PodcastProf G MarketsLiaquat Ahamed

Will The AI Boom End Like The Railroad Crash Of 1873?

The guest compares the AI buildout with past investment booms, arguing that debt-financed expansion and companies’ fear of falling behind could drive capacity beyond demand. He points to historical railroad financing and the 1873 crash as examples of how funding shocks, excess capacity, and tighter credit can damage markets, while noting that crises are hard to time. The discussion also raises current risks, including inflation constraints, limited fiscal room, and the possibility that weaker AI investment or profits could trigger a broader correction.

Artificial IntelligenceAI InfrastructureMarket ValuationsMarket VolatilityPrivate CreditInflation

Preview · 17 more points in the app

Read the Full Summary

Key Takeaways

The guest argues that AI infrastructure financing is shifting from hyperscalers’ earnings toward debt as cash flows become insufficient and some businesses turn cash-flow negative; Nvidia is described as effectively lending to customers, while large language model companies have not shown a near-term path to profitability. He also identifies Chinese open models as a possible source of sudden pressure on AI companies’ profits.

The guest puts projected AI investment next year at roughly $1.5 trillion, or 3–4% of GDP, within the range of earlier U.S. investment booms that added roughly 2–4% of GDP and ended in excess capacity. He raises the possibility that data center construction could create excess capacity and push down token prices.

The guest argues that individually cautious tech companies can collectively overbuild because each fears losing market share by slowing construction. He says large investment booms can crowd out other investment and absorb more than 3–4% of GDP in financing capacity, while the current financing demand is appearing alongside a roughly 150-basis-point rise in U.S. long-term interest rates.

The guest describes the railroad boom as a major shift in bond-market financing: from around 1850, bonds increasingly funded railroads rather than governments, and most railroad funding was borrowed. The speaker says France’s billion-dollar fundraising was 10 times oversubscribed and that the available capital helped fuel bubbles in German and Austrian stocks.

GEO-POLITICS

Local Opposition — The guest identifies local political opposition to data center construction as an unpredictable risk to the buildout. He compares it with the conflict that disrupted Jay Cook’s Northern Pacific financing, while cautioning that historical events can introduce unexpected complications.

“I suppose the modernday equivalent is the political protest against the construction of data centers. So it's not sitting bull, but it's a lot of people not wanting local communities…”—

TECHNOLOGY

Debt-Financed Buildout — The guest argues that debt-financed overbuilding is a key risk in the data center boom. Hyperscalers initially funded construction from earnings, but the buildout’s scale has pushed them toward debt, much of it opaque; Nvidia is described as effectively lending to customers.

“Yes, very much so. Because the key risk in the data center boom is overbuilding and overbuilding finance by debt…”—

Collective Overbuilding — The guest argues that individually cautious investment decisions can still produce collective overbuilding. Each tech company fears losing market share if it slows construction, creating a prisoner’s dilemma that can push total capacity beyond demand.

“It's the collectivity of their decisions that leads to overbuilding because what it would require to prevent overbuilding, it would require some sort of collective action.”—

Investment Scale — The guest estimates that the current investment boom remains below the 5%-of-GDP scale reached in some past booms. He says projected investment next year is roughly $1.5 trillion, or about 3–4% of GDP.

“this current boom is not at 5% of GDP. We're pro, you know, next year we're projected to put in I don't know $1.5 trillion…”—

ECONOMY

Past Investment Booms — The guest says earlier U.S. private-investment booms—in the 1920s, late 1990s, and mid-2000s—added roughly 2–4% of GDP in investment and lasted two to four years. He says these episodes ended with excess capacity, rising interest rates as companies competed for capital, and crashes.

“the 1920s when private investment jumped by 5% of GDP for a few years, the dotcom bubble of the late 1990s, the real estate bubble of the mid 2000s…”—

Crowding and Rates — The guest argues that large private-investment booms can crowd out other investment and absorb more than 3–4% of GDP in financing capacity. He says AI has become the dominant stock-market story, while financing demand is emerging alongside a rise of about 150 basis points in U.S. long-term interest rates.

“Essentially, it starts crowding out all other forms of investment. And we're already beginning to see the market, you know, the stock market is overwhelmed by AI…”—

MARKETS

Railroad Financing — The guest says the 20-year boom before 1870 coincided with GDP doubling and trade increasing fivefold as railroad construction spread across Europe, the United States, and Argentina. From around 1850, the bond market shifted from primarily financing governments to financing railroads, with most railroad funding borrowed.

“[2:51] GDP doubled in those 20 years. trade went up five times and fueling the boom was a new technology. The railroads and that was a truly global phenomenon…”—

French Fundraising — The speaker says the fine imposed on France did not end the investment boom; speculation continued. France’s billion-dollar fundraising was 10 times oversubscribed, revealing substantial available capital that helped fuel bubbles in German and Austrian stocks.

“[5:11] the Prussians imposed a giant fine on France of a billion dollars which is equivalent to $1.5 trillion today…”—

Rothschilds’ U.S. Exit — The speaker describes the Rothschilds as major lenders to railroads and infrastructure projects worldwide by 1873, but says they did not enter the U.S. market. Their decision followed unrecovered losses on 1840s canal bonds after American states claimed sovereign immunity.

