't-

   26. LALeopold Aschenbrenner6:39

       But they... I mean, really, you know, once, once you're trying to get into the sort of hundreds of billions-

   27. DPDwarkesh Patel6:42

       Yeah.

   28. LALeopold Aschenbrenner6:42

       ...I mean, eventually you got to get to, like, $100 billion a year revenue.

   29. DPDwarkesh Patel6:44

       Okay.

   30. LALeopold Aschenbrenner6:44

       And I think this is where it gets really interesting for the big tech companies, right? Because, like, their revenues are on order, you know, hundreds of billions, right? So it's like $10 billion? Fine, you know, and it'll pay off the, you know, 2024-sized training cluster. Um, but, you know, really when sort of big tech, it'll be gangbusters, is 100 billion a year. And so the question is sort of, how feasible is 100 billion a year from AI revenue? And, um, you know, it's a lot more than right now, but I think, um, you know, if you sort of believe in the trajectory of the AI systems, as I do, and which we'll, we'll probably talk about, it's not that crazy, right? So there's... I think there's like 300 million, you know, ish Microsoft Office subscribers, right? And so they have Copilot now, and I don't know what they're selling it for, but, uh, you know. Suppose you sold some sort of AI add-on for 100 bucks a month, um, and you sold that to, you know, a third of Microsoft Office subscribers subscribed to that. That'd be 100 billion right there. You know, $100 a month is, you know, a lot.
2. 21:20 – 41:15

   AI 2028: The return of history

   1. LALeopold Aschenbrenner21:20

   2. DPDwarkesh Patel21:20

      Uh, I want to ask some questions about this. I think-

   3. LALeopold Aschenbrenner21:21

      We can tell you. Yeah, yeah, yeah.

   4. DPDwarkesh Patel21:22

      Yeah, le- le- let's zoom out. Okay.

   5. LALeopold Aschenbrenner21:23

      All right.

   6. DPDwarkesh Patel21:24

      So suppose you're right about this.

   7. LALeopold Aschenbrenner21:25

      Yeah.

   8. DPDwarkesh Patel21:25

      Um, and I guess you, this is because of the 2027 cluster, we've got 10 gigawatt, uh, 2027, 10 gigawatts, some, uh-

   9. LALeopold Aschenbrenner21:33

      '28 is the 10 gigawatt year.

   10. DPDwarkesh Patel21:35

       Okay, yeah.

   11. LALeopold Aschenbrenner21:35

       I mean, maybe it'll be pulled forward. Who knows, right?

   12. DPDwarkesh Patel21:37

       Okay, sure. Uh, something-

   13. LALeopold Aschenbrenner21:38

       Yeah.

   14. DPDwarkesh Patel21:38

       Uh, and so I guess that's like 5.5 level by 20- or 2027. Like, what- whatever that's called, right? Um, what does the world look like at that point? Uh, you have these remote workers who can replace people. What is the reaction to that in terms of the economy, politics, geopolitics?

   15. LALeopold Aschenbrenner21:55

       Yeah, so, you know, I think 2023 was kind of a really interesting year to experience as somebody who is like, you know, really following the AI stuff, where, you know, before that, it was just-

   16. DPDwarkesh Patel22:05

       Wh- what were you doing in 2023?

   17. LALeopold Aschenbrenner22:06

       I mean, OpenAI.

   18. DPDwarkesh Patel22:07

       Oh, okay.

   19. LALeopold Aschenbrenner22:07

       Yeah, yeah, yeah. And, and, uh, and, um, you know, it kind of went, you know, I mean, you know, I was, I was thinking about this and, you know, like talking to a lot of people, you know, in the years before, and it was this kind of weird thing. You know, you almost didn't want to talk about AI or AGI. You know, it was kind of a dirty word, right? And then 2023, you know, people saw ChatGPT for the first time, they saw GPT-4, and it just like exploded, right?

   20. DPDwarkesh Patel22:27

       Yeah.

   21. LALeopold Aschenbrenner22:27

       It triggered this kind of like, you know, uh, you know, huge sort of capital expenditures from all these firms, and, and, and, you know, the explosion in revenue from NVIDIA, and so on. And, um, you know, things have been quiet since then. But you know, the next thing has been in the oven. And I sort of expect sort of every generation these kind of like g-forces to intensify, right? It's like people see the models. Um, there's like, you know, people haven't counted them so they're going to be surprised and it'll be kind of crazy. And then, you know, revenue is gonna accelerate. You know, suppose you do hit the 10 billion, you know, end of this year. Suppose it like just continues on this sort of doubling trajectory of, you know, like every six months of revenue doubling. You know, it's like you're not actually that far from 100 billion. You know, maybe that's like 26. And so, you know, at some point, you know, like, you know, sort of what happened to NVIDIA is going to happen to big tech. You know, like their stocks, their, you know, that's going to explode. Um, and I mean, I think a lot more people are gonna feel it, right? I mean, I think the, um, I think 2023 was the sort of moment for me where it went from kind of AGI as a sort of theoretical abstract thing and you'd make the models to like, I see it, I feel it. And like I see the path. I see where it's going. I like, I think I can see the cluster where it's trained on, like the rough combination of algorithms, the people, like how it's happening.

