223. Is The AI Boom Like 1929 Or The Dot Com Bubble?

0:02This episode is brought to you by HubSpot. Now did you know most businesses only use 20 % of their data. That's like playing a football match with one fifth of the team. It's going to be hard to win, right? Unless you have HubSpot. Their customer platform gives you access to untapped insights in emails, call logs, meeting transcripts, and all the other data you need to grow your business. Because when you know more, you grow more. Visit HubSpot.com today.

0:42Hello and welcome to The Rest Is Money. Steph's not with us today, but I'm delighted to be joined by Andrew Ross Sorkin, distinguished journalist from The New York Times, who I'm not sure if we've ever met, but I've certainly known you by reputation for... Same here. Goodness knows quite how long. Your excellent book, 1929, is a re-examination on the basis of an enormous amount of evidence that you've amassed connected to that extraordinarily important crash. What was your motivation for sort of immersing in what I think certainly within the last 150 years is still the most important crash? The truth was that for me, I had read a lot of these books, trying to frankly understand that period better actually after 2008 because people would ask me to compare 2008 to 1929.

1:39At least initially, I didn't have a great answer. And there were great books. Kenneth Galbraith wrote the most famous one, The Great Crash in 1955. And there were a number of others, some books written in the 30s and 40s and another good one that was written, I think, in the 70s. but none of them at least to me brought you in the room with the characters with the people in a way where you really felt like you understood who they were and what their motivations were and what their incentives were I always loved books like Barbarians at the Gate and Den of Thieves yeah brilliant books and and books of this nature that sort of made you feel the sort of immersive view of it all.

2:20And I thought, there's so many interesting, fabulous characters, that if you could tell the story through their lives, you really could understand this period in a whole new way. And that's really what I sort of grasped in trying what I sought to do. The challenge of it, of course, is these people are not alive. And so I really had to spend an extraordinary amount of time just collecting letters and memos and diaries and transcripts and depositions and all sorts of other materials so that when you're in the room and you're seeing them quoted talking to each other those are real quotes so if they say you know if i say this is what they thought in the moment it's because that's what they thought in the moment as documented somewhere but finding all of that material was the was the great sort of puzzle mystery i was trying to put together and you were able i think to get hold of some material that had never been published before?

3:13A whole bunch of material. In fact, the sort of great treasure map for me was convincing the New York Federal Reserve to give me their board minutes from that period for reasons that are almost inexplicable. They had never been made public prior to this period. And we got a big back and forth, you would call it a row, of trying to get these minutes from them. After a lot of debate, they finally actually had a lawyer go through them and send me redacted minutes. So what was still sensitive? Well, so that's what I said. How could this be? And so six months later after that, then they sent me the actual minutes without the redactions.

3:55And that was just out of interest because it's there was actually, but what was it that they were worried about giving? I don't know what they were worried about. There was a lot of information about banks, mostly small banks, not big banks, that were seeking lots of help from the Fed at that point, as you might imagine. But they named them by name, and they had the exact amounts of money that they were seeking. And you could see the conversation that was happening. And for whatever reason, I think back then, they had told these board members that this was strictly confidential. It's no longer confidential, because the minutes today actually are made public.

4:27I believe actually everything from the 50s or 60s on is made public. But for whatever reason, back then, they didn't want to make them public. But by having those minutes and all the details from them, I was then able to go and find and so so much deeper in terms of knowing where all of my characters were at any moment, who they were talking to. And then I was able to go find archives of those individuals and their friends and family. So we ought to remind people that what we're really talking about is a period of astonishing boom in the stock market and a boom that was financed to a large extent by debt, by borrowing.

5:08This was what traders and economists would call leveraged investment. And that was the reason why that shock was, for example, so much greater than the bursting of the dot-com bubble back at the turn of this century because back in 29, everybody and his wife was buying shares on margin. You know, when shares fall and you've got no collateral, you're left with what in many cases was ruinous debt. And if you then look at the bursting of the dot-com bubble, there just was less debt associated with that kind of, even though it was pretty hard going for people. I wanted, therefore, to bring us to where we are today.

