Title: The $1.5 Trillion of Hidden Debt Fueling the AI Boom | Robin Wigglesworth of FT Alphaville Show: Monetary Matters (host Jack Farley) — The Monetary Matters Network Guest: Robin Wigglesworth — editor of FT Alphaville (Financial Times); author of "Trillions" and of "A Fabulous Debt: The Epic Story of How Bonds Built the Modern World" (Penguin Random House; publication 2026-09-22, so this is a pre-publication interview) Date: 2026-08-16 (published) URL: https://youtu.be/kZXjQZWUHH0 Length: 65:16 Note: YouTube auto-captions. Fillers (um/uh/"you know"/"I mean"/"sort of"/stutters/false starts) removed; wording otherwise verbatim — no paraphrase, no reordering, every (mm:ss) cue kept in place. Auto-caption NAME garbles were corrected in the body and are mapped here. VERIFIED: "Alphavville" = Alphaville (FT); "core we've" / "cororeweave" / "Coreweee" = CoreWeave (NASDAQ: CRWV); "Larry Frink" = Larry Fink (BlackRock); "Jay Cook" = Jay Cooke & Co. (the 1873 failure); "Gregor McGregor" = Gregor MacGregor; "pouet" / "poyet" = Poyais (MacGregor's invented country); "gobles" = Goebbels; "Sachin Della" = Satya Nadella (Microsoft CEO); "in 1973/1873" = the book "1873: The Rothschilds, the First Great Depression, and the Making of the Modern World"; "Ilio Leoad Aamemed" = Liaquat Ahamed, its author (also author of "Lords of Finance") — verified via the Monetary Matters episode listing; "Bur" = Michael Burry; "KKKR" = KKR; "Eerie Canal" = Erie Canal; "Grunder Crack" = Gründerkrach; "Bismar" = Bismarck / Otto von Bismarck; "liable" = LIBOR; "tener" = tenor; "as back securities" = asset-backed securities; "Charlie Mer" = Charlie Munger; "Blue Owl" = Blue Owl Capital (NYSE: OWL); "Hyperion" = Meta's Hyperion data centre, Richland Parish, Louisiana. LEFT AS SPOKEN / UNVERIFIED: "depreci favorable rates" (garbled), "needing you 20 30,000 million working capital" (garbled figures — left approximately as spoken), "USPS" = USPs (unique selling points), "to each of the road" (garbled, probably "to each their own"). NEVER turn an unverified garble into a ticker.
00:00 I'm joined today by Robin Wigglesworth, editor of Alphaville, the Financial Times financial blog, and author of A Fabulous Debt: The Epic Story of How Bonds Built the Modern World. Robin, you've been doing some work on the off-balance sheet hidden leverage of the hyperscalers, Meta, Google, Microsoft. So we're reading, oh, hundred billion dollars of capex.
00:21 Google's doing that. Microsoft's doing $150 billion. This is so much money. Oh my god. But actually, it's looking like it's almost guaranteed to be way way higher. You've been looking at the numbers and just from the first to the second quarter the guarantees, the lease obligations and so forth off balance sheet again went from roughly one trillion to roughly 1.5 trillion.
00:42 What are your thoughts? What are we looking at? >> No, it's fascinating. It is one of the biggest capital markets events of our lifetimes really. We've seen massive capex booms before like the railways in the 19th century is the classic parallel that people draw — transformative technology very expensive to build.
01:04 What's unusual of course, railways back in the day used to be almost like venture capital ideas, they were very VC-ish. Today it's like major large money machines that are doing it. Google, Alphabet makes lots of money, Meta makes lots of money, Amazon makes lots of money. And for a long time the money they were pouring into data centers, they could just fund it from their free cash flow.
01:30 Google search, Amazon, Facebook itself just prints money. So it's easy. But the scale is just becoming so massive that they've increasingly turned to the debt markets, as we now see. There are actually some signs of indigestion — the sheer scale of the bond sales. We're talking multiple hundreds of billions of dollars both last year and already this year; we've already smashed last year's record for the hyperscalers' bond sales.
01:56 And they're getting more creative, and look, creativity in finance can be a good thing. I find a lot of this stuff fascinating, but it can also be quite dangerous, as you know, Jack. And first it was structuring some of the bonds as leases. So essentially, take a great example, Meta is building a huge data center in Louisiana called Hyperion.
02:21 And rather than pay to build it, they're only investing 20%. They're buying 20% of it, but they are guaranteeing that they will lease that data center for 20 years. So the lease payments will essentially cover the costs of that company itself, like a JV with Blue Owl.
02:41 And they'll sell those bonds to other investors. So it's off-balance sheet. It doesn't come up as a bond or a debt or a loan for Meta, but of course it's on the hook for paying this lease for 20 years. And this has inspired a lot of the other hyperscalers. So we've seen massive amounts of these kind of lease structures happen.
03:02 So that's what's gone to $1.5 trillion. Up from less than a trillion last year and nothing, de minimis, a few years ago. And crucially, 500 billion or so of that you can see as — the leases have started. So you can see them in the financial accounts. They won't appear as debt but you'll see the payment obligations on the balance sheets.
03:29 But a trillion dollars of that is for leases that haven't even started yet. And that doesn't appear except as a footnote. So Goldman Sachs, that's where I got the numbers from. They did God's work in going through all the filings to find that stuff. What I did then was I started looking at the purchase commitments, because these companies have also promised to buy obviously chips and equipment, cooling, stuff like that, but obviously power as well.
03:55 These data centers need electricity and they need quite often guarantees that they will get that power, and that has gone from again also roughly $1 trillion earlier to $1.5 trillion. And these are quite often payment obligations they can't squirrel out of. So they walk, talk and quack a bit like debt but they don't actually appear as debt.
04:18 And I think it's fascinating. Maybe this is, they're betting the house on AI and I hope this all works out, but I'd feel maybe slightly more comfortable if they structured some of this as more plain vanilla debt and let the debt markets do the talking. >> Well, if they'd structured it as plain vanilla debt, Goldman and you wouldn't have to do all of this work to figure out.
