Home
+
Published : 12 Oct 2025, 08:49 PM
Even the tech titans admit we are in a bubble. There will be plenty of losses to go round, they say, but the massive investment in artificial intelligence will leave everyone better off.
In short, what we’re seeing is a “good” bubble rather than the debt-fuelled “bad” version that blew up during the global financial crisis of 2008. Such wishful thinking invariably appears during speculative manias. The truth is that while booms often accelerate the development and deployment of new technologies, they also produce painful busts.
OpenAI boss Sam Altman acknowledges that some people are going to lose “a phenomenal amount of money” when the AI bubble bursts.
Jeff Bezos made a similar point last week. The excitement around the new technology has become so great, says the Amazon.com founder, that both good and bad ideas are getting funded. Still, says Bezos, when the dust settles AI will deliver “gigantic benefits” to society.
The notion that certain bubbles are inherently benign is the thesis of the recently published book “Boom: Bubbles and the End of Stagnation” by Byrne Hobart and Tobias Huber. The authors distinguish between that they call the “innovation-accelerating bubble” and those driven by financialisation and easy money. The former, in their view, “is not simply a collective delusion but an expression of a future that is radically different from our own.” Bubbles are useful, the authors say, because they induce people to take more risk.
“Every financial mania,” they write, “requires some suspension of disbelief and unshakeable faith that the idea at its core will pan out…these delusions are more rational than they appear, if only with hindsight.”
Bubbles create their own “reality distortion fields” which bend the real world towards the speculators’ transcendent vision of the future. The speculators’ “fear of missing out” can be seen as the coordinating mechanism that unleashes new technologies on the world.
Over the past 250 years, the world has experienced a succession of bubbles that have transformed civilisation. While the authors of “Boom” focus on the bubbles’ long-term advantages, they ignore the severity of the busts.
The British canal mania of the 18th century, for instance, brought about the commercial crisis of 1793. A quarter of a century after craze for man-made waterways, one in five canals was still unable to pay dividends. Nevertheless, the new waterways enabled coal to be transported more cheaply, helping to propel the Industrial Revolution.
Britain’s railway mania of the 1840s brought even more significant economic changes. The construction of the world’s most extensive rail network eventually ushered in a multi-decade era of prosperity. But the bubble also resulted in a massive misallocation of capital in ill-conceived railway lines and contributed to the banking crisis of 1847, which left investors nursing massive losses. By January 1850, railway shares were down on average 85 percent from their peak.
The great expansion of the US railroad system took place several decades later. Between 1865 and 1873 some 30,000 miles of track were laid. The new railroads created a continental market for American-made goods, helping the United States replace Britain as the world’s dominant economic superpower.
The costs in the intermediate term, however, were immense. In September 1873, a bank heavily involved in railway finance, Jay Cooke & Company, failed. The ensuing financial panic ushered in America’s first “Great Depression” – a downturn that continued for the rest of the decade and was accompanied by high levels of unemployment and widespread social unrest.
The “Roaring Twenties” were a period of extraordinary technological transformation, which saw the rapid spread of electrification, motor cars and radio. This boom was, of course, followed by history’s most famous depression.
The dot-com bubble of the late 1990s followed a well-established path. After the stock market peaked in March 2000, the Nasdaq Composite Index lost nearly 80 percent of its value and remained below its previous high for 15 years.
Amazon was the standout winner among the early crop of listed dot-coms, but its stock fell by more than 90 percent. Although several telecom carriers, including WorldCom, went bust, the fibre-optic cables laid during the boom provided the bandwidth for successful streaming companies such as Netflix and YouTube.
Of all these previous “innovation-accelerating” bubbles the internet mania was followed by the mildest economic downturn. That’s because monetary and fiscal policy provided a buffer.
Federal Reserve Chair Alan Greenspan slashed short-term interest rates to a low of 1 percent by 2002, while the US government’s fiscal balance went from surplus to deficit, providing a stimulus equivalent to 5.5 percent of GDP. Unfortunately, Greenspan’s easy money policy also ignited the credit and real estate booms that sparked the global financial crisis a few years later. Viewed from a medium-term perspective, the aftermath of the dot-com bubble was extremely destructive.
AI could turn out to be more transformative than any of these earlier technology bubbles. However, it is also more speculative. Whereas railways, cars and internet were proven technologies during their bubble periods, the same cannot be said of self-teaching computers.
We are witnessing a multi-trillion-dollar experiment to arrive at artificial general intelligence, where technology reaches human-like levels of reasoning. If this experiment fails, any benefits are unlikely to justify the immense costs.
While earlier bubbles involved the construction of long-lasting infrastructure, the expensive graphic processors that drive the AI revolution are vulnerable to technological obsolescence. Furthermore, the construction of AI data centres is increasingly being funded with debt.
For the moment, AI-related capital spending and the wealth effect from the soaring stock market are boosting US economic growth by around 3 percent of GDP, according to Julien Garran of The MacroStrategy Partnership.
Yet past technology bubbles have stopped abruptly. The authorities are less well-positioned to arrest a future downturn than they were in 2000. The US government is already running a massive fiscal deficit. The Federal Reserve’s balance sheet is still bloated by past bond-buying operations and inflation remains above its target. Investors are as heavily exposed to AI as their forerunners were to internet and telecoms stocks in early 2000. They may be about to discover there’s no such thing as a “good” bubble.
[Edward Chancellor is a Reuters Breakingviews contributor]