6:01There's definitely elements of a bubble in terms of this great AI boom. How much of this AI boom, how much of a sort of an incipient bubble that we can see is like 29 and therefore highly dangerous or a bit more like the dot-com bubble where it'll be painful but not ruinous? So I think we don't know yet. I think the truth is that the 1920s look a lot like now in terms of the great euphoria around the technology itself. I mean, by the way, back in the 20s, we were talking about automobiles and telecommunications and radio. I mean, the stock RCA was the NVIDIA of its time. It was like a meme stock.

6:44What was happening in the 20s was people had access to extraordinary leverage. You would go into a brokerage house, which, by the way, were popping up on the streets like Pret-a-Mange is on every corner. You'd just walk in or to a hotel, and you'd put down a dollar, and they'd give you 10. I mean, that was 10 to 1. So I don't think that is going on today at all. In terms of this AI bubble, what I don't know about the leverage piece is it's not coming from the investor class per se. It's more coming in how these transactions to buy data centers, build data centers, how much leverage is in that system right now.

7:21And just part of that, the bit that I'm increasingly anxious about is, you know, what we traditionally call vendor financing. Yes, vendor financing. Red flag. You know, where you're basically getting the creators, in this case, quite often of GPUs, these chips that power AI, effectively financing the AI businesses themselves. There's no question. OpenAI currently does not have enough money to pay for the amount of chips that they are committing to buy. The only reason they're able to buy them is because NVIDIA is financing this. So that is clear and that's a flag. But I actually think there's a lot more leverage in the system that is not even understood, which is to say that the energy complex that is going to power all these data centers, all the buildout, that is being financed by enormous amounts of debt.

8:10Similarly, some of the construction costs, some of the real estate companies that are building some of these things. I mean, there's a whole sort of subset of debt in the system that we don't know about. And the distinction that's even harder to understand, I think, is so much of it lives in the shadow banking system. I was going to ask you about that. How much of this is this booming industry of private credit? It's all private credit. And we don't know what the marks are. We don't have the disclosure. You know, even unlike 2008, you know, we don't know where the leverage is. And so I think there's a sort of big question mark if, in fact, some of these AI companies, big and small, can't pay their bills, sort of how does the whole system – does it collapse?

8:53Does it get just extended out into some – with private credit, some of these funds can get extended out into sort of the zombie-like operation. So it's not that necessarily it pops. It could be like a hiss for a very long time. Yeah, it could. And then the only other piece of this that I sort of throw a big red flag on is, you know, back in 1929, there was a budget surplus in the United States. The government had practically no debt. And so when you think about what happens in a crisis, the right thing to do, I think we've learned in a crisis, it's what they didn't do in 1929, is you flood the system with money.

9:30I think we learned that in 2008. That's the better path, at least from a practical perspective. The problem with that, though, is today what I don't know is given how much government debt we have, if at some point, you know, the bondholder community and the investor class raises their hand and says, we really don't like this anymore and sort of can spiral you into sort of an even deeper crisis. So I was with a group of fairly big investors the other day. There was this consensus, and I just wonder what you think about it, that if the worst came to the worst, because so many of these AI bosses have got close connections either with Trump personally or with people in the White House, they are all persuaded that if the worst came to the worst and there was a bit of a crash, that Trump would move heaven and earth to bail them all out.

10:22Do you think that's right? The truth is I think they probably are right in that thus far at least the only meaningful governor, I would argue, over this administration, meaning the Trump administration, thus far has been the market one way or the other. We saw it with the bond market back when he announced the tariffs back in April. We saw it just actually a couple weeks ago with the equity markets when he made those comments about tariffing China. and he has backed off every time he's felt that the quote-unquote marketplace was going to turn on him. So I do think, perhaps unlike other politicians, he would be more inclined to try to placate some of these business leaders and others to prevent a crisis.

11:09Interestingly, you go back and look at what happened during the pandemic. I mean, he was the president of the United States during the pandemic. And it was during his office that, you know, this enormous amount of money was just poured into the system. You know, bailouts for everybody. I mean, bailouts for everybody. Think about how politically unpalatable bailouts were after 2008 around the world. And then the pandemic happens and we just spent money like drunken sailors and nobody said a word. Yeah, I know. It is interesting. And it is, of course, one of the stark contrasts with 1920 29, where, you know, you get a market crash that then leads to the Great Depression and unemployment of an incredibly painful sort, not just in America, but across the world.