04:40 It is really interesting. So it's really a transformation from a pure complete AAA double A investment grade counterparty like Microsoft, and it is using that but through a much less investment grade data center developer or a neocloud, to then — they're the ones who are actually spending the money and they report to investors this giant backlog which CoreWeave just reported, and that giant backlog is basically these off-balance sheet commitments that the Microsofts and the hyperscalers have made. Yeah, you
05:10 can say this is entirely disclosed. These companies aren't hiding the fact that they're pouring hundreds of billions of dollars into capex and that money has to come from somewhere. They're not hiding this. But I do think >> [snorts] >> they could perhaps have chosen a slightly more transparent approach to this.
05:31 Rather than wanting to preserve the optics of pristine balance sheets, they are increasingly getting creative about how they structure, how they pay, how they disclose it. Just literally going through the 10-Qs to look at the purchase commitments. Some of these companies, Google for example, actually was admirable — they don't break it up, but you can search for it and find it fairly easily.
05:52 On some of the other companies, I had to spend quite a lot of time digging it out. It's not easy. And I think that's unfortunately >> who's the most complicated? >> I guess all the others except Google. Some of them disclose it. But they don't disclose it in a uniform way. They use different words in different places.
06:11 So it's hard to compare one quarter to the next quarter. And some of them don't disclose it at all really. They just say they have material upcoming payments or promises to make payments. >> So it's a bit of a mixed bag really. Google maybe because it's the biggest. Google has purchase commitments now of $800 billion.
06:33 >> So that's chips, memory, equipment, cooling, electricity, the whole jamboree. >> But that's almost half the total. I think maybe their accountants got a little bit worried and thought we need to be quite transparent about this. >> And do you have a rough sense of over how many years the majority of that is scheduled for? Because if that's scheduled for the next three years then the Google capex number for expectations is too low.
06:57 >> So Google, using them again, they broke it up a little bit more transparently — around $200 billion of that $800 billion is what they call short-term. They don't define it there but I'm pretty sure that'll be over the next 12 months or so. So their payments are coming due on the leases.
07:16 If the leases haven't started yet, it's an accounting thing. Once a lease is started, you actually can put a right-of-use asset on your balance sheet and then put the liability on the other side. So it is transparent. This isn't them necessarily doing something bad or changing anything or doing anything differently from how people have done it forever.
07:36 I just think that the scale of it is now such that it's a whole new world, and maybe these off-balance sheet liabilities — because they are financial liabilities that are in many cases extremely hard to squirrel out of; Meta's guarantee for example of the leases for the Hyperion data center are incredibly strong.
07:57 I don't see how they can squirrel out of them. Then they should be more transparently disclosed and then investors can make their minds up, and broadly speaking I think people are okay and understand that they are spending this money, these companies are hiding it, but I wish some of the gamesmanship could maybe be cut out and them breaking out a little bit more obviously >> Robin, what do you make of the deal, the memorandum of understanding that Nvidia made with five or six giant alternative asset management firms,
08:30 Blackstone, BlackRock, KKR, to finance chips and recognize that these are an investable asset class. I asked Claude before this what percentage of that 500 billion is going to be debt versus equity. They said roughly 80% debt 20% equity. So you're the debt guy, so this is good, we're speaking to you.
08:52 >> I know, this is a debt cycle. There's lots of hoopla around the IPOs of SpaceX and Anthropic and OpenAI to come but really this is a debt cycle. That actually makes me more worried. Fundamentally, incredible transformative technologies come around occasionally and sometimes they come true.
09:10 The railways were genuinely transformative, as were canals or telegraph poles or the internet. But when these big investment cycles are primarily equity financed, they can break bad but it's generally fine. The end of the dotcom bubble, the stock market dropped 50% peak to trough; economically it was a nothing burger, it's hard to disentangle from the effects of 9/11. Debt-fueled capex cycles are very different — even when the underlying premise comes true and AI transforms the world, quite often they
09:42 end in tears. On this specific deal, one thing I would say is that it's very easy to push out press releases saying we're going to lend or invest X or Y or Z into this or that, and we'll see what actually materializes. Clearly there is a lot of heat in this area now and everybody wants to be seen to be leaning into it.
10:07 But it's going to be really interesting to see what they actually do and how they structure it, because I think a lot of these investment firms in particular are going to be very careful about how they protect their own balance sheets but also the balance sheets of their investors. So I'd urge people at this point in the cycle to take press releases with a pinch, maybe a fistful of salt.
10:31 >> Yes. What do you think that these AI securities are going to look like? Larry Fink literally said — I don't know if he was on vacation — these AI securities. What's an AI security? Well, I think it's just compute and I think that's quite an interesting thing, and I think this story is throwing in a lot of different things, but it's the transformation of compute — you can buy, lease a certain amount of GPUs I guess, or how you structure it, that can be turned into an asset class. I think there is a
11:03 journey there. Just because you say something is an asset class doesn't make it so. When it does become so, the SEC typically has something to say and will have a view about it. But I do think that is the directional travel, that in the same way that water or a commodity is a tradable asset class, at least the futures on these things.
11:26 I can see us getting compute futures as well and that becoming roughly investable. My wariness is that just because something is an investable asset class doesn't mean people should be investing in it. We'll hear lots of stuff about oh, you need to be diversified and this is democratizing access to whatever everybody has, and to paraphrase, who was it, Goebbels, the Nazi propagandist who said when I hear people talk about culture I reach for my gun — he was not a nice man but whenever I hear people talk
12:00 about democratization of something I tend to reach for my metaphorical gun, because it's usually a code word for jamming something down the necks of retail investors that are not really quite ready to digest. So I actually have great hopes for compute futures and turning compute into some sort of tradable asset class.
12:19 That sounds cool and interesting to me. But I think there's a journey still to make on that. And I'd worry about what happens along the way as well. And something like a commercial real estate building, yeah, it can be risky, but a building is generally going to hold its value and generally appreciate over time.
12:40 So a loan to value of 50% — you lend 50% of what the property is worth — is pretty appropriate. And you're lending to pipelines of oil infrastructure; these are things we understand. And for something like compute, yes, the market right now is super hot.
12:56 And Google is buying compute from SpaceX for a super super high amount of money. And on the depreciation argument, like the Michael Burry argument, basically every single data point of the past nine months has not supported the Michael Burry depreciation point, like depreciation favorable rates.
