What 150 Years of Bubbles Teach Us About the AI Boom

An illustration of the New York Stock Exchange closing its doors on its members during the financial panic of 1873, published in Frank Leslie’s Popular Monthly in New York, circa July 1884. (Photo by Paul Popper/Popperfoto via Getty Images)

Observers failed to see the impending downturns in 1873, 1929, and 2000. And money is pouring into AI at a much higher rate than it did in those market crashes.

Key Points

Historically, major market crashes like those in 1873, 1929, and 2000 were preceded by observers who failed to see the impending downturns.
By 2032, investment in the artificial intelligence buildout is estimated to reach 3.63% of U.S. gross domestic product.
The projected AI investment rate is higher than the 2.24% of GDP spent on the 1870s railroad expansion and triple the dot-com boom.

There have been lots of Irving Fishers through the years.

Fisher was the renowned economist who, on Oct. 15, 1929, said stocks had reached “what looks like a permanently high plateau.” Nine days later the market crashed, dropping almost 90% in three years, and it took 25 years to regain Fisher’s “plateau.”

Fisher isn’t alone. The inability to see the proverbial train headed for the station has featured in every crash since the Panic of 1873, which was actually caused by the bursting of an investment bubble in railroads. The dot-com bust of 2000, the subprime-mortgage crisis of 2007-09—there are more.

Today, with stocks hovering near record highs on soaring investment in artificial intelligence—some say it’s already a bubble —it’s worth looking back on these disasters, and their run-ups, to see what contemporary observers missed. Maybe it will help us see the next train coming.

“Cooke’s Crash,” was the New York Herald’s headline on Sept. 19, 1873. “A Financial Hurricane in Wall Street.”

Jay Cooke was the nation’s leading banker, “having accomplished the labors of Atlas,” as his hometown Philadelphia Inquirer put it Jan. 18, 1872, selling government bonds to finance Civil War efforts. Success was expected to follow as he applied his nationwide sales network—precursor to the modern investment bank —to the national railroad boom.

But he overshot, laying out $1.5 million ($42 million in 2026) for construction of a second transcontinental line, the Northern Pacific Railway, long before returns would be seen.

Jay Cooke & Co.’s bankruptcy was just the start.

“A Flood of Failures,” the Pittsburgh Gazette wrote Sept. 20, 1873. “Runs on Numerous Banks; Business at a Standstill.”

The cause was overexpansion, with national trackage doubling in a matter of years. Railroad investment from 1870 to 1890 accounted for 2.24% of U.S. gross domestic product, according to a Brookings study. That is higher than any single investment, from canals (0.66%) to highways (1.13%), to telecoms (1.10%)—at least until artificial intelligence. More on that below.

The boom brought thousands of newcomers to the market, and many got burned.

“I have nothing left, John. It was all in there,” the New York Herald reported one man telling another in front of Cooke’s Manhattan office.

The bursting of the railroad bubble was followed by a six-year downturn in the 1870s, which was called the Great Depression at the time.

By the 1920s, a newly prosperous America was enjoying the fruits of its industrial revolution. Homes were electrified, telephones installed, and factories churned out waves of consumer marvels.

As the economy roared, the Dow Jones Industrial Average rose sixfold, to 381.17 on Sept. 3, 1929, from 63.90 on Aug. 24, 1921.

“The shelves are not piled up absorbing bank money, and the people are buying from hand to mouth, some days motors, other days long hosiery, other days radios, and other days refrigeration machines,” Clarence Barron, this magazine’s founder, who lived through the 1870s depression, said on June 27, 1927. “[P]eople have the money to follow the fashions and buy what they please.”

Barron joined Fisher and others in dismissing concerns of a bubble.

In fact, today’s Fed estimates that as much as a fifth of the market had been bought on margin. Investors could put just 10% down, while brokers fronted the rest with money borrowed from the banks. The securities served as collateral. You can see where this is going.

When the market dropped, many neophyte investors weren’t prepared for the margin calls demanding more cash. Also squeezed were hardcore speculators, whose money might be tied up with numerous brokers.

As the crash rolled through Wall Street—Black Thursday to Black Tuesday, Oct. 24-29—thousands went bust. When the banks made their margin calls, brokers sold shares into the reeling market, worsening the rout. The system collapsed.

CIRCA 2000: Anxious traders take a break outside the New York Stock Exchange. It was another bad day for stocks with the Dow index sinking below the 10,000 level. (James Keivom/NY Daily News/Getty Images)

The Great Depression that followed all but erased memories of the one in the 1870s. It also prompted generations of economists, from Milton Friedman to Alan Greenspan to Ben Bernanke, to focus on avoiding another depression.

It’s worked, so far, albeit while ushering in an expansionary new era of national debt. Avoiding investment bubbles has proven more difficult.

The 1990s dot-com bubble foreshadowed today’s AI boom, as great pools of capital chased a promising but unproven new technology—the internet—driving company valuations to worryingly high levels.

James K. Glassman and Kevin Hassett played Irving Fisher’s role. In September 1999, months after the Dow surpassed 10,000 for the first time, they published Dow 36,000, arguing stocks were undervalued.

“Tomorrow, stock prices could immediately double, triple, or even quadruple and still not be too expensive,” they wrote.

Within months, the bubble burst. By November 2000, the 280 firms on the Bloomberg US Internet Index had fallen a combined 75% from their respective highs. One-time darlings like Pets.com, WorldCom, and Global Crossing failed.

The tech-heavy Nasdaq plummeted 78% over the next two years. The Dow staggered sideways before cratering to 6547.05 on March 9, 2009, after the subprime-mortgage crisis. It wasn’t until Nov. 2, 2021, that it finally reached 36,000.

Today, money is pouring into AI at an unprecedented rate. Brookings estimates that, by 2032, investment in the AI buildout will amount to 3.63% of GDP. That is half again as much as the railroad expansion of the 1870s, and three times the dot-com boom.

AI has the potential to produce a surge in economic productivity, say supporters, including President Donald Trump, who called it “the oil of the next 20 to 25 years—it’s bigger than the internet.” Detractors say it’s a threat to jobs and maybe even humanity itself.

Yet even if AI does become as big as the internet, it may still leave many losers behind, like Pets.com investors, or the man who had everything in Jay Cooke & Co. That is how crashes go.

Write to editors@barrons.com

参与讨论