Over the past several quarters, corporate earnings in the major equity indices appeared to move in only one direction: up. Investment in artificial intelligence (AI) was a key driver. Many investors now ask whether this is a new bubble—not so much in valuations as in expectations for future earnings growth. It is therefore worth examining more closely how markets might absorb a potential transition from AI’s investment phase to its application phase.
Earnings per share in the S&P 500 have grown much faster than the overall economy in recent years. Much of this increase is directly or indirectly attributable to AI investment. At the same time, