00:00:02 Speaker 1: Bloomberg Audio Studios. Podcasts. Radio. News. 00:00:08 Speaker 2: What are you reading, Matt Levine? 00:00:09 Speaker 1: You know, I'm reading Absalom, Absalom. It's a little much. 00:00:12 Speaker 2: Many people think that's one of the greatest novels to ever do it. 00:00:16 Speaker 1: I know. I know. 00:00:16 Speaker 2: Did you read it because of the Dwight Garner review of The Disappearers that mentioned it as one of the best? You did. 00:00:22 Speaker 1: Well played. Wow. From downtown. This is like such not money stuff content. Yes. Sylvia accurately identified why I am reading Absalom, Absalom. It's like a little much, man. It's a little much. Okay. 00:00:34 Speaker 2: Absalom, Absalom is your fiction. What's your non? 00:00:37 Speaker 1: My usual division is fiction, nonfiction, and math or finance. But then also, I usually try to have two paper books going and one book on my phone so I can read it while walking the dog. And so right now, my phone book is also fiction. It's Colson Whitehead's Sag Harbor. 00:00:50 Speaker 2: What drove you to that? Wistful feelings about the end of the summer? 00:00:54 Speaker 1: So yes. So well-identified, but no, there's a much more specific answer. 00:00:57 Speaker 2: Give me one more. 00:00:58 Speaker 1: Which is that I was In Montauk, Reading the East Hampton Star, which featured a write-up of a book event that Colson Whitehead did at, I don't know, the East Hampton Public Library. Right, for the new one. And it was just a good, he sounds fun, and so I was like, I have late summer nostalgia, so I bought Sag Harbor from my phone. Hello, and welcome to the Money Slip Podcast. I'm Matt Levine. I wrote the Money Slip column for Bloomberg Opinion. I'm joined by. 00:01:25 Speaker 2: I'm Sylvia Killingsworth. I edit the OnBooks newsletter. 00:01:29 Speaker 1: Sylvia is our returning guest co-host. 00:01:32 Speaker 2: Had so much fun. 00:01:34 Speaker 1: The last time. So in May of this year, China announced a crackdown on basically cross-border trading from its brokerages. So some Chinese brokerage firms, their stocks dropped a lot because they got in trouble and they were restricted from doing some of the trades that had made them a lot of money. And just before that happened, people bought a lot of puts on several Chinese brokerage stocks. And Susquehanna International Group, which is a big options market maker, sold a lot of those puts. And they went to court in the US in June and said, people who are buying these puts, we have statistical data that makes it look really suspicious that they bought these puts. There's no news. They just bought a lot of very short-dated. 00:02:27 Speaker 2: They must have known. 00:02:28 Speaker 1: Because they paid off a lot of money. 00:02:29 Speaker 2: Somehow. 00:02:29 Speaker 1: They must have known. They must have been. 00:02:31 Speaker 2: Somebody must have leaked somewhere. 00:02:32 Speaker 1: It guesses who they were. Right. Because like insider trading laws are weird. You need to have a fiduciary duty. That would be illegal. You need to be in breach of some duty to someone. And so they're like, they must have been either Chinese regulators or employees at these brokerage firms and had some duty to keep it confidential. And instead they traded on it. And so we want the money back. And they sued like 100 John Doe defendants. They don't know who they are. 00:02:53 Speaker 2: 100 Johns Doe. 00:02:54 Speaker 1: Yeah. they're like customers of a couple of like Chinese brokerages. And I think of interactive brokers that you asked for. And they basically like went to court. First of all, you get the money back, but secondly to like immediately get an injunction to like freeze these accounts. Yep. And this month they lost the preliminary injunction. Basically the judge was like, Well, the case can go forward maybe, like, it's fine, but, like, we're not going to freeze all these accounts because he. 00:03:19 Speaker 2: Just didn't... It's impossible to know which one's who's what. 00:03:21 Speaker 1: He just didn't, like, entirely believe that they were all insider trading. 00:03:24 Speaker 2: All 100 of them are holding hands and insider trading together. 00:03:26 Speaker 1: Right, right. 00:03:27 Speaker 2: So what was happening? If not all 100, some number of them, but the rest were kind of just following the pack. 00:03:33 Speaker 1: Yeah, it's unclear, right? Like, there could be no insider trading. Like, I think, like, the circumstantial evidence is reasonable that maybe someone was insider trading. 00:03:41 Speaker 2: But all 100 of them, eh. 100 is even kind of a fake number. Sure, but we're calling it 100. 00:03:46 Speaker 1: You just say John Does 1 through 100. Sure. But yeah, all 100 love them, let's say. We don't know. I've written about this before. If you get accused of insider trading and you are not a U.S. national and they freeze your account, it is probably not wise to get on a plane and fly to New York and say, I'd like my money back, because they might arrest you for doing the insider trading. But if you're totally innocent, you can take a shot. And so, in fact, a couple of– I don't know if they flew to New York. But a couple of the like John Doe's came to court. 00:04:18 Speaker 2: And used their real names. 00:04:19 Speaker 1: Yeah. And we're like, no, no, I'm not John Doe. I'm this guy. And here's why I bought these put options, right? And it's like, you know, I was on message boards and I saw people worrying about a potential crackdown. And I saw people buying put options and I thought, huh, that's weird. I saw volatility going up. And so like some of that, you know, it's always possible that that is like self-serving, right? And that like people are Sure. They can just say that. But to be fair, these people are not like, I'm a Chinese securities regulator, just coincidentally. It's like they're not Chinese securities regulators. Maybe they were tipped by insiders, but quite possibly their story is true and they saw someone else buying put options. 00:05:02 Speaker 2: I was just doing what that guy was doing. I don't know why he was doing it, but I have a good guess. 00:05:06 Speaker 1: Right. And people email me about this sometimes. To some people, it feels wrong to buy options. Because you see a big options trade and you have a good guess about why. If you see, looking at public stock tapes, if you just see a public trade and you're like, that must be insider trading, and then you go trade on it, people are like, that's insider trading too, but I disagree. I think that is totally fine. 00:05:30 Speaker 2: So trading on what appears from the outside to be insider trading, you're saying is essentially still outsider. 00:05:36 Speaker 1: Yeah. If you're trading on public information, that happens to be. 00:05:38 Speaker 2: The public information is I saw it in the tape. 00:05:41 Speaker 1: And like, you know, if you know, like if like the CEO of the company calls you and like, hey, I'm going to do a big options trade because we're getting acquired. Then like, that's bad. Right. But if you just. 00:05:50 Speaker 2: If you're looking at the shape of the trade. 00:05:52 Speaker 1: If you just look at, if you're just a momentum trader, then you're fine. 00:05:54 Speaker 2: You're riding the wave. 00:05:56 Speaker 1: And so, you know, some, you know, between. one in 99 of these Susquehanna customers were probably doing that. And Susquehanna's point, which I think is fair, is we can't really tell who is who. Right. 00:06:11 Speaker 2: And these people made money at our expense. 00:06:13 Speaker 1: Right. And the judge's point, which I think is also fair, is like, I didn't tell you to sue 100 John Does. Right. You sued the people who did it. 00:06:20 Speaker 2: Right. Tell me, give me the evidence of them actually doing it and then we can talk. 00:06:24 Speaker 1: I do think that like, you know, maybe the case proceeds and they like, you know, subpoenaed the brokerages and they get some information and maybe they can. 00:06:31 Speaker 2: What are they looking for? They're looking for like, oh, I texted my buddy from college and that's how I learned that, you know, they're looking for. 00:06:37 Speaker 1: Or just like literally the names of the accounts, right? Where like, if you like match the name with like the head of the Chinese securities regulator, like, oh, that guy might be insider trading, right? But yeah, I think it's very hard to like correlate a bunch of traders in a foreign country with like who you think might be insider trading. I think it's the sort of thing that a securities regulator can look into. And I think securities regulators might be looking into this, but it's very hard for Susquehanna. 00:07:03 Speaker 2: It's hard to identify the needle. 00:07:06 Speaker 1: Yeah. It's hard to prove who insider traded and who didn't. And like they have, you know, they do the best they can with like statistical, like doesn't this trade look weird. Right. 00:07:13 Speaker 2: But overall Susquehanna's case is like, but somebody did probably most likely. And don't you see that that's why we lost so much money. I mean, I know there are market makers that they have to, but like, Isn't there some point when you're selling those out-of-the-money put options where you're like, hmm, we're selling a lot of these. This is weird. 00:07:33 Speaker 1: So, one, I think I talked about this with Kate on this podcast. Ages ago, Jeff Yass, the founder of Susquehanna, did an interview with the Market Wizards series of books where he talks about this problem. In the 90s, getting picked off by selling a bunch of call options right before a merger. And then sometimes it was the insider traders who then gave the money back because the SEC came after them. And he's kind of like, yeah, but like we're a market, like it's our job. 00:07:58 Speaker 2: We have to do it. 00:07:59 Speaker 1: We can't just be like, oh, that's a big trade. 00:08:01 Speaker 2: We can't do it. They're not going to not do it, but they surely have some higher order thought about it. 00:08:06 Speaker 1: Right. But then the other thing is, and I've written about this, like the simple story here is that people kept coming to Susquehanna to buy puts and Susquehanna kept selling them the puts and then it was short a lot of puts and then like bad stuff happened and I lost a lot of money. But that's not like true. Like that's one part of the story. But the other part of the story is like they're an options market maker. Options market makers like just naturally immediately hedge. Not always. It's not like, but like for the most part, if you sell put options, you then go and short some stock to hedge your risk on the put options. But that's not the only possible hedge. You could lay off the put options, right? That you could sell a hundred put options to a customer and then buy back a hundred put options from other customers or from another market maker or something like that. 00:08:49 Speaker 2: Right. 00:08:49 Speaker 1: And so if you're a Cisco head and you're bringing this lawsuit, You're counting each put option that you sold to a customer and saying, this customer was maybe insider trading, so they should give us the money back. But you're only counting those. You're not counting the put options you bought. 00:09:04 Speaker 2: Right, right. 00:09:05 Speaker 1: Because those aren't insider trading. And so, okay, it's certain that they're partially hedged, right? Sure. And the judge quotes them admitting in the court papers, like, yeah, we don't deny hedging, but that goes to damages. We'll discuss that later, right? 00:09:20 Speaker 2: Yeah, yeah. 00:09:20 Speaker 1: But the customers made on the order of $ 70 million. This is what they allege. So Susquehanna lost $ 70 million just on those trades, right? Just on the puts that they sold to the customers. How much did they make on their hedges? It would be shocking if it was less than $ 35 million, right? But it would not be shocking if it was more than $ 70 million, right? There's no allegation that they were like, net short these puts, right? They're buying some puts from some people, selling some puts to some other people. Maybe, maybe they made money that day. I don't know. But they still can sue for the money they lost. 00:09:52 Speaker 2: Sure, sure, sure. But it's slightly obscuring the fact that they were also having a hand Right. 00:09:58 Speaker 1: Like what I wrote is that like, okay, so like Citadel Securities, another big options market maker, joined Susquehanna. They're like, we also sold options. 00:10:07 Speaker 2: Yeah, yeah, yeah. 00:10:08 Speaker 1: And we also got taken by this, right? 00:10:11 Speaker 2: Pity them. 00:10:12 Speaker 1: So both Citadel Securities and Susquehanna were selling options that day, right? Because they're market makers, right? Also, they were both presumably hedging. I don't know how they hatched. Classically, you delta hedge, you sell some stock, right? So maybe they just sold stock. But it's very possible that if you sell a lot of options, you'll be like, I'm going to buy back some options, right? Maybe they just have customer flow both ways. They have customers buying, they have customers selling, so they buy back from the customers who are selling. Maybe not, though. Maybe one or both of these market makers was like, we have sold a lot of put options. We don't like our risk position. Let's buy back some put options. put a bid in the market. It's not impossible that like Citadel Securities bought some put options from SIG or that SIG. 00:10:58 Speaker 2: Bought some put options from Citadel. 00:10:59 Speaker 1: Securities, right? And then like you have a mess, right? And it's like they should be suing each other. So I don't have any evidence. 00:11:05 Speaker 2: Some of the John Does are named. 00:11:07 Speaker 1: I don't have any evidence of that happened, but like it's, you know, totally possible part of the story, right? It's not like everyone who was buying put options that day was an insider trader, right? Like some of them were, you know, just people in the market. Like possibly... Pacing the frontier. 00:11:38 Speaker 2: Pacing the frontier. What a terrible... You should talk about that. 00:11:40 Speaker 1: No, it's great. 00:11:41 Speaker 2: Pacing is one of those verbs where it could mean either one, right? Like tabling? 00:11:45 Speaker 1: Yeah. 00:11:46 Speaker 2: They mean it to say slowing down. 00:11:47 Speaker 1: Sure. 00:11:48 Speaker 2: But pacing can also mean keeping up with, like in a running. 00:11:51 Speaker 1: To me, pacing as a verb mainly means like walking back and forth because you're nervous. 00:11:56 Speaker 2: Oh, huh. 00:11:58 Speaker 1: Like you're pacing. 00:11:59 Speaker 2: Pacing the frontier, like, oh, oh no. 00:12:02 Speaker 1: Oh, the air is going to kill us. 00:12:04 Speaker 2: It's going to kill us all. It's going to turn us into paperclips. Right. 00:12:07 Speaker 1: No, anyway, sorry. Dario Amadei. 00:12:10 Speaker 2: Furiously walking back and forth on the frontier. Wearing a cowboy hat. 00:12:15 Speaker 1: The CEO of Anthropic. published an essay over the weekend called Pacing the Frontier, basically being like, we should slow down, take a deep breath, and take our time a little bit on the continuing development of frontier AI models because otherwise they might kill us all. I don't think he said otherwise they might kill us all, but that was kind of implied. 00:12:33 Speaker 2: Well, this is, to be fair, sort of like the founding, theoretically, was supposed to be the founding ethos of, at the time, OpenAI, which Dario Amadei was a part of. I'm in the middle of reading Kevin Roos' AGI Chronicles, and... First of all, all of them all used to work together at the same place, whether it was Google or OpenAI. But they all, Dario Amadei especially, their main thing was like, we're so concerned with how AI is going to potentially ruin everything. So we need to be the ones to do it so that Google doesn't do it. 00:13:01 Speaker 1: Every AI developer is someone who's worried about AI killing us all. 00:13:05 Speaker 2: That's like the main identity. 00:13:06 Speaker 1: You can't get into AI without starting from the thought. 00:13:09 Speaker 2: What if AI kills us all? Exactly. Right. That is the number one starting point. 00:13:13 Speaker 1: There's so much about this essay. So many people are being like, this is all cynical marketing, and they don't believe this. 00:13:20 Speaker 2: What if it's genuine marketing? 00:13:21 Speaker 1: People have different views on this, but I think there's clearly a marketing benefit to conveying how powerful your AI is by saying, oh, what if it goes a little? I think, is there cynicism in the AI industry? Is Sam Altman being somewhat cynical? Maybe. 00:13:37 Speaker 2: Sam is a different person. 00:13:39 Speaker 1: Do I think that most people at The big AI labs are genuinely worried about AI killing us all? Yes. I don't think it's all marketing. Right. 00:13:46 Speaker 2: No, that's like, you definitely get that sense reading Kevin's book that like, that's the starting point for everyone is like, oh my God, this is so powerful. It's going to obliterate us. How do we control it? Also, I need to be first. Right. It's very paradoxical. 