13:18 But just because Michael Burry's been wrong doesn't mean that there's going to be a glut at some time and that lending a trillion dollars against this thing is a good idea. >> Well, fundamentally it's a lending decision if you're lending towards it. The chips do degenerate. There is a half-life to how long you can keep them and, for example with SpaceX putting up in space, how are you going to do maintenance? How are you going to replace chips that burnt out? But I would say that look,
13:44 finance lending officers mess this up all the time but broadly speaking that's what markets are really good at; that can be priced as long as it's a known risk, and this is very much a known risk, people can price it in. People can adjust and sometimes they'll make a mistake and they'll get their faces ripped off and other people will make money but that's the way of markets.
14:10 It's what makes this system so dynamic. So I'd be worried about the sheer amount of money going in there and some of the return expectations and this sense of FOMO that seems to be everywhere these days. But that doesn't mean that the end destination isn't the right one. It's just how much money we lose along the way and who loses it and when.
14:35 Robin, in your book, you've been doing a lot of work on close to a millennia of reading of financial history. I'm curious about the trends and patterns you've noticed about the following question: pricing power. Whenever a new industry emerges, there often is tremendous pricing power, as there is in AI and semiconductors right now.
14:59 Throughout your many many centuries of reading of history, what tends to happen to that pricing power? >> It tends to erode. There are so many examples of this. I do worry for example that right now Nvidia is the picks and the shovels distributed to the entire AI revolution. But it obviously depends on its own supply chain, and I do wonder about the assumption that nobody else can create GPUs at scale and quality of an Nvidia ever.
15:34 Because right now we're pricing that in, and pricing power tends not to last forever. In a capitalist system people respond to incentives, and monopolist-like pricing power tends to not last very long. Sometimes it can last for a few years. But it never lasts forever as far as I know. >> What's the most extreme debt cycle that you studied in the book? The GFC, the lead up to the GFC, because it was the culmination of a debt bubble in every part of the world and in every sector. Sometimes it's governments,
16:14 sometimes it's companies, sometimes it's households. In this case, actually, governments weren't for the most part levering that much up. They were doing a little bit, but it wasn't too bad. But it was pretty much everywhere. Everybody thinks that our banks were uniquely terrible or our politicians or our government was uniquely ineffectual, but in reality it was a global phenomenon and the scale of it was just wild.
16:41 And one of the reasons when bonds and debt becomes particularly dangerous is when essentially it's been so long since a previous crisis that you treat it as money or money-like. It's one of the USPs, the ultimate selling points of bonds originally, was that you could use it as collateral, as money, for certain things.
17:03 It was because government bonds — and then over time people started using high-grade corporate bonds like IBM or Microsoft, very solid, you can use that as collateral for loans. But then of course in the 2000s people started using asset-backed securities and initially those were super solid as well, and eventually we take things too far and they were not the equivalent to money; in fact you couldn't trade them and some of them were close to worthless. And that I think is what really transforms almost like a
17:34 humdrum market downturn or an economic recession into something nasty, really bad. It's not when you invest in something that's risky and it blows up in your face. That's fine. That's just risk and reward. They're part of it. If I invest in junk bonds, look, if they break bad, if they default, I can't complain about that.
17:54 Maybe I did something stupid. Maybe the company did something stupid. That's fine. It is when you invest in something you think is super safe, or you base your entire investment strategy or the business model of the bank around something that you thought was super safe, proved not to be so, and that's what happened I think in 2008. It wasn't just the scale of the debt bubble, it was how people treated it. That's what transformed it into such a horrific financial disaster. But it probably isn't my favorite crisis because there are so many to
18:25 choose from, really demented ones. >> We'll get into some demented ones. So a principle you're saying is basically financial crises are caused not by perceived risky assets, but by assets that are perceived to be safe but that are risky. You mentioned junk bonds. Now the more polite term of course is high yield.
18:43 And it's funny, obviously the real action of risky credit lending was junk bonds when it was invented in the 1980s, but now all of that risky lending, a lot of it has migrated from the high yield bond market to the private credit market. So the high yield bond market is so-called safe relative to the private credit market.
19:01 I'm sure the private credit people would disagree. What do you make of the rise of the private credit asset class? And what have you made of the jitters in the market? And I'm curious, to what degree do you think they are real versus just headlines and not much substance to them? >> Oh, they're real.
19:21 I've been borderline obsessed with private credit for a long time and I've had many arguments with people in the industry, and the nuance — I think it's a fantastic asset class. I think it's a fantastic idea. I hope it grows and grows because I actually think it does derisk the financial system. That's not just marketing from the executives in the industry.
19:40 I think it's great if we take basically these bundles of risks, which is what every loan constitutes, and that is in the investment ecosystem, in the non-bank system. I think that's a great thing. But as we know, whenever people get over optimistic people do dumb things — people have invested way too much money in private credit based on very backward-looking numbers and the illusion of safety, or just frankly the lack of volatility which is just an artifice because of the lack of mark-to-market accounting. So yeah, high
20:18 yield frankly — it's not safe but it's far more solid I'd say now than it ever has been. The ratings is an obvious way to look at it: over half the market is double B now. But just generally the quality even beyond ratings I think is far more solid and it throws off cash.
20:36 The technicals have been great. And private credit has picked up all the dicey stuff and I think that's great. That's where it should be. But that does mean there's been dicey stuff happening there. For me, the real wake-up moment was when I was still in the United States and I started getting cold calls and offers for private credit lines. Me, as a journalist.
20:58 That's just astonishing, right? Nobody should lend any money to any journalist ever. >> To lend you money, not for you to be an investor. Really? Okay. >> People were offering me term loans, 20, 30,000, million working capital, 3-4 percentage points above LIBOR.
21:15 Incredible spreads. I didn't actually take them up on it, but I just thought, when we're getting to that kind of spray-and-pray approach to origination — there was so much money flooding into private credit. And it's very much: how do you find the borrowers? How do you find high-quality borrowers? Well, actually, in the end, you don't need to find high quality borrowers.
21:33 You just need to find borrowers to take the money so you can earn fees on it. It's very similar to what we saw in 2008. Not in scale, of course, but the idea that you just want to make mortgages because that's how you got paid. You got paid by sourcing mortgages and making them and then hopefully the risk is worn by the next guy.
21:51 So I think in private credit too much money flooded in too quickly. It has been deployed — a lot of it still dry powder, hasn't been deployed — but it was in some cases deployed poorly, and there is a default cycle that is probably going to be far worse than what the backward-looking numbers look like. Private credit looks great if you look at the historical data because frankly you're not looking at the market today then.