“and by 1873 they were the primary lenders to railroads and other infrastructure projects around the world.”—

FORECASTS

Capacity and Prices — The guest says excess capacity from the railroad boom eroded profitability: by the end of 1872, only one-quarter of railroad companies could pay dividends. He raises the possibility that data center construction could increase capacity and push down token prices.

“The second is by 1873 it is clear that there has been overbuilding and it shows up in the form of excess capacity on railroad lines and a erosion in…”—

Fiscal and Dollar Risks — The guest says the U.S. budget deficit is 6% of GDP and suggests fiscal policy may have limited room to respond. If foreign inflows into U.S. equities reverse, the dollar could face pressure. Monetary easing could worsen a dollar decline and complicate the Federal Reserve’s inflation goals.

“We've we've got a budget deficit that's at 6% of GDP. you know, are they we've probably run out of bullets in terms of fiscal policy…”—

Preview

The episode opens with Liaquat Ahamed’s view that the current AI buildout is already manic, while cautioning that historical comparisons cannot pinpoint exactly when a boom will turn.

Interview with Liaquat Ahamed

The host introduces Ahamed and his book on the 1873 crisis, framing the discussion around whether today’s data-center investment resembles the railroad boom.

How did this boom begin and how did it turn into a bubble?

Ahamed traces railroad expansion to globalization, bond-financed investment, and successive European and American booms fueled by abundant capital after the Franco-Prussian War.

Who were the Rothschilds and Jay Cooke and why were they so central to this story?

Ahamed explains how the Rothschilds financed governments and infrastructure while Jay Cook democratized bond sales and took on the Northern Pacific project.

What triggered the crash?

The crash began with European market declines and intensified when Jay Cooke could not complete Northern Pacific financing, undermining confidence and freezing railroad construction.

How did falling stock prices trigger a depression that lasted for years?

Railroad bond defaults weakened the economy, but Ahamed argues that Germany’s move to gold and the resulting credit contraction transformed the downturn into a prolonged depression.

Ad break

The show pauses for sponsor messages before returning to the interview.

Do you think this is an apt comparison?

Ahamed sees a strong AI comparison in debt-financed construction, opaque financing, possible overcapacity, declining returns, and tighter funding conditions.

Why was it that they landed in that position of overbuilding?

Railroads overbuilt because competing firms each feared losing market share by slowing down; their individually rational choices produced excess capacity collectively.

Does every transformative new technology inevitably lead to a market crash?

Ahamed lists earlier US investment booms that ended in excess capacity and crashes, while emphasizing that their timing is difficult to forecast.

Do you think we have more room to run?

Ahamed says AI investment could continue, but its scale remains below some past extremes when measured as a share of GDP.

Why should we use percentage of GDP to determine whether we’re at the top?

Large investment booms can crowd out other activity and absorb savings, pushing up interest rates and creating broader financial strain.

What were the key warning signs before the crash, and are we seeing similar ones today?

Ahamed cautions against reading warning signs mechanically, pointing to local opposition to data centers and the possibility that AI companies’ profitability will erode.

Is that essentially what happened in the railroad boom as well?

The discussion connects AI firms’ reliance on outside capital to railroad financing; Ahamed says crises often take longer to arrive than expected, then unfold rapidly.

Were there any winners coming out of the railroad crash?

Ahamed says the Rothschilds avoided speculative fads, while JP Morgan and other investors gained by acquiring railroad assets after the crash.

Ad break

The interview pauses for another set of sponsor messages.

How might monetary policy shape this next chapter, and what concerns you most?

Ahamed worries that a large equity market, consumption wealth effects, limited fiscal room, and foreign capital flows could complicate policymakers’ response to a downturn.

Do you agree that AI is the biggest risk facing financial markets today?

Ahamed agrees that weaker AI investment or profits could trigger a correction and questions whether authorities have enough capacity to contain another crisis.

Is there a boom-bust cycle that you think most

Ahamed considers 1873 a useful comparison but notes that the 1929 crisis also offers lessons, especially about policy constraints and the danger of inflation.

Are there any characters in the 1873 story who you think make perfect analogous links?

Ahamed cannot identify a perfect modern counterpart to Jay Cook, but sees hubris and unrealistic thinking across the financial market as important parallels.

What are we supposed to take away from history here?

Ahamed says the lessons depend partly on whether AI proves transformative; a confident company might wait out the boom and acquire assets after a shakeout.

Whereabouts in that cycle do you believe we are in this moment?

Ahamed judges today’s market to be in a manic phase, while stressing that the boom could continue and that its precise endpoint is unknowable.

What might unfold if AI went through a downturn like the railroad industry did?

A downturn could begin when a major AI player cannot raise funding, when competition erodes profits, or when an unpredictable geopolitical event intervenes.

What drew you to history, and how has it shaped your view of the world and markets?

Ahamed says financial losses during the Asian crisis drew him to financial history, where he enjoys telling broad economic stories through individual people.

Credits

The host gives Ahamed’s credentials and book details, thanks him, and closes the episode.

Related Episodes

Important Disclaimers

Not Investment Advice: For education only — not financial, tax or legal advice.

AI-Generated: Made by AI from source transcripts and may contain errors.

No Guarantees: Past performance does not predict future results; investing carries risk.

Our Commentary · Request Removal