   22. DPDwarkesh Patel23:35

       Yeah.

   23. LALeopold Aschenbrenner23:35

       And I think, you know, most of the world is not, you know, most of the people who feel it are like right here, you know?

   24. DPDwarkesh Patel23:38

       (laughs)

   25. LALeopold Aschenbrenner23:39

       Right? Uh, but, but, you know, I think a lot more of the world is going to start feeling it. Um, and I think that's going to start being kind of intense.

   26. DPDwarkesh Patel23:47

       Okay, so right now, who feels that you can, you go on Twitter and there's these GPT wrapper companies like, "Whoa, GPT-4.0 is going to change our business"?

   27. LALeopold Aschenbrenner23:54

       Yeah, well, I mean, I'm so, so bearish on the wrapper companies, right? Because like they, they're the ones that are gonna be, like the wrapper companies are betting on stagnation, right? The wrapper companies are betting like you have these intermediate models and it takes-

   28. DPDwarkesh Patel24:02

       Sure, sure, sure.

   29. LALeopold Aschenbrenner24:03

       ... so much work to integrate them. And I'm kind of like, I'm really bearish because I'm like, we're just gonna sonic boom you, you know? We're gonna get the unhobblings, we're gonna get the drop-in remote worker, and then, you know, your stuff is not gonna matter.

   30. DPDwarkesh Patel24:12

       Okay, sure, sure. So th- that's done. Now, um, who, so th- th- SF is paying attention now, or th- this crowd here is paying attention. Who is gonna be paying attention in 2026, 2027? And p- presumably, this is, these are years in which hundreds of billions of CapEx is being spent on AI.
3. 41:15 – 1:09:09

   Espionage & American AI superiority

   1. DPDwarkesh Patel41:15

   2. LALeopold Aschenbrenner41:15

      Yeah.

   3. DPDwarkesh Patel41:16

      Well, let's go back to the $100 billion revenue whatever, and so these companies-

   4. LALeopold Aschenbrenner41:20

      Trillion dollar cluster.

   5. DPDwarkesh Patel41:21

      Yeah, the companies are w- um, deploying, bu- trying to build clusters that are this big.

   6. LALeopold Aschenbrenner41:25

      Yeah.

   7. DPDwarkesh Patel41:26

      Where are they building it? 'Cause if, if you say it's the amount of energy that would be required for a small or medium-sized US state, is it then Colorado gets no power and it's happening in the United States? Or is it happening somewhere else?

   8. LALeopold Aschenbrenner41:35

      Oh, I mean, I think the, I mean, in some sense this is the thing that I always find funny is, you know, you're talking about Colorado gets no power. You know, the easy way to get the power would be like, you know, displace less economically useful stuff. You know, it's like whatever, buy up the aluminum smelting plant and, you know, that has a gigawatt and, you know, we're gonna replace it with, with the data center because that's important. Um, I mean, that's not actually happening because a lot of these power contracts are really sort of long-term locked in. You know, because obviously people don't like things like this. And so it sort of, it seems like in practice what it's, what it's requiring at least right now is building new power. The, um, that might change and I think that that's when things get really interesting when it's like, "No, we're just dedicating all of the power to- to the AGI." But anyway, so right now it's building new power. 10 gigawatt, I think quite doable. Um, you know, it's like a few percent of, like, US natural gas production. Um, um, you know, I mean, when you have the 10 gigawatt cl- training cluster, you have a lot more inference. So that starts getting more... You know, I think 100 gigawatt, that starts getting pretty wild. You know, that's, you know, again, it's like over 20% of US electricity production. Um, I think it's pretty doable, um, especially if you're willing to go for, like, natural gas. Um-

   9. DPDwarkesh Patel42:32

      But- but- but-

   10. LALeopold Aschenbrenner42:33

       I do, I do think, I do think it is incredibly important, incredibly important that these clusters are in the United States.

   11. DPDwarkesh Patel42:38

       And why does it matter it's in the US?

   12. LALeopold Aschenbrenner42:41

       Um, I mean, look, I think there's some people who are, you know, trying to build clusters elsewhere. And you know there's like a lot of free-flowing Middle Eastern money that's trying to build clusters elsewhere. Um, I think this comes back to the sort of like national security question we talked about earlier. Like, would you, I mean, would you do the Manhattan Project in the UAE, right? And I think, I think basically, like, putting, putting the clusters, you know, I think you can put them in the US, you can put them in sort of like allied democracies. But I think once you put them in kind of like, you know, dictatorships, authoritarian dictatorships, you kind of create this, you know, irreversible security risk, right? So I mean, one-... clusters there, much easier for them to ex-filtrate the waste. You know, they can, like, literally steal the AGI, the super intelligence. It's like they got a copy of the, uh, you know, of the, of the atomic bomb, you know, and they just got a direct replica of that. And it makes it much easier to them. I mean, w- we are ties to China. You can just ship that to China, so that's a huge risk. Another thing is they can just seize the compute, right? Like, maybe right now, they just think of this... I mean, in general, I think people, you know, I think the issue here is people are thinking of this as a, you know, ChatGPT, big tech product clusters. But I think the clusters being planned now, you know, three to five years out, like, they will be the, like, AGI super intelligence clusters. And so anyway, so like, when things get hot, you know, they might just seize the compute. And I don't know, supposedly if we put like, you know, 25% of the compute capacity in these sort of Middle Eastern dictatorships, well, they seize that, and now it's sort of a ratio of compute of three to one, and you know, still have some more. But even like, even, even only, only 25% of compute there, like, I think it starts getting pretty hairy, you know? I think three to one is, like, not that great of a ratio. You can do a lot, um, with that amount of compute. And then look, even, even if they don't actually do this, right? Even if they don't actually seize the compute, even if they actually don't steal the weights, um, there's just a lot of implicit leverage you get, right? They get, they get the seat at the AGI table. Um, and, um, you know, I don't know why we're giving authoritarian dictatorships the seat at the AGI table.

   13. DPDwarkesh Patel44:24

       Okay, so there's gonna be a lot of compute in the Middle East if these deals go through. Uh, first of all, who's, who, is it just like every single big tech company is just trying to figure out what-

   14. LALeopold Aschenbrenner44:33

       Not everyone. Some.

   15. DPDwarkesh Patel44:34

       ... okay, okay. Well, I guess there's reports, I think Microsoft are-

   16. LALeopold Aschenbrenner44:37

       Yeah, yeah, yeah, yeah.

   17. DPDwarkesh Patel44:38

       Um, uh, which we'll get into.

   18. LALeopold Aschenbrenner44:40

       Yeah.

   19. DPDwarkesh Patel44:40

       So they, UAE gets a bunch of compute because we're building the clusters there.

   20. LALeopold Aschenbrenner44:44

       Yeah.

   21. DPDwarkesh Patel44:45

       And why... So l- let's say they have 25% of the... Why, why does the compute ratio matter? Um, is it... If it's about them being able to kick off the intelligence explosion, isn't it just some threshold where you have 100 million AI researchers or you don't?

   22. LALeopold Aschenbrenner45:01

       I mean, you can do a lot with, you know, 33 million extremely smart scientists. Um, and you know, again, a lot of this stuff, you know, so first of all, it's like, you know, that might be enough to build the crazy bioweapons, right? And then you're in a situation where like, now wow, we've just like... They stole the weights, they seized the compute. Now they can make, you know, they, they can build these crazy new WMDs that, you know, will be possible with super intelligence. And now you've just kind of like proliferated the stuff and you know, it'll be really powerful. Um, um, and also, I mean, I think, you know, three to... Three acts on compute isn't actually that much and so the, um, you know, the, um, you know, I think a thing I worry a lot about is, I think everything... I think the riskiest situation is if we're in some sort of like really tight-neck feverish international struggle, right? If we're like really close with the CCP and we're like months apart. Um, I think the situation we want to be in, we could be in if we played our cards right, is a little bit more like, you know, the US, you know, building the atomic bomb versus the German project. Way behind, you know, years behind. Um, and if we have that, I think we just have so much more wiggle room, like to get safety right. We're gonna be building like, you know... There's gonna be these crazy new WMDs, you know, things that completely undermine, you know, nuclear deterrents, you know, um, intense competition. And, um, that's so much easier to deal with if, you know, you're like, you know, it's not just... You know, you don't have somebody right on your tails. You gotta go, go, go. You gotta go maximum speed. You have no wiggle room. Um, you're worried that at any time, they can overtake you. I mean, they can also just try to out-build you, right? Like, they can might, they might literally win. Like, China might literally win, um, if they can steal the weights 'cause they can, they can out-build you, um, and they maybe have less caution, uh, both, you know, good and bad caution. You know, kind of like whatever unreasonable regulations we have. Um, or you're just in this really tight race. And I think is... That sort of like... If you're in this really tight race, this sort of feverish struggle, I think that's when sort of there's the greatest peril of self-destruction.

   23. DPDwarkesh Patel46:47

       So then presumably the companies that are trying to build clusters in the Middle East-

   24. LALeopold Aschenbrenner46:50

       Yeah.

   25. DPDwarkesh Patel46:50

       ... realize this. What is, uh, is it just that it's impossible to do this in America, and if you want American companies to do this at all-

   26. LALeopold Aschenbrenner46:55

       Yeah, yeah.

   27. DPDwarkesh Patel46:56

       ... then you do it in Middle Eastern or not at all?

   28. LALeopold Aschenbrenner46:57

       Yeah.

   29. DPDwarkesh Patel46:57

       And then you just like have China build a Three Gorges Dam cluster.

   30. LALeopold Aschenbrenner47:00

       Yeah. I mean, there's a few reasons. One of them is just like people aren't thinking about this as the AGI super intelligence cluster. They're just like, "Ah, you know, like, cool clusters for my, you know, uh, for my ChatGPT."

Episode duration: 4:32:06