22:18 Now, it's just a very different market. But that's part and parcel of finance. We want these things to happen. You want cycles. You want ups and downs. You want people to learn their lessons, and they will. And hopefully at some point private credit will dust itself off, learn from this in the same way that securitization has, and come up with a better mousetrap afterwards, and that is actually a good thing that will stick around for a long time.
22:44 >> What is the issue with the mousetrap? And what could be better about the mousetrap — and the leverage. So two, three things that will blow up anybody. I do think you can make a case that private credit could be sold to retail investors, but it has to be done exceptionally carefully.
23:13 And I don't believe in semi-liquid offerings. If you're going to invest in an illiquid asset class that touts liquidity as one of its main selling points, do not do it even in a semi-liquid format. If you invest in loans with a 5-year tenor, then you should be locked up for 5 years. Because retail investors, whatever they say, retail investors can be like people like me or people worth quite a few million or even billion, but people pull their money out when they're
23:44 afraid. And these structures aren't built for that. So I hope more credit migrates from the banking system and into the non-bank financial system, that the non-bank financial system — private credit firms, bond funds and so on — lock up investor money for a bit longer.
24:03 We've built an entire financial system around the idea that one-day liquidity is some sort of god-given human right and it isn't and it shouldn't be. It's actually dangerous. I think even mutual funds should have, my personal view, longer lockups. You should not be able to pull your money out daily because it actually leads to suboptimal outcomes for both you the investor and the fund manager, because they have to make decisions knowing that money can go in and out on any given day. And private credit, that's
24:33 particularly acute. So a better mousetrap — there are many small little fiddles. I would prefer levered investment vehicles that invest in highly levered companies not be levered themselves. So BDC >> zero to zero leverage. >> Yeah. So zero leverage ideally. Again, term leverage, if a BDC sells a 10-year bond and invests in some similar maturity assets.
25:01 Look, it's not ideal but it's fine. >> For the public BDCs, I've done a little bit of research and a lot of it is termed. The BDC is just one more transparent slice of the private credit industry, that quite a lot of institutional investors that put money into private credit. >> As the returns started falling because there was capital gushing in, certain return expectations, they lever up their investments in these funds, and again, done judiciously,
25:32 done carefully with no recourse. Maybe that's smart, but it makes me feel uncomfortable when you basically lever up an investment in a highly levered vehicle. Anyway, so that's more on the institutional side. I hope nobody's borrowing money from the brokerage and yoloing into BDCs, but to each of the road.
25:56 >> And also, there's a reflexive dynamic you referenced, that when money floods into an asset class, it makes returns look really good. So the private credit loans that were made in 2018, a ton of money flooded in in 2022 to refinance those loans. So defaults were very close to zero. So even if on a fundamental basis nothing changes, defaults will likely be higher as inflows go down, which they look like they are going to. >> That's very apparent in the equity market, right? Inflows will push the asset classes up, and in BDCs, or in private credit, there's a different nuance: it's not that the loan value will suddenly go to the moon
26:28 suddenly because there's more money going in, but yes it will give the private credit fund manager far more flexibility in how they manage bumps along the road. But only to a certain extent. It's one of the reasons why we've seen the increase in payment in kind — payment in kind is a completely viable and acceptable and important tool in many cases.
26:54 It is the right one to use for companies growing very quickly but don't want to send cash out the door right then. But I think it's unambiguous that lots of private credit funds have been using PIK as a way of deferring the pain essentially. The danger is of course a lot of these companies, and this is where for example the default cycle comes in.
27:16 It's not just the fact that the defaults have been kept probably artificially low because of the money coming into the market, but also the recovery rates are somewhat, I'd say, fantastical, the assumptions. So typically, let's say in a high yield bond you might get 70, 80 cents on the dollar. Depends on where you are in the cap structure of course, but people have penciled in, I suspect, unrealistic recovery rates when a lot of these companies are not going to have any recoveries whatsoever. Let's say if
27:45 they're in the software industry, where there are no plants and factories and roads and trucks, right? It's just, if the company isn't good, it blows up and there's nothing there for you as a creditor. So it's going to be fascinating to watch. So I tend to be on private credit, stepping back.
28:05 I tend to be on those guys that I think it's going to be a bad default cycle. It's started already, but it's getting masked, but it's not going to be catastrophic. And the asset class deserves to survive and thrive once more once it's been through a few of these. >> Definitely. And in a crisis, I think some of these public BDCs are probably going to go to 30 cents or 40 cents of net asset value.
28:26 And for investors with the stomach, there could be opportunity there. Yeah, completely. Buying dollars for pennies is a classic way of making a killing. The problem is when you time it, of course, and when the BDCs go further, because people always think something can't fall any further than it always can, unfortunately.
28:48 But I'm not yoloing into BDCs, go that way. Robin, everything we've talked about — the off-balance sheet debt, the hyperscalers, Nvidia, very murky, unclear what's going to look like, private credit. What themes are present in there that are present throughout the history of the rise of the bond market and in debt that you wrote about in your book?
29:09 Well, in A Fabulous Debt, I talked a lot about how we associate bonds with safety, and quite often that is true, but sometimes that safety can lull people into a false sense of safety, and they do stupid stuff. And also bonds are just as, maybe not just as, but are also susceptible to these kind of bouts of mania that we see in the stock market most obviously.
29:36 So whenever a transformative new technology comes, that typically manifests itself in both the stock and the bond market, and sometimes the most dangerous development happens in the bond market. A classic case were the canals and banks of the United States in the early 19th century.
29:55 So after independence the US was rebuilding itself. It was building canals. All these states were borrowing money for banks. New York famously started with the Erie Canal, which was just a transform — it was like the Apollo program of the era and it was a huge success and they sold lots of bonds and both the investors and the state made a killing out of it.
30:14 But that encouraged a debt bubble, a bondish bubble, that ended up with half the United States being in default, the individual states all bankrupt, and that was quite a nasty crisis that took some time. >> So it was the state governments that built the canals and they were the borrowers. Okay.