11:54And it didn't have to happen. I mean, that's the thing. The crash of 29, and the book is about the crash, but it's really actually about sort of the years that follow the crash as well. The crash itself was, to me, just the first domino, and then a series of other dominoes, which were otherwise terrible policy choices by the Federal Reserve, by the president, everything from trying to raise taxes in the middle of all of this to implementing tariffs to not spending more money. All of these things were sort of ultimately the dominoes that led to 25 % unemployment in the United States. They were.

12:35And obviously, and much of this was the sort of genius of a number of economists, especially Keynes, in understanding that governments do have a role, they can stimulate. And then we've got other economists like Minsky, again, who've understood, increased our understanding of how these bubbles form. And obviously, central banking understanding of how you can create money in certain circumstances and it not be inflationary. So, there's been huge sort of intellectual breakthroughs since then. But we shouldn't underestimate, certainly in a country like the UK, so much of the UK's economic stagnation, because this is an economy that's more or less stagnated since the crash, stems from the fact we were massively over-reliant on a city of London, a financial services industry.

13:25and we've not been able to re-engineer our economy sufficiently well subsequently to regain the kind of economic momentum that most British people think we deserve and it's one of the reasons why our politics has fractured in the way that it has. In America, obviously you have had an economy that has been much more vibrant in recent years than ours But what do you see as the sort of lasting damage of 2007-8 in America? I think the lasting damage is actually political damage more than anything else. And how would you define that? In the context of the undermining or the questioning of everything, of institutions, of expertise, of all of...

14:21So is it connected to the rise of Trump? I think it's connected to the rise of Trump. I think it's connected to the rise of populism. And I think that there's been a real sort of rethink about who you're supposed to listen to and who's really for the people. And that has clearly impacted things. Which is wrong, this, because we're recording actually on the day that Mamdami has become mayor of New York. Rise of Mamdami as well. And there's something quite striking about a city like New York, which, you know, is, I suppose, the nearest you have in America to the equivalent of London with its dependence on Wall Street and financial services.

15:03It is the capital of capitalism being now run by a socialist. Which is which is fascinating, isn't it? And so, again, I mean, is that in some way? Is that the long tail, the long shadow of what happened in 2008? I genuinely think that there is a long shadow of 2008 politically. I think it led to the rise of Trump in 2016. I think it's led today to the rise of Mamdani in now 2025. So yes, I do think that there's a rethink of what the establishment has done and almost a backlash against the establishment across the board now in the United States. And one other just quick mention, though, you suggested that our economy, the United States, you know, has managed to bounce back better than the UK.

15:55And the truth is it has, but it hasn't. If you really look at what's happened in the United States, so much of what's happening in terms of our growth has been powered almost exclusively, at least over the last couple of years, by all of the AI spending. If you were to actually remove AI spending, in fact, Jason Furman from Harvard, the economist, just did a fascinating study. If you take out all of the AI spending, you have growth that is literally flat. I think it's up 0.1%. So what's happening is it's almost papering over the rest of everything else that's happening in the United States economy in this moment.

16:31And I think that's another sign or at least red flag to consider given the concentration of growth is coming from one place and almost exclusively that place. Let's unpick all of that in a minute or two. This episode is brought to you by Wealthify. If you've ever thought I should really do more with my savings, you're not alone. Most of us are busy juggling work and family life, so investing can feel complicated. Thankfully, Wealthify makes it simple. You put your money in, their experts take care of the rest. No picking your own stocks, no jargon, just a managed investment plan you can keep track of 24-7.

17:12Open a Stocks and Shares ISA, Personal Pension or a Junior ISA for the kids and with over 100 ,000 customers and the backing of Aviva, you know your money's in experienced hands. Plus, when you deposit or transfer to a Wealthify Stocks and Shares ISA, you could earn between£50 and£1 ,000 cash back. Please note that minimum investments apply and cash back varies by the total amount of your deposit or transfer. So for investing, savings and pensions, the smart money's with Wealthify. Open your account today at wealthify.com. Wealthify is authorised and regulated by the Financial Conduct Authority.