00:14:00 Speaker 1: I joke about this because like, I think a lot about how much AI researchers get paid. Oh God. 00:14:06 Speaker 2: And how- More than- everything. 00:14:09 Speaker 1: Right. And like how in the financial industry, there's like a traditional, like people have their number, right? And if you get paid $ 30 million, you can stop working. And not everyone, right? But like you need some method of motivating people to continue working. 00:14:21 Speaker 2: But they don't care about money. Well, they do and they don't. 00:14:25 Speaker 1: The fact that they get paid $ 30 million just proves that they do. 00:14:29 Speaker 2: Isn't it like 75 sometimes? 00:14:30 Speaker 1: Whatever, yeah. I mean, it's huge numbers. But like the point is that In the financial industry, people think very hard about how to calibrate salaries so that you keep working even. 00:14:38 Speaker 2: Though you're making $ 30 million a year. 00:14:40 Speaker 1: And I don't understand how it works in AI. And I have joked that one way it works is that because the AI researchers believe that their product will kill everyone on Earth, that is somehow motivating. I see. So it's not the money. 00:14:54 Speaker 2: It's the singularity. 00:14:56 Speaker 1: Yeah. And I've joked that when I worked at Goldman, if you had told everyone that our work was going to kill everyone on Earth, That would probably not make us more of a way to do it. But then I thought about it for a minute. I was like, well, maybe, maybe, maybe. 00:15:09 Speaker 2: It's going to be the derivative. 00:15:12 Speaker 1: But it's just like a weird, right. A weird culture of everyone being like, oh man, I'm so excited to go to work at the Killing Everyone on Earth factory today. But yeah. 00:15:22 Speaker 2: Okay. So Dario wrote this memo where he's like, we got to hold hands and not kill us all. 00:15:28 Speaker 1: Right. So. if you want to be cynical, there's like a very simple cynical story, which is that like, like one model for what's going on in AI is that the model capabilities are kind of ahead of deployment in a lot of areas, right? Like there's just like so much stuff where companies could add economic value by using existing AI models that they just haven't gotten around to yet. Cause like it's too new and they're not trained on it and they need more forward deployed engineers to like, you know, so you have like, Anthropic and OpenAI doing all these partnerships with private equity firms to basically roll out AI to companies, right? So if you just turned off model development for the next year, Anthropic and OpenAI could do a lot of business. They could sell a lot of products, right? And they'd make a lot of money because their unit economics are pretty good. 00:16:21 Speaker 2: That's going to work at the making... Money factory. That's not going to work at the killing everyone factory. 00:16:27 Speaker 1: Like AI right now is in kind of a great place, which is that it's commercially useful. 00:16:31 Speaker 2: There's all these applications. 00:16:32 Speaker 1: It has barely scratched the surface of its commercial use. I mean, people will disagree with that, but I think a lot of people think that. And it's probably not going to kill us all right this second. And so what a great place to be. But because there is a competitive pressure to build the best model every frontier lab, instead of only just going around sending out forward deployed engineers to like sell more compute to accounting firms is spending bajillions of dollars training their next model. And will the next model kill us all? 00:17:04 Speaker 2: Right. They're on the frontier specifically seeking AGI. 00:17:06 Speaker 1: Sure. 00:17:07 Speaker 2: Trying to get to whatever point. 00:17:08 Speaker 1: But like, but like, so right. 00:17:09 Speaker 2: So one, you can have the factory doing the making money over here, but everyone always needs to have the engineers at the frontier. Like the competition is like still being the best. 00:17:18 Speaker 1: Yeah. But like, but like one aspect of that is like that competition might lead to robots that kill us all. regardless of that, that competition is very expensive. It requires a lot. 00:17:26 Speaker 2: Of data centers, right? 00:17:28 Speaker 1: It requires a lot of training. 00:17:29 Speaker 2: So many chips. 00:17:30 Speaker 1: And if you just turned off that competition, first of all, we'd have one more year before robots kill us all. But second of all, they'd be much more profitable, right? And so the simple cynical story is that the frontier AI labs would have a better business if they didn't have to race to train their models, right? Right. 00:17:50 Speaker 2: So there's a business case you're saying for what Amadeus Yeah. 00:17:54 Speaker 1: And by the way, the biggest critic of this is Donald Trump, who called Jensen Huang to be like, it's a hoax. Donald Trump saying something is a hoax is the most terrifying thing I've heard about AI existential risk. But no, another critic of it is Jensen Huang, who does not run a Frontier ALAP. Right. 