30:31 Yeah, typically. So they looked at what New York had done with the Erie Canal and then borrowed a lot of money on their own balance sheets, because obviously they had very little debt because the United States as a federal government had assumed all the post-revolution debts. They sold lots of bonds to investors in Britain and the Netherlands and Italy and France and Germany and some of the United States, and they started banks, they built canals, they started gingerly to industrialize, but they just borrowed too
31:00 much money. There's actually a great scene in A Christmas Carol by Charles Dickens where Ebenezer Scrooge — it's not unfortunately in the Muppets version, which is my favorite, I watch it with my kids every Christmas — but where Ebenezer Scrooge has a nightmare and he wakes up in a cold sweat because he's had a nightmare that all his securities have been transformed into United States securities, which was Dickens's joke about how US bonds had by then all become worthless. Like half the
31:29 states pretty much had all defaulted and some of them never repaid back their debts ever. So at the time, this is in the 1840s, the US was synonymous with, I guess, Argentina today, like a country that just defaults all the time. But those canals were valuable. Most of the states dusted themselves off and we've never had quite that violent a spate of state or municipal defaults in the United States since then.
31:57 Same thing in the late 19th century with the railway mania. That was just massive. If you talk about AI data centers today, that's a few trillion dollars, but the equivalent, if you scale it to the size of the US economy in the 1870s and 1890s and so on to the present day, we're talking the equivalent — railways issued the equivalent of around $10 trillion of bonds.
32:26 It was the biggest capex explosion in history. And a lot of those railways went bust and investors, quite often again in England and in the Netherlands and France and Germany and Spain and Denmark, they lost their shirts, but the railways were still there. And that literally physically knitted together the United States and transformed the economy.
32:50 Which goes to show that these manias look very painful — after the 1871 financial crash when lots of railways went bankrupt, it caused the collapse of a bank called Jay Cooke, which was the equivalent of JP Morgan going bankrupt today overnight. It was catastrophic at the time and it caused what was long called the Great Depression until the actual Great Depression happened, and we now call the downturn in the 1870s the Long Depression.
33:21 But it still transformed the United States because all those railways were still there and I think it shows manias and financial crisis, although painful, sometimes they're a good thing, that the weird thing is that the optimal number of financial crises is arguably not zero, as painful as they are to live through.
33:41 >> That's an interesting argument probably. Basically to guarantee that there would be no financial crisis, you'd have to have regulation, speculation basically be banned. And I could see definitely the downsides of that. Robin, I understand how someone could get into a mania about a stock.
33:59 They buy the stock at 100 and it goes to 900 and they get extremely emotionally very excited, but just in terms of, I can't wrap my head around a mania, a credit mania. I understand they exist, but the idea of earning SOFR plus 4% on a risky thing, it just doesn't really get my blood pumping. Maybe something's wrong with me.
34:19 >> No, sadly, there have been meme bonds. But there aren't any meme bonds around today. I guess maybe TLT is the closest. >> Oh, yeah. >> Or the levered version of TLT. No. So, back in the day, most bonds were actually perpetual bonds. >> Mhm.
34:40 So they were quite — they lasted until the government or sometimes a company paid them back. Governments especially issued perpetuals and they were quite often sold at a discount. So they were sold at, let's say, 90 cents on the dollar at an interest rate of four, but then of course >> 90 or 1 >> 90, yeah, or 60 cents on the dollar or whatever, right.
35:02 But they were sold at a discount, which is why you could have price appreciation for the bond as well. And this is a different era. People didn't have Bloomberg terminals. It was quite difficult for even some smart bankers to calculate literally what should be the right price for this bond.
35:20 So you'd see bonds trade way above par. We see that in modern day, but it just shows that the bond could go up and down a lot. So people could get quite excited, and in an era where what else could you buy to make money. So let's say if you're in Change Alley, it's the Wall Street of Britain in the 17th and 18th century, and you're buying a bond for a newly independent Latin American country.
35:46 Well, you might be buying, let's say, a Brazilian bond, this new fantastical country you've never heard of called Brazil, but the banks are saying it's fantastically full of potential. You're buying that at, let's say, 50 cents on the dollar. And then you're also getting the coupon. Maybe you're also getting the equivalent of 10 cents a dollar on interest all the time.
36:06 So you're getting that plus the price keeps going up because everybody else is discovering this new country called Brazil. So that's why you can get wrapped into it. In the 19th century there was a famous fraudster called Gregor MacGregor that literally invented an entire country so he could sell a bond, and he just took the money and ran to France.
36:24 But people didn't know better back then. >> Fraud is the business that has the highest profit margin. >> Yes. Very much so. So if you can get away with it, Gregor MacGregor made out like a bandit. But to be fair, his efforts — there are frauds and then there are frauds. This guy invented an entire country.
36:41 He invented a capital, a coat of arms, an entire system of government, geography. Had maps made. He had songs made. He basically invented an entire country out of cloth and managed to trick hundreds of people to literally move to this country and also invest in the country's bonds. But they ended up at something called actually the Mosquito Coast.
37:01 And most of them died there sadly. So quite a tragic end, but the joke is that the difference between tragedy and comedy is time. So hopefully after 200 years we can laugh at the debacle of Poyais and Gregor MacGregor. >> Yeah, I don't think I want to go to the Mosquito Coast. >> No, it's not nice.
37:21 It's somewhere in Guatemala now. But that's where he invented this country of golden honey everywhere apparently. Robin, one thing when you mentioned how many of the canal bonds went bad, how many of the railroad bonds went bad, but ultimately, okay, the old man and the family made the loans and eventually the grandson was able to recover 70 cents on the dollar because he held it.
37:42 It just made me think that people holding the bonds in their closet drawer and then eventually being paid back, that is much more stable than a highly sophisticated financial institution holding these securities on leverage, which is what BDCs are basically, even though a lot of it is term debt. >> Yeah.
38:00 The reason why we always call — financial crisis back in the day used to be called panics because usually it was banks that held these loans, these bonds, and even though a bond is technically designed, it's supposed to be tradable, quite often when everybody wants to sell and nobody wants to buy, well, good luck trading it. And there was no deposit insurance. They had deposited money, they borrowed money themselves. So that's why banking crisis and panics, they were intertwined for a long time. Now it is different. But
38:28 yes, sometimes if you buy something unlevered, you can lose money, but you can only lose what you put into it. And that's why leverage is so dangerous and has shown that again and again and again in every major and minor market cycle. >> So earlier you talked about the great financial crisis 2008 GFC, but you said it wasn't one of your favorites.
