AI Bubble Shows Signs of Deflating as Wall Street Reassesses Valuations

The artificial intelligence boom that has lifted American stock markets to historic peaks is showing the first visible cracks, according to a growing chorus of Wall Street analysts who see the hallmarks of speculative excess. John Higgins, chief economic adviser for financial markets at Capital Economics, laid out the case plainly this week: the market has entered the late stages of an AI bubble, and his firm's forecast is that it will begin to rupture sometime in 2027. Capital Economics is also predicting a correction—a drop of at least 20 percent from recent highs—in the S&P 500 next year.

The concern centers on a fundamental mismatch between what the market is pricing in and what the underlying economics can support. James Reilly, a senior markets economist at Capital Economics, pointed to the gap between how fast leading AI companies are expected to grow their earnings and how fast the broader U.S. economy is actually expanding. By multiple measures, he said, the valuations have become stretched to a degree that echoes the dot-com bubble of the early 2000s. The firms driving the AI rally may well prove transformative and profitable, Reilly acknowledged, but the market is betting on returns that exceed what the fundamentals can reasonably deliver.

The scale of the capital flowing into AI development underscores both the opportunity and the risk. Goldman Sachs projects that global spending on AI-related projects will reach $1 trillion in 2026 alone, with $581 billion of that in the United States. That torrent of investment has fueled an extraordinary two-year run for technology stocks, as investors have piled into companies positioned at the forefront of the AI revolution, betting on the outsized profits that are expected to follow.

Yet the picture is more complicated than a simple warning. Economists caution that identifying bubbles in real time is notoriously difficult, and that technological revolutions routinely attract heavy investment and speculative fervor in their early phases without necessarily collapsing. Greg Daco, chief economist at EY-Parthenon, noted that distinguishing between legitimate enthusiasm for a transformative technology and irrational excess is not straightforward. New technologies do promise to reshape how we work and live, and that promise naturally draws capital and optimism.

Kenneth R. French, an investment strategist at Dartmouth College's Tuck School of Business, expressed skepticism about the bubble thesis altogether. Investors commonly rush into hot technology stocks long before the full economic impact of a new technology becomes clear, he said, but that does not necessarily mean the current enthusiasm is misplaced. It is entirely possible, French suggested, that in five years people will look back and realize they were too pessimistic about AI's potential, and that the technology will prove even more important than current expectations. AI is already moving the needle on corporate earnings, he noted, which means the market may actually be underestimating its positive effects. "We don't have enough information to judge if these prices are right or wrong, too high or too low," French told CBS News.

A separate concern has begun to shadow the AI investment narrative: warnings from researchers and industry leaders about the risks posed by uncontrolled AI development. Some prominent figures in the field have called for a slowdown in how quickly the technology advances. These safety concerns could weigh on technology stocks regardless of whether the underlying valuations are justified. But Daco drew an important distinction: worries about the lack of proper safeguards around AI are not the same as concerns about whether the technology will generate the profits investors expect. The guardrail question is about controlling the technology itself and preventing excesses in its deployment. The valuation question is about whether the returns on investment will materialize as promised. They are related but separate anxieties, and conflating them obscures what is actually at stake in the market's current reassessment.