00:18:17 Speaker 2: He's the chip man. 00:18:18 Speaker 1: He sells chips. And his business is much better if they're constantly training the next model and buying new chips. So he is a critic of this for straightforward. I mean, I think that everyone has genuine beliefs. Sure. But also, like, his straightforward business interests are in racing to train models as much as possible. And Dario's and Sam Albin's are maybe not doing that. The other thing I wrote about is like, okay, so like arguably OpenAI and Anthropic have some business interest in not training models that fast so that they can, you know, capture some revenue from their existing models. And as several people, including David Sack said, you can just not, you just do that. You don't have to train the next model that will kill us all. You can just not do that. But there is competitive pressure to train the next model that will kill us all because one, Some customers want the most advanced model, and it really helps to be on the top of the leaderboard of who's got the best model. And then two, you're competing for researchers too, and the researchers want to be at the best lab, not the worst lab. 00:19:22 Speaker 2: It feels like a crucible of capitalism. 00:19:25 Speaker 1: Right. The interesting thing is that because there are competitive pressures, no one entirely feels like they can unilaterally disarm. But there's some chance that if you got the four leading AI labs in a room, they could agree to disarm. Right. And there's other non frontier competitors who could catch up to them, but maybe it'd be okay. But if you got like the four competitors in a room and they agreed to like slow the rollout of their products. that looks like a very classic antitrust conspiracy and so it's not clear they can do. 00:19:55 Speaker 2: That and so conspiracy to save humanity well this is the thing. 00:19:59 Speaker 1: Like so i read about this like there's this amazing lawsuit it's like against blackrock because. 00:20:04 Speaker 2: It's like yeah it's incomprehensible but. 00:20:06 Speaker 1: Like basically there's this idea that like ESG investors led by BlackRock, tell coal companies to stop producing coal. And why do they tell coal companies to stop producing coal? Well, they will tell you because coal creates pollution and climate change and is bad for us. And so as responsible stewards of capital and of the environment, they would like to produce less coal. But the plaintiffs in this lawsuit, which is several state attorney generals, They say maybe, but the effect of telling coal companies to produce less coal is that coal prices go up. And the coal companies have- Colluded. Yeah. Essentially, you have a cartel to sell less coal at higher prices. It's like OPEC for coal. And the takeaway from this lawsuit is that even if you have a good pro-humanity reason for conspiring to restrict the supply of your product, it's still potentially an antitrust conspiracy. 00:20:58 Speaker 2: And I quote, trying to save the world is no excuse for an antitrust conspiracy. 00:21:02 Speaker 1: Right. And, like, by the way, that has to be wrong. Like, obviously, trying to save the world is a very good excuse for an antitrust conspiracy. 00:21:13 Speaker 2: Yeah, the main excuse. 00:21:14 Speaker 1: Yeah. Maybe not under, like, traditional antitrust law. But, like, if you're like, the robots were an hour away from killing everyone on Earth, so I'd... Turn them off. Me and the CEO of my competitor held hands and turned them off. Like, great, do that. Yeah. And then you go to jail for antitrust conspiracy. 00:21:30 Speaker 2: Worth it. 00:21:31 Speaker 1: But, right, like... So Darius SA does have a footnote being like, with appropriate changes to antitrust law, right? They need permission to get together in a room to agree to. 00:21:41 Speaker 2: Turn off the killer alerts. Right, which is why you wrote that they can't just do it themselves. They sort of need the government to nod at them and say. 00:21:46 Speaker 1: Yeah, but it's like this weird situation where the people building AI are like, this is going to kill us all. 00:21:50 Speaker 2: And Donald Trump is like, no, it's a hoax. 00:21:52 Speaker 1: So I don't think they're going to get regulated. 00:21:56 Speaker 2: A couple of things that are really interesting about the Kevin Ruse book, which I'm only halfway through, but one of the really amazing. 00:22:00 Speaker 1: Is it out by the way? 00:22:01 Speaker 2: It is not out. The publication date is October 6th. I got an advanced copy because, you know, I edit a newsletter. But one of the more interesting... I mean, it is a sort of narrative history of how we got here and the race to AGI. And something that is really kind of funny to me is Amadei's Pacing the Frontier memo. Like, the entire book is like, oh, and then there was the big this memo, and then the big that memo, and the say AGI memo, and the all of these instances are marked by documents written by humans, which is so interesting to me because all that we talk about now with AI is like, oh, was it written with AI? The foundational inflection points and dramatic tensions in the book, every moment there is an associated memo. 00:22:44 Speaker 1: I was thinking about that. I find myself really allergic to writing that sounds like stereotypical AI writing. And so as I was reading Daria's memo, I was like, did Clara write this? 00:22:54 Speaker 2: And I'm like, no, it doesn't sound like AI at all. 00:22:57 Speaker 1: All of these people are good writers. 