38:51 What is one of your favorites in the book >> and why? >> Oh god, it's like choosing my favorite child. I know. It's very difficult. I do like Gregor MacGregor and Poyais. It's just incredible, right? But 1873, the railway mania crash, is epic because it was epic in size.
39:11 It was epic in its casualty. Jay Cooke was the John Pierpont Morgan before John Pierpont Morgan. He was titanic. He was the guy that bankrolled the North's victory in the Civil War. And suddenly he just went bankrupt out of the blue because he'd gone over his skis on transcontinental railway bonds. He'd decided against his better judgment initially to back one of these big companies and it just soured on him.
39:38 So because the mix of the enormous ambition of these transcontinental railways, because they weren't just like one, they were like a series of Apollo programs all happening at the same time, and it did genuinely transform the United States into what we now know today. It used to be a coastal country.
39:56 It was north and on the eastern coast and the west coast and it went up and down, but suddenly it became a country that changed its axis. It was west to east. You could actually travel from California to Maine in a few days at least, or at least a week, rather than months it would take before. So I think the mix of both the economic impact, the political importance — this really did transform, it united the United States physically properly for the first time — and how nasty it ended. It was a gigantic financial
40:32 crisis that we don't remember that much these days, but it was huge almost everywhere. Lots of companies went bankrupt in the United States. Expressions like hobo came from that era because there were so many homeless people and soldiers, unemployed soldiers also after the civil war, that lost their employment at the railway lines.
40:57 So I think that's probably my favorite, but change >> a hobo on the railroad tracks. You can't have that if there's no railroads. >> No, exactly. >> It seems to me like, as speculative as data center buildout is, once the data centers are built, they are producing revenue now. Seems to me that railroads back then were a little bit more expected, like actually literally you have to have a guy putting the wooden tack in and then foot by foot across the entire country, and then it has to be built, then the train has to be built.
41:26 You have to market it before the revenue. That does seem to be a greater endeavor than building a data center, which is now very very difficult and takes time and tons of capital of course, but it seems a little easier. >> Yeah. Don't forget, I think the difference between railways in Europe and the United States is an interesting one, because in Europe railways connected existing towns.
41:49 You built a railway from Liverpool to Manchester for example, or from Berlin to Paris. In the United States, railways built towns. It created entire towns. It created entire states. Bismarck, the city, is literally only named Bismarck as a marketing gimmick for the company that built that railway line, the Northern Pacific, as a marketing gimmick to appeal to German investors, because the chancellor of Germany at the time was called Otto von Bismarck.
42:19 And these were in the middle of nowhere. As you know even better than me, the United States is a vast country and back there very little of it was settled. So it's incredibly hard work. Obviously, this is manual. People had to literally hammer down the nail.
42:38 They had to dig out the roads. You have to keep it smooth as well, right? So it's backbreaking work. Then there are all the ravines, the mountains, the forest, everything you have to go. And this in the middle of nowhere, it was lethal. Thousands of people died. As much as the data center construction is pretty epic today, I am not aware of mass casualties in the construction of a data center in New York yet, and this was the equivalent of building
43:09 the pyramids essentially. Very epic, hugely dangerous and incredibly lethal, but transformative in the long run. >> Yes. And you're using the word epic in the British sense, or the way the British people use the word great. Like it doesn't mean that it's a good thing. It just means it's big at scale. >> Oh yeah.
43:28 No, I think the railways are good. Pyramids, lots of slave labor there as well. In the railways, a lot of it was free workers, but not always. And they were treated incredibly shabbily, especially like lots of workers were imported from China for example and would basically be killed in the thousands.
43:49 Lots of Irish workers, and it was a positive thing in the long run, but not quite up to modern labor standards, put that way. But yes, epic in the titanic country-transforming projects that unfortunately do sometimes always have a darker side as well. >> So a few months ago I interviewed Liaquat Ahamed, the author of the book 1873.
44:23 A few months later I'm listening to the Microsoft earnings call and CEO Satya Nadella says we at Microsoft, the executive team, we're reading 1873, so we're thinking about this. So let's say in a few months the next Microsoft call, the team, they say we've been reading A Fabulous Debt: The Epic Story of How Bonds Built the Modern World.
44:40 What are some lessons that you think they should know? The people who are spending hundreds of billions of dollars, borrowing hundreds of billions of dollars, and probably according to the off balance sheet lease commitments it's going to be over a trillion, a trillion and a half as you say.
44:55 What are the lessons that they should know? >> Well, Liaquat's book is phenomenal. It's really good. I tackle the railway mania and the US. My book is a bit more US-centric. His is more global and focuses maybe a bit more on Europe and the Gründerkrach in Europe, which is spectacular. >> But I hope people realize that bonds are an incredibly powerful financial technology.
45:23 It's the financial technology. It's loans 2.0. Both banks and bonds were born in Renaissance Italy a thousand years ago. But it's only now really that the bond market has, I'd argue, supplanted the banking system as the dominant credit engine of the global economy. And the reason why actually some of the basic building blocks haven't changed that much over the hundreds of years is because it's incredibly powerful.
45:47 It's fixed interest so you can calculate things easily, and it's tradable, and that gives you — and it's decentralized. It's the original, it's the OG decentralized finance, because you can sell not just to one or two banks, borrow from a club of banks. You can sell bonds to thousands, even millions of investors indirectly.
46:04 That's why you can pool individually tiny pieces of savings into one big gushing river. And that's what the hyperscalers are doing. So I hope a Microsoft or any of these CEOs and CFOs reading it would realize that actually bonds — you can iterate on the fundamental technology and people are and do, but it still works. And that transparency, the sobriety that comes with doing something through public fixed income markets rather than leases, opaque financing arrangements, private credit loans, loans negotiated
46:43 off the side. That comes at a cost. That's flexibility. That's great. But if you have big projects like the railways, the most valuable thing to do is to just sell bonds. The bond market is supremely able to handle that, and has shown that again and again and again, ranging from Napoleonic wars, canals, railways and AI centers to today. And I'd much rather that goes into the public fixed income markets than stay in the shadows.
47:18 >> Why is it in the shadows? And you talk about this flexibility. The private property people, they say, oh, our borrowers love flexibility. I don't even really know what that actually means. >> I agree. Flexibility sounds great. Like, yeah, you want flexibility? Yes, definitely.
