Author of '1873' Liaquat Ahamed Talks New Book

Liaquat Ahamed

AI infrastructure spending matches railroad-era scale and is largely debt-financed, like the 1870s railroad boom.

He compares the AI boom to the railroad mania. In 1870s America, about 5% of GDP went into railroad construction, which he says would be about $1.5 trillion in today's terms, comparable to the money now going into data centers and AI infrastructure. In both eras the money was mostly borrowed: the US bond market dwarfed the equity market, and about 80% of railroad money came through bonds. He argues the same debt-financed pattern is beginning to show up now.

“Today's terms would be $1,500,000,000,000. So comparable to the amount of money that's going into data centers and AI infrastructure. What was also true then and is beginning to be true now is most of it was debt financed.”at 2:38

Liaquat Ahamed

Circular financing in AI is hidden leverage: great on the way up, terrible on the way down.

He describes circular financing as 'essentially a hidden form of borrowing', which is leverage. He draws on the pattern from past crises: borrowing heavily during a boom is what kills you, because when the cycle turns the debt bites on the way down. He concludes the arrangement is great on the way up and terrible on the way down.

“Circular financing... It's essentially a hidden form of borrowing. So it's leverage and the thing we've learned in crisis after crisis is the thing that kills you is when you borrow a lot of money in a boom... it's great on the way up, and it's terrible on the way down.”at 16:12

Liaquat Ahamed

Yes, AI is in a bubble; timing the burst is hard, and it could run two more years.

Asked whether AI is in a bubble, he says yes, but says the timing is the hard part. Past infrastructure booms ran two to four years before unwinding, and this one is about two years old, so it could last another two years. He points to a pattern: over-investment erodes profitability, as by 1872 only one in five railroads was paying a dividend. He warns to watch for the same erosion in AI as profitability starts to decline.

“Yes. But getting the timing right of when it's going to burst is the giant challenge. Previous infrastructure booms have taken anywhere from two to four years to play themselves out. So I don't know. We're two years into this one. It could go on for another two years.”at 5:24

Liaquat Ahamed

Boom-driven borrowing is pushing yields up; rates rose 150bp in a year, with crowding out globally.

Booms typically push interest rates up, and he says rates have already risen about 150 basis points in the last twelve months. He says this is a global phenomenon, not just American. Everyone is trying to get their hands on capital while governments are borrowing heavily. He argues that the resulting competition for capital also raises real interest rates and crowds out workers and construction resources, which he says is already happening.

“interest rates have gone up by a 150 basis points in the last twelve months and the other thing to remember is this is not just in The US this is a global phenomenon... everyone's trying to get their hands on as much capital as they can... at a time when governments are borrowing an enormous amount.”at 7:30

Liaquat Ahamed

US equity market is at a historically bloated ~250% of GDP, with foreign inflows adding to the overhang.

Equity markets never reached this size in a financial crisis before. The equity market peaked around 110% of GDP in 2000, while the US equity market is now about 250% of GDP. In the last twelve months about a trillion dollars has flowed into US equities from foreigners. He counts private equity as well, noting public equities are only about 75% of the story, which he calls a giant overhang.

“Until 2000, the size of the equity market got up to a 110% of GDP... So The US equity market is 250% of GDP and on top in the last twelve months a trillion dollars has come into The US equity market from foreigners.”at 24:39

Liaquat Ahamed

Monetary policy mistakes, not the bust itself, turn boom-bust cycles into crises.

He argues that the economy is surprisingly resilient to boom-bust cycles, which are terrible but do not necessarily lead to a massive financial crisis. What converts a bust into a crisis is a monetary policy mistake. He cites 1873 as one example and 1929 as another, when authorities let banks go under. He says that did not happen in 2008, which is why that episode was different.

“The economy is surprisingly resilient. What really converts a boom bust cycle into a crisis is a mistake in monetary policy... That's what happened in 1929 when they let the banks go under. That didn't happen in 2008.”at 29:09

Liaquat Ahamed

Bailing out bankers but not homeowners after 2008 fueled left and right populist resentment.

He says bankers were blamed for the crisis but did well afterwards. In 2008 the bankers were bailed out while homeowners were not. That created enormous resentment, which showed up on the right in the Tea Party and on the left in Bernie Sanders and Occupy Wall Street. He says the result is an unholy coalition of left and right saying the system does not work, which he links to the populism seen globally.

“If you go back to 2008, who was bailed out? The bankers. Who was not bailed out? The homeowners... it created enormous resentment... both from the right wing... and the left wing, so Bernie Sanders and Occupy Wall Street, and so you get this unholy coalition of both the right and the left saying this system does not work.”at 32:05