The U.S. stock market hit a record high on a surge in artificial intelligence (AI)-related stocks. But debate is growing over whether the rally is a rational revaluation driven by technological innovation or a speculative overheating reminiscent of past bubbles. Some indicators showed overheating comparable to the period just before Wall Street's 1929 crash or the 1999 dot-com bubble. In particular, as Nvidia's market capitalization more than doubled from its April low this year and at one point surpassed $5 trillion, Central Bank officials and some investors warned that the AI rally could weigh on financial stability.
On the 30th (local time), the Financial Times (FT) of the United Kingdom reported, citing data analytics firm PhaeNon, that the 10-year price-to-earnings ratio (PER) of the Standard & Poor's (S&P) 500, calculated on a cyclically adjusted basis, was higher than just before the 1929 Great Depression and exceeded the level just before the 2008 global financial crisis. Based on long-term data since the 1840s, the only time the stock market was more overvalued than now was during the dot-com bubble, the analysis found.
Some investors argued, however, that current stock prices reflect the potential for productivity innovations that AI could bring. They said expectations that AI adoption could quickly lift corporations' revenue were priced in ahead of time. But forecasts on the productivity gains diverged widely. The U.K. Office for Budget Responsibility noted that while AI clearly has the potential to raise global productivity, its actual impact is highly uncertain.
Studies analyzing how much AI can raise the annual productivity growth rate also showed wide variation. Related research put the productivity increase between 0.1 percentage points a year and as much as 3.4 percentage points. That meant there is both a possibility of only a marginal improvement and a possibility of more than doubling the existing growth rate.
Another concern cited was that concentration in technology stocks within the stock market has risen sharply. The so-called five mega big tech corporations—Amazon, Alphabet, Microsoft, Meta and Oracle—accounted for about 19% of the S&P 500's market capitalization. Adding Nvidia and Broadcom, which have effectively seized control of AI chip supply, pushed the weight close to 30%. The market as a whole has shifted to a structure overly reliant on the earnings of a few corporations.
Since Oct. 2022, when expectations for AI commercialization began in earnest, the ratio of the S&P 500's market capitalization to U.S. gross domestic product (GDP) has jumped from 142% to 214%. During the same period, the weight of technology stocks more than doubled from 44% to 101%. The structure of economic growth has also become more dependent on the technology institutional sector. According to the Organisation for Economic Co-operation and Development (OECD), corporations' investment in information-processing equipment such as data centers accounted for most of U.S. GDP growth in the first half of this year.
Ben Inker, a manager at investment firm GMO, warned, "If expectations for AI wobble, serious shocks could ripple across the market." He said much of today's AI investment thesis rests on conviction, and that conviction will be tested at some point.
There was also a counterargument that historical comparisons do not necessarily lead to a bubble bursting. It was noted that even though market concentration was lower than now during the dot-com bubble, a prolonged slump still occurred. Conversely, during the 19th-century British railway boom, investment and corporate value were far more concentrated than today, but railways ultimately took hold as core infrastructure that changed the economic structure itself.
Gareth Campbell, an economic historian at Queen's University, said, "The railway boom was not just a speculative target, but a technology that changed the entire way people moved," adding, "Countless railway projects were halted at the time and stock prices plummeted, but the railway industry itself remained central to the economy in the long run." He added, "AI likewise may face short-term corrections, but in terms of the technology's fundamental influence, it could follow a path similar to railways."