47:31 You want freedom? Yeah, definitely. But in practice, it comes at a cost. >> Broadly speaking, >> if Microsoft wants to sell 10 billion, wants to borrow 10 billion to build a new data center, what is the cheapest way for a large mainstream public company to do so? Is it to sell to like 10 private credit firms or maybe a handful of private credit firms to do it without a rating, doing it quickly? We can do it opportunistically that way.
47:59 Yeah, sure. But you're definitely going to pay a lot less to borrow by just issuing a plain vanilla for general purposes corporate bond. And the reason why they aren't doing that is because they want to maybe obscure how these companies have become, have gone from being lean mean cash machines into being capex hungry utilities.
48:23 And maybe that pays off. The returns of some of these data centers are pretty phenomenal right now. But they're not doing it for purely financial reasons. And I think flexibility is probably a convenient excuse to hide that this is more about making them seem healthier than they really are. Yeah.
48:48 I think one thing that CoreWeave, the neocloud, is doing is delayed draw term loans. So, oh, you don't have to actually borrow the money until you need it. So it's like a credit line. Yeah, the real king of debt I would say is CoreWeave, that just reported. I've never seen a bigger gap between EBITDA and net income loss.
49:08 It is quite extreme and it's a little railway-like. What do you make of CoreWeave's massive massive borrowings? >> Yes, it's heavily indebted. There is in every cycle one or two or maybe a handful of outliers. Look, I'm not worried about Facebook and Alphabet or Amazon going bust. >> Yeah.
49:30 >> They — yeah. If they take all these liabilities on balance sheet, it's not great for investors. I worry about the financial hangover, but it is fundamentally different in that this is not the error. These companies do have solid real products and they're just shoveling all that money and a bit extra into AI.
49:51 And even if AI somehow goes to zero or nothing happens, I think it's manageable. It's okay. But there will be of course in any cycle a few extreme outliers that just borrowed way too much money, did too many dumb things, were too invested in this or didn't have any other products or fallbacks.
50:10 Essentially, people are still going to be going on YouTube even if Google wastes a few hundred billion dollars on AI data centers. And that's going to save them. With a CoreWeave or some of these other companies, do they have that backup? Maybe crypto mining, I don't know, but I'd worry about those essentially more than I do the big hyperscalers.
50:33 There are a few of the hyperscalers that look a little bit dicier, but broadly speaking, they're probably okay. Yeah, you perhaps referring to Oracle, definitely the most indebted relative to its revenue. >> Yeah, Oracle, in hyperscaler terms there's the rest and then Oracle. Oracle is not a tiny bad company or anything like that, but it's just it doesn't have nearly the financial and corporate heft of the others.
51:03 And it's clearly the weakest of the litter. >> How long do you think this capex bubble burst? I think we all know that this is not going to be infinite. Trees don't go to the sky. There will be a bust, a correction. Do you think it's going to be soon or in a few years? >> Obviously, I have no clue.
51:25 >> People signing the check don't know. Yeah. >> Yeah. It's like you say, trees don't grow to the sky. Capex bubbles can continue for a long time until it becomes very obviously unmanageable. Right now there are a lot of people in that industry and this is maybe both what worries me but also can keep the show going for a lot longer.
51:43 There are a lot of people now with a vested interest in keeping this going, that the AI industry has become remarkably incestuous with just an incredible tangle of financing agreements, co-investments, supplier and customer relationships that bind it all together but also can keep things going for a long time.
52:09 They all have an interest in managing this and that makes me worried about what the endgame funding looks like. But it also means it can continue for a while longer. And then I guess the chicken answer: it just comes down to the technology, to what extent AI genuinely is transformative. Is it glorified chatbots or is it going to cure cancer? Is it going to put people on Mars? And where we fall on that spectrum is probably what's going to decide just how much those investments pay off. But the scale is
52:46 pretty astonishing now. [snorts] Okay. So, as someone who is a journalist and is talking to people all the time and is very well informed, what are you hearing about how the revenue is at OpenAI and Anthropic? I think that literally over half of what matters is that topic. Are you hearing good things or bad things or medium things? >> So I haven't spoken to anybody directly about the revenues at OpenAI and Anthropic.
53:13 So I only know what my colleagues have reported in the paper. I think it's broadly understood that Anthropic looks financially a lot healthier than OpenAI. And that's one of the reasons why they're probably going a little bit more aggressively for an IPO now. But I'd question with private companies how real sometimes revenue is — not like fending numbers but just, if you just look at the net income of some of the hyperscalers, the public companies now, look at how much is actually classified as other income,
53:51 which is essentially revaluations of their investments. Like yes >> a lot of the money that Microsoft and Google and Amazon have are basically marking up the value of their stakes in Anthropic, OpenAI and SpaceX and other companies. If you take that away, then some of those earnings look a little bit not bad, but definitely not as good.
54:12 >> Yes. >> And with OpenAI and revenue, how much of that is actually cash, like free cash by rules, everything? And until I've seen the accounts, I don't know. Even when we've seen the accounts, sometimes it's hard to know. But I can tell you I'm really looking forward to the S-1s for OpenAI and Anthropic.
54:31 That's going to be a popcorn moment for me. >> Definitely. Robin, how durable do you think the credit rating agencies are? So, Moody's, S&P, Fitch, the former two, which are publicly traded companies and up until recently were viewed by the Compound Bros, the hedge funds, as these extremely durable businesses.
54:52 Their valuations have fallen a lot because, oh, AI could displace them. I just wonder, you having spent so much time researching and writing this book, just your insight on the value or lack thereof, like is it just a sticker that really isn't that valuable over time? Do you think, in a giant credit cycle, are people going to be like, oh my god, I need my Moody's rating before I buy it?
55:15 >> So that's a great question actually, because I spent a lot of time, a depressing amount of time, writing the book thinking about this. There's an entire chapter just on the history of the rating agencies and it is weird how many cowpies they've stepped in over the years and how they endure. And I think that's the secret to answering your question: that yes, I don't know about 2070, that's a long way off, but I think people will be shocked at the resiliency of their business model
55:47 because people don't actually pay S&P and Moody's and Fitch for their credit work. It's not like if you're the CIO of PIMCO and you sit there, well, I'm going to look at what Moody's says about this bond. You care about the rating for investment mandate reasons.
