Analysts have cast doubt on the profit forecasts of AI companies. What does this mean for investors?

The earnings growth factored into the consensus is driven by capital expenditures on artificial intelligence by the largest cloud service providers, and these investments may slow down, warn strategists quoted by Bloomberg. Excessive bets on AI have made the market vulnerable to sell-offs, and investors may need to temper their return expectations, the agency notes.

“Technology-related earnings have soared simply because U.S. hyperscalers continue to invest record amounts in the AI ecosystem,” Bloomberg quotes Dario Perkins, managing director at research firm TS Lombard, as saying. Perkins expects corporate profit growth to slow in 2027 for three reasons: the positive effect of rising energy prices will fade, growth in AI spending will slow, and the U.S. economy will once again be driven by an increase in employment rather than by gains in labor productivity.

If, in response to the results of the U.S. midterm elections, hyperscalers slow the pace of their capital expenditure growth, analysts will lower revenue and profit forecasts for companies across the entire data center supply chain, warns Joachim Clement, a strategist at investment bank Panmure Liberum. He also noted that investors have “run out of steam” and that few new buyers are entering the AI market, according to Bloomberg.

In recent weeks, stock prices have fallen relative to expected earnings due to emerging doubts about the rollout of AI, Bloomberg notes. Valuations for the Philadelphia Semiconductor Index (SOX) have fallen to the lower end of their 10-year range, and the index’s premium over the S&P 500 has disappeared. At the same time, analysts’ consensus forecast calls for earnings growth of 29% in 2026, 19% in 2027, and 16% in 2028. Following the decline in price-to-earnings ratios, analysts may also lower their consensus forecast, the agency believes.

Following the Democrats’ victory in the midterm elections on November 3, investors are wondering how quickly AI data centers can be built. According to Barclays’ estimates, based on data from energy companies, in 2027, 80% of data center projects with a high probability of implementation will be located in states with Democratic governors, whereas this year that figure is less than 30%. Market participants fear that stricter building permit regulations and rising costs for connecting to power grids will cause chip manufacturers to postpone orders until future years. Because of these concerns, their stocks were at the center of last week’s sell-off, which Bloomberg called the biggest crash in tech stocks since the pandemic.

According to FactSet, the S&P 500’s forward P/E ratio (price-to-expected-earnings ratio) fell to 17.7 on November 5, a more than three-year low. There is a prevailing view in the market that such a decline indicates the absence of a bubble. Andy Constant, founder of Damped Spring Advisors, disagrees: the bubble is not in stock prices, but in earnings forecasts, according to MarketWatch.

There are already early signs that companies may have to lower their forecasts for AI infrastructure spending: spending by Google’s parent company, Alphabet, and Amazon on data centers in the last quarter came in slightly below expectations, according to MarketWatch.

However, Christopher Barrett, a fund manager at Carmignac, does not see any slowdown among data center builders and chip manufacturers. “At the same time, valuations have fallen significantly and look attractive, provided that end-user demand for AI doesn’t drop sharply,” Bloomberg quotes him as saying.

This article was AI-translated and verified by a human editor