17:50Offer registration closes on the 2nd of March, 2026. T's and C's apply. With investing, your capital is at risk. Seit Torben über die stabile Plattform von Flatex traded, fühlt er sich wie Torben Gecko. Für alle, die wie Torben Börse im Blut haben. Bis 28. Februar, Depot eröffnen. Wir übernehmen deine ersten Order-Provisionen bis zu 500 Euro. Flatex, besser richtig handeln. Investieren birgt Verlustrisiken. Bedingungen gelten externe Kosten fallen an. Welcome back to The Rest is Money with me, Robert Perston. I'm delighted that Andrew Ross Sorkin is still here. So I suppose there are two questions that follow from your absolutely correct statement that American growth at the moment is massively dependent on AI investment.

18:35I suppose there are two things to say. I mean, yeah, it might be a bubble, right? But this is still real investment. And I suppose, I mean, one of my concerns would be, obviously, it is also fueling a big phenomenon that's been true, actually, of much of the West for the last 30, 40 years and seems to me to be being accelerated, which is the kind of growth that we're seeing at the moment, the kind of booms we're seeing are definitely increasing inequalities. Is that something you worry about in the context of the, in a sense, the structure of the US economy? Essentially, it is just a great machine that makes a limited number of people unbelievably wealthy.

19:19But you also have large numbers of people who don't believe that the economy works for them. And that's another reason why you get so. I mean, there is this paradox that you've got the wealthiest, you know, we think president in history. But nonetheless, he talks the language of unfairness all the time. And I imagine it's going to become exacerbated even more, which is to say, here we are talking about, is there an AI bubble and will there be a popping of the AI bubble? Let's say there's not a popping of the AI bubble. Let's say the AI bubble continues apace. Well, in great success, if in fact these valuations of these AI companies are to be believed, the only way they're going to be achieved, these valuations, is if they create shocking amounts of productivity.

20:04Well, what does productivity really mean? It means growth at a lower cost. How do you get to a lower cost? It means getting rid of people like you and me. And then you start to think to yourself, in success, you ultimately may have failure. Whatever probability you put on AI, either enhancing productivity to the extent that vast numbers of people aren't needed or indeed just purely, totally just replacing humans. And there is, you know, there's a there's a meaningful probability that will happen. If you're running a government at the moment, you should be preparing for what would be the mother of all economic and social shocks.

20:42You'd have to think about your welfare system, your education system, everything. Right. None of them are having this debate at all. So, I mean, you know, we all knew a pandemic would happen at some point. Nobody prepared for it. We know that there is a meaningful probability that AI will totally change our economies in a way that is destructive for millions of people's lives, certainly in the short term. And you've got to prepare for that, right? But the complicated part is I think we don't know. I don't think we know what it would look like. I think we could scenario plan for, you know, five different sort of variations of how it might play out.

21:18But I think we don't really know yet what it's going to be. Yeah, but you cannot know. OK, so let's now I'm going to take us back to the thing that sort of in a sense is our, you know, our sort of bond, as it were, which is the crash. Right. We all knew, OK, that the kind of lending that was going on in the subprime market in the US was totally unsustainable. And anybody with half a brain knew that the value at risk models of the big banks were ludicrous in the way that they were basically saying there are no risks here. And we knew this for months and then in some ways, you know, years before the crash happened.

21:55And yet our regulators and our governments didn't prepare, you know, because they thought, you know, basically, well, who knows when it's going to happen? And, you know, we don't know precisely what's going to happen. So nobody prepared. Right. And this is the same. I mean, this, in my view, is a potentially way more worse social and economic outcome. And even if we don't know precisely what's going to happen, there should be at least some contingency planning going on. Oh, I don't disagree with you. The question is, what would those contingencies genuinely look like? So there's sort of two parts to this.

22:26To me, one is the how do you prevent whatever might be a popping of this bubble piece of it, which is to me around disclosure and private credit and trying to understand what's happening in that marketplace. But the other is a social issue. I don't know if it's social so much as an economic issue, which is, OK, let's say that there is mass unemployment in the United States or the UK or elsewhere. Are we talking about UBI? Are we talking about tax cuts to allow people to afford? Are we talking about training? What does the training look like? You know, if everybody's unemployed, there is no income tax.