00:22:58 Speaker 2: Yes, no, there is zero. I mean, they're all philosophers, right? And that's like the most of what they're doing is they're getting together. They all live in this one house on Delano Street or whatever. And they're all like talking and staying up through the night. And like, they all write these original foundational pieces of text. And I just think it's just such an interesting idea that like some of the most important documents of like the race to building this thing that is intelligent are these like human handwritten hand sewn hand forged you know flags planted in the ground about you know this must be done or in order to do this this is where we are this is a watershed moment yeah. 00:23:34 Speaker 1: But that's just like just we're just in like a like a phase transition right like most of the code most of the algorithms behind modern ai are were invented by humans because who else would have done it, right? But like now we're at a point of recursive self-improvement where like now like the AI is programming the AI. All these people grew up in a, they didn't grow up in a world where AI would write papers for them. And like they've thought a lot about this. And the idea of going to Claude and being like, write a memo about restricting Claude is stupid. But like the next generation will just do that. They'll be like, hey Claude, what should I do? 00:24:08 Speaker 2: I think Claude will be great at that. 00:24:11 Speaker 1: He'll be like, no, like great for like, Claude will do a perfectly adequate job. 00:24:15 Speaker 2: We won't need to write the memos because Claude will already be our overlord. I know. 00:24:19 Speaker 1: Like Claude will do a perfectly adequate job of laying out the reasons for restricting the pace of AI improvement, but not stylishly. No, no. 00:24:27 Speaker 2: I mean, unless you trained it on, you know, some of the most stylish writing out there, like money stuff columns. 00:24:33 Speaker 1: Claude has ingested a lot of money stuff. 00:24:35 Speaker 2: Claude has ingested all of it. Let's talk about pensions. 00:24:53 Speaker 1: Okay. 00:24:54 Speaker 2: I was told that this was a thing that used to exist. You used to have a job in one place for 40 years. And then you would retire. And in exchange, you got a pension at the end. 00:25:07 Speaker 1: Yeah. 00:25:07 Speaker 2: It was very simple. Everyone had them. The world worked. 00:25:12 Speaker 1: Right. There were some problems with it. One of them is that if you didn't have a job for 40 years, if you moved jobs, pensions were more complicated. Another problem is actually like there's this like notion of a golden age where everyone had a defined benefit pension, but like, in fact, like the majority of people did not have defined benefit pensions. And then another problem is that occasionally pensions would go bankrupt, right? Because the idea of a pension fund is like, you know, you work at a company, the company agrees to pay you some number of dollars, some formula, something every month after you retire. And how it meets those obligations is like sort of in the first instance up to the company. And so it like sets aside some money and invests it. And if it invests it in dumb stuff and loses all the money, then you don't get your pension, which is bad. And so like ERISA, the retirement security law was passed to kind of clamp down on pension fund bankruptcies. But there was a set stream of liabilities. And so the company had to invest to meet those liabilities. And over time in the U.S., that has been like largely replaced by like the 401k. 00:26:18 Speaker 2: Why did that happen? What was wrong with the defined benefit pension besides the fact that it sort of never really existed in this imaginary wonderful way that we thought? 00:26:26 Speaker 1: Well, a couple of reasons. One is that if you're moving jobs, like a 401k is yours. 00:26:29 Speaker 2: Yeah, you're like, wait, I need to leave this company to go to different jobs. But like you can't take your pension with you. 00:26:34 Speaker 1: Right, because the pension like belongs to the company. So it's complicated to have a pension if you're moving jobs a lot. Whereas a 401k, you can move with you. It's just an account that belongs to you. Another reason is that the pension itself is a risk, right? Because there is a mismatch between the company has promised you a monthly dollar amount of benefits, and those benefits are 20 years in the future. And so it has to invest such that it will be able to meet that dollar amount of benefits. 00:27:02 Speaker 2: Better get it right. 00:27:03 Speaker 1: Yeah. One way to do that is to buy treasury bonds that will pay out exactly the amount of benefits, but that's very expensive, right? Another way to do it is to buy stocks that you hope will grow to meet the amount of benefits. Don't get it wrong. But you have, there's risk there. The thing about a 401k is that it's not, in the general case, a promised stream of benefits. It's just a dollar amount. It's just a big pool. Yeah. So if you put money in a 401k, you have the money in the 401k. If the stocks go down, they go down. But there's no, like, funding gap, right? Right. You're never, like, I mean, there is. Sure. There is, right? 00:27:35 Speaker 2: Maybe the number is as big as you want. 00:27:36 Speaker 1: It to be. Yeah. You can convert. either like literally by buying annuities or just mathematically by like multiplying by 4% or whatever, you can convert your balance in your 401k into a stream of income in retirement. And if that stream of income is not sufficient to support your lifestyle, then you have a funding gap. But it doesn't like look like that. It's not like on financial statements. It's not like a company being like, oh, we have a billion dollar funding gap. It's just like, oh, the number in your 401k is lower than you'd like. So it creates less concentrated risk, right? There's no, you're never gonna have a pension bankruptcy if you have 401ks. But then the other reason And I know this is true, but it's just like a thing people used to say is that the move from pension funds to 401ks was like good for Wall Street, right? Because the idea is that you take all of these people who are like, you know, working and like hoping, you know, counting on a pension for retirement and you turn them into individual investors. You turn them into people who, instead of just like counting on their employer to fund their retirement, they're now playing the stock market and they become customers of wall street. 00:28:34 Speaker 2: Right. And they're like, is some part of this also that like, Oh, now I can do the dials on my own 401k. 00:28:38 Speaker 1: Oh yeah, totally. Right. And so not everyone does the dials on their own 401k. Like the most popular 401k investment is target date funds. 00:28:45 Speaker 2: Set it and forget it. 00:28:46 Speaker 1: Yeah. Target date funds, which are truly set it for their, They're like not even index funds. They're like a slider of index funds where it's like, you know, mostly stock index funds and becomes mostly bond index. 00:28:54 Speaker 2: Funds over time. Right. 00:28:55 Speaker 1: So it's not turning the dials on your own 401k, but like you could, you could. There's all sorts of like possibilities for investments in 401ks. And so like the idea back in the day was like, oh, this is like creating a lot of customers for Wall Street. 00:29:08 Speaker 2: Got it. 00:29:09 Speaker 1: But what I've argued is that first of all, pensions, it doesn't really create customers for Wall Street because if a company goes from having a pension to having 401ks, Sure, all those 401k investors are now customers of Wall Street. 00:29:23 Speaker 2: Right, it doesn't change the shape of the customers. 00:29:26 Speaker 1: So you don't have the big pension customer, you have like a thousand small 401k customers. And in some ways that's better for Wall Street because you can sell the 401k customers all sorts of other services. 00:29:36 Speaker 2: More customers. 00:29:38 Speaker 1: Yeah, and maybe they'll be less sophisticated customers. in some ways. But the advantage of the pension is it actually turns out to be a really good customer for Wall Street. Because the pension can, because it has a very long time horizon, can do kind of spicy stuff that a retail stock investor wouldn't do. And so pensions are classically great customers for private equity and private credit and all sorts of other investments. And I think that 30 years ago, you would have said, well, probably these retail investors will pay higher fees to Wall Street than a big, sophisticated pension that can bargain for its fees. But what has happened in the last few decades is that the 401k has become just index funds, targeted, cheap indexed. public equity investment and like in fact it turns out retail investors pay very low fees like over time because there has just been a lot of pricing pressure from like indexing and pension funds like sure like they get a volume discount and their indexing is very cheap but like they can do you know put money into a 2 in 20 private equity fund and so the thing that I've been writing about recently is that I feel like there is a financial industry push to like revive the. 00:30:57 Speaker 2: Pension fund reinvent the pension fund yeah. 00:30:59 Speaker 1: Because like what you see is like this huge push to put private credit and private equity into retirement funds and yep how do you do that well you can just. 00:31:12 Speaker 2: You get trump to say it's allowed so. 00:31:14 Speaker 1: That happened right legally it's fine you can put whatever you want you put crypto sports betting everything um seriously But there's a question of what vehicle you use. And so you just be like, well, we'll sell private equity to retail investors. How did that go? Well, so I've been writing about this for a while. And you hear people at private asset managers say, look, retirement savers can take some risk. They can take some illiquidity risk because they have a long time horizon. They don't need their money for 30 years, so they should be able to take some illiquidity risk. They should be buying illiquid stuff. They should be buying private equity, private credit that pays some premium in exchange for not being able to sell it that day because they don't need to sell it that day. They're retirement savers. And like that has a certain theoretical appeal. Sure. It's true. And that's why pension funds can do it, right? Pension funds also have 30-year time horizons. 00:32:07 Speaker 2: But don't people like to, you know, maybe borrow a little bit from their 401k sometimes? 00:32:10 Speaker 1: Yeah, like it turns out that people do want liquidity in their retirement savings, even though like, you know, at some like first principles level they shouldn't because like, yeah, like emergencies happen Or just like the experience of being able to manage your own money means that you might change your mind. 00:32:25 Speaker 2: You might be like, oh, I want to move to something else. 00:32:27 Speaker 1: And when you go to a like institutional asset manager and you say, we'll pay you a premium for locking up your money. They're like, oh yes, we understand that. With retail investors, it's somewhat harder to convey that idea. And we've seen that this year with like the private credit PDC stuff where all these private business development companies raise money from retail investors. And they're like, we'll earn a liquidity premium by investing your money in private credit And then there was some bad news about software companies, and all these investors were like, we'd like our money back. And the BDCs were like, no, no, you're not supposed to. 00:32:56 Speaker 2: No, no, we told you only 5% of you are allowed to take them out every quarter. 00:32:59 Speaker 1: They were sort of premised on the idea that you wouldn't really want your money back, and then everyone wanted their money back. And there's just a misunderstanding, right? Where the private credit funds are like. 00:33:10 Speaker 2: Is that a liquidity mismatch? 00:33:12 Speaker 1: It's not. I don't like that. 00:33:13 Speaker 2: It's a psychological liquidity mismatch. 