56:08 But the credit work you do yourself and that's clearly, like with AI, a lot of that is happening. A lot of that happened before the current excitement about large language models. I've been covering AI before it was cool — natural language processing and machine learning. I used to cover quants all the time and it was fascinating to see how people were learning to automate the ripping apart of a credit prospectus and putting it into your own models and then automating all that.
56:37 And this was 10 years ago. But the rating matters, not because you want Moody's to tell you what to think of this investment, because they famously don't try to give investment advice. They just give a probability of default. The credit rating agency ratings are a lot better than people think.
57:00 There are outliers when people say, well, this company was rated A half a year ago and it went bust. But they are highlighted because they're actually pretty rare. Broadly speaking, the letter-based model as a signifier of chance of default is actually pretty accurate. Even the financial crisis, all those shoddy securitized monstrosities that were given AAA ratings — well, actually even quite a lot of the AAA tranches ended up being money good.
57:28 They traded down maybe to 20 cents on the dollar, but a lot of those actually were pretty okay. >> And AAA companies, AAA governments — there aren't that many of them around these days, but it tends to work. And I think the reason why the rating agencies actually endure, will continue to endure, is because of this phrase I once came across, somebody in the industry used, but he talked about the need for a language of credit. Like we need shorthand. We're humans. We're
57:59 both very smart and very stupid at the same time. And we like these shorthands. We like rules of thumb. We like simplistic models. And it's just nice to have something like, this is a single B, that's a double A, that's a triple C. And the reason why the rating agencies, despite having very different — they talk up all their difference, they still have basically the same letters as well.
58:26 And that's because it gives us a cohesive language to talk about credit. And sometimes it's wrong, like all language can be, all models don't work. There's a famous British statistician who said that all models are wrong, but some are useful. The rating agency models are not as wrong as people think [snorts] and it's still pretty useful.
58:48 And as much as you can automate all sorts of cool with AI, I think that will endure, and the craving for just a brand, a name like a Moody's or an S&P, is going to stay there. And in fact, in places like the United States it's enshrined in law. Despite all the controversy around the financial crisis, the nationally recognized rating agency designation is still there.
59:13 It's still in the books. And that's why it's one of the most stubborn oligopolies in the history of business probably. >> And if you're an insurance company buying something, you have to buy something, a certain percentage of your assets have to be investment grade, even if the rating is totally wrong.
59:33 And also, I think of the Charlie Munger anecdote about how he was in World War II. And I think he was tracking the weather and he ultimately was saying to his superior like, hey, my forecasts are really bad. You shouldn't be asking me for these forecasts.
59:48 And the military people said, we need these forecasts for our military planning. So even though the forecasts are wrong, we still need them. >> We still need them. We still need something like that. Yeah. It's like so many things in finance that look weird — in the world really, they look weird or dumb or dangerous.
1:00:04 Quite often you still come to this: well, if it didn't exist we'd have to invent it. Ratings, as weird and dumb as they sometimes can seem, we still need something like that, and if they didn't exist we'd have to reinvent them all over again. The insurance issue is quite interesting, drawing back to private credit there. Of course there is always a danger of shopping around for the greatest rating, and broadly speaking the big three have done a pretty good job over time.
1:00:33 Not always, but over time, as much as they could be a little bit more commercial, let's say, certainly before the financial crisis. Broadly speaking, not letting the standards erode too comically far. But clearly, I do worry about so-called private label credits that insurance companies are getting on private credit loans and saying they're investment grade when really the reality is, I suspect, a lot iffier.
1:01:03 >> What does an investment grade private loan really mean? >> I think most of the big serious insurance companies are very aware of this issue and are aware of it, and if they do use private label credit ratings, they take it with a pinch of salt or they know the issues, here be dragons maybe.
1:01:27 But there are also a lot of private insurance companies that are owned by private equity. >> Yes. >> And those private equity insurance companies sometimes own also some of these private label companies. And I do worry about the tangled private capital ecosystem of private credit, private equity, private ratings and private-equity-owned insurance companies.
1:01:52 I think that is something that could at some point bear watching as well. Have you looked into these things called funding agreement-backed notes? >> No, but it sounds amazing. Tell me more. >> It is basically when an insurance company, probably a life insurance company, issues debt, but the debt that they're issuing, they can call it a policy, a life insurance policy.
1:02:16 >> Yeah. >> Yes. No, actually, I do remember reading about this and I was delighted to learn about it. It shows that there is nothing more creative on this planet as a financial engineer who wants to optimize risk and reward and game the system to do so. One of the dangers of my job, journalism, and your job and I guess everybody's job, is that we look very much backwards and it's always cooler to seem pessimistic and cool.
1:02:48 Or this is the next big thing and this is the next co or whatever. >> And luckily those kind of crashes like 2008, they don't happen very often. I actually have literally in front of my desk a little cartoon that shows — it's from 2008 — that shows somebody going onto an airline and the captain comes across the tannoy saying, oh, there's a bit of turbulence. Buckle up.
1:03:14 And there's a passenger who screams, oh my god, we're all going to die. And the passenger next to him says, look, it's a financial journalist. Don't worry. He's just panicking. And I just have it there in front of my desk to remind myself not to always think everything is the next 2008. Not everything is a big crisis.
1:03:32 So these notes, look, I think the optics are bad. The fundamentals are probably not great. Is it going to be a disaster? Probably not because, yeah, it might be bad, but we can't have reward without risk. You can't make money without losing money. That's what keeps the train on the roads.
1:03:56 And people sometimes create stupid things, invent new things, game the rules, and they get their faces ripped off. It blows up in the next downturn. But the good inventions, they survive and they evolve and they thrive. Securitization is one of them. Securitization was a dirty word not that long ago, just a few years ago.
1:04:19 And now we're looking, even the Europeans are talking about it like, oh my god, we wish we had America's mortgage-backed security market. Oh my god, that would be amazing. And I remember when even American politicians were badmouthing it. Stupid things happened in 2008, but we learned from it. And at some point we'll realize what was really stupid that we're doing right now, what was actually just fine and what was just moderately stupid.
1:04:50 And then I get to write a book about it a few years after that. So it's all gravy for me as a financial journalist. >> People should buy the book, A Fabulous Debt: The Epic Story of How Bonds Built the Modern World. Buy it for yourself. Buy it for your kid. Buy it for your parents, grandparents. Thanks so much.
1:05:08 >> No, thanks for having me on, Jack. >> Thank you. Just close the door.