23:03So let's just say, you know, part of the answer is certainly for a period, welfare support, a universal basic income, you know, a system where governments essentially, you know, finance people's living standards as the whole economy adjusts. But at the same time, there's no income tax coming in from millions and millions of people because they haven't got jobs. In those circumstances, somebody like democratic socialist Mandami is sort of probably in a better space than a Trump, because ultimately, if those who are essentially the owners of this technology who are reaping, presumably in these circumstances, rewards beyond our wildest dreams don't then make a contribution back, we're in trouble, aren't we?

23:41So this argument, I mean, we're not in a world, particularly in America, where it's easy to talk about redistribution. But in that kind of world, you have to talk about redistribution, don't you? Oh, I think you'd have to have some form of redistribution. It's very hard to fully contemplate the magnitude of it. And even if Bill Gates, Jeff Bezos, Elon Musk and Sam Altman or whatever, put all of their money together. And then transferred it to the people. And literally gave it to the people. It would still last, in truth, only, you know, probably six months. I mean, it's really, I think that's, it's a very interesting question.

24:22You know, it's a terrifying question. Some bubbles, which when they burst, nonetheless leave you with really important infrastructure. Right. So the railway bubbles. Right. Nobody's complaining. We didn't have the railways today, even if there was a lot of excess and people lost a lot of money. I would say, actually, although it's perfectly clear that quite a lot of the digital revolution has done social harm, thinking about social media companies and the mental health problems and all of that, nonetheless, despite the fact there was a ridiculous boom and overvaluation in the stock market around those original dot-com businesses, we are left with real companies and real industries.

25:09I don't believe, if I'm completely honest with you, that the financial boom of 2007-8 left us with anything valuable. Do you? Oh, no, I don't think the financial boom left us with anything valuable. The internet boom left us with something valuable, fiber in the ground. Yeah, totally. Oh, that's also true. Which was real infrastructure. And I think this boom is going to leave us with infrastructure. Well, here's where it gets more complicated. It gets more complicated because these data centers are basically buildings with chips in them. and the chips do not age well. That is actually the biggest distinction.

25:45So the fiber in the ground has great value. The train tracks have great value. Three, four years from now, you have to depreciate those chips as quickly as humanly possible, and you have to replace them. In four years from now, you will not want those chips. And that is a real meaningful question as to what value all these chips are actually going to have in the future. And what about all the power stations? And by the way, that will have value. We will need those. So there's part of the infrastructure that will have value and part of it that may not have value. Unfortunately, the biggest cost component of these data centers is the chips themselves.

26:21Now, I'm going to bring you back to, I suppose, the sort of one of the biggest things that exercises certainly leaders in this country, but also across Europe. I mean, you know, we've described an American economy that is flawed, but nonetheless has a momentum that Europe lacks. I mean, as somebody who sort of looks at us from a little bit of a distance, what would you describe as the single biggest differences that mean that, you know, in the end, the AI boom and it might be a bubble is focused on America and not on Europe? Why do you think that is? It's a cultural issue. And it's actually articulated by Winston Churchill, who makes a cameo appearance in this book in 1929, when he's literally in New York as the stock market is crashing.

27:19And by the way, he also got bitten by the bug of stock speculation and lost his shirt in the process. He happens to actually see somebody, he thinks, jump out of a window from his hotel room at the Plaza Hotel in New York. And this is 1929. This is October 1929. He happens to be there. This is before he was the prime minister. And he writes after seeing this person fall to his death about the American economy and this idea of just the American culture of optimism. and basically suggests that there's an optimistic nature to Americans and to this thing called the American experiment and effectively argues that the British people, frankly, should be more optimistic, that you actually need a bit of this sort of speculative feeling in the culture to create great innovation.

28:26In fact, I think he would argue that speculation is a feature, not a bug, that you need sort of that kind of optimism to get there. And it's interesting because, look, I'm a journalist. I, to some degree, am a professional skeptic. That is my job. But it has paid a lot better dividends over the last hundred years to be a professional optimist than to be a Cassandra. I mean, that has always been the truth. Of course. And but, you know, I used to live here in the UK. This is 20 plus years ago. And it's true. I think as a people, the culture is such that we're always sort of looking at what's the downside, always at the risk in a way that people in the US don't always spend their time with that emphasis.