00:33:15 Speaker 1: Yeah, it's like the retail investors, it turns out, wanted more liquidity than the funds were offering them. And the funds weren't offering them that much liquidity because the funds really were making long-term investments. And it really was a bad idea for them to sell all their long-term investments to meet redemptions. And so you have that problem where it's actually hard to sell illiquid private investments to individual retirement savers. But selling illiquid private investments to retirement savers is both theoretically a good idea because they have long time horizons. They should not care about liquidity. They should be willing to get paid to take some liquidity risk. And it's also just like a good economic proposition for the financial industry because you can't make money giving people public equity exposure and charging them one basis point. So the way to like make money in active management, the way to make money as like an investment manager is to do something in the private's world. And so you see this big push for investment managers to do stuff in the private's world. And the way to do that is to make it look more like a pension. So there's a Wall Street Journal article today, kind of following up on a Bloomberg article from March that I also read about, about basically BlackRock pitching, I don't want to say a pension plan. No, no, no, no. 00:34:24 Speaker 2: I want to read to you the exact Wall Street Journal headline. BlackRock plans to make the corporate 401k look more like a pension. 00:34:31 Speaker 1: Yes. 00:34:32 Speaker 2: It's still a 401k, but it looks like a pension. 00:34:34 Speaker 1: Yeah. Okay. Right. 00:34:35 Speaker 2: It's a pension-like 401k. 00:34:37 Speaker 1: Well, what they say is they want to keep the benefits of a 401k, which is portability. 00:34:43 Speaker 2: Yep. 00:34:44 Speaker 1: Meaning that you don't have to do that one job forever. And also some amount of individual control with the benefits of the pension, which is professional management and long time horizons. And so what does that mean? I don't know. I'm not sure they know. But my general view is that it means putting private assets into it. I think the main thing that it means conceptually is that instead of telling you the number of dollars in your account, the long-term goal is to tell you the number of dollars you'll get each month in retirement. 00:35:15 Speaker 2: An annuity? 00:35:16 Speaker 1: Yeah. 00:35:16 Speaker 2: Which is a maybe more practical way to focus your mind on. 00:35:19 Speaker 1: Sure, sure. But like the way you implement that in a 401k is like you have some investment account, let's say a target date fund, maybe a target date fund that includes a great heaping helping of private assets. And your target date fund targets your retirement date. And when you hit the retirement date, it switches from a target date fund into an annuity. Like you just use the balance in the target date fund to buy an annuity. And the annuity pays you X thousand dollars a month until you die or for some fixed period of years. That's a pension, right? I mean, that's like to you that has the experience of a pension. Except that you don't know what the number of dollars every month is until you retire because they buy the annuity when you retire. But there are ways around that. You can estimate. You can buy forward starting annuities. You can structure around that if you are a professional asset manager. 00:36:09 Speaker 2: I am not. 00:36:10 Speaker 1: Right. So, like, I think the idea is, like, this is all baby steps because it's all optional because people have 401ks and some people are used to 401ks and they're like, I just want to own index funds. 00:36:20 Speaker 2: Right. 00:36:20 Speaker 1: But, like, the idea, I think, would have to be that in a perfect world, a company would offer its employees to push a button and the button is, like, BlackRock will give you X thousand dollars in retirement and, like, that's that. And, like, BlackRock... figures out how to get there. 00:36:40 Speaker 2: Right. 00:36:40 Speaker 1: And the way they get there is keeping helpings of private assets. 00:36:43 Speaker 2: Right, right, right. 00:36:44 Speaker 1: And the structure of it is they buy assets and hold them until retirement and then they- It's like. 00:36:49 Speaker 2: An elegant way to say don't worry about the short-term kind of stuff. Don't worry about the software stocks in the- Right. 00:36:57 Speaker 1: It's like focusing people's minds away from the dollar balance in their 401k, which went down today because of AI. 00:37:04 Speaker 2: Right, right. Yeah, yeah, yeah. 00:37:06 Speaker 1: Focus it on, like, you get a long-term pension, don't worry about it. 00:37:09 Speaker 2: Yeah, yeah, yeah. So psychologically a pension. 00:37:11 Speaker 1: Yeah, right. And that was the Money Stuff Podcast. I'm Matt Levine. 00:37:22 Speaker 2: And I'm Sylvia Killingsworth. 00:37:23 Speaker 1: We'll have fill-in guest hosts for a few months while my usual co-host Katie Greifeld is on parental leave. You can find my work by subscribing to the Money Stuff newsletter on Bloomberg.com. 00:37:33 Speaker 2: And you can find me by subscribing to the OnBooks newsletter at Bloomberg.com. 00:37:37 Speaker 1: We'd love to hear from you. You can send an email to moneypod at Bloomberg.net. Ask us a question and we might answer it on the air. You can also subscribe to our show wherever you're listening right now. And leave us a review. It helps more people find the show. The Money Stuff podcast is produced by Anna Mazarakis and Moses Andam. Our theme music was composed by Blake Maples. Amy Keene is our executive producer. And Cheryl Brumley is Bloomberg's head of podcasts. Thanks for listening to the Money Stuff Podcast. We'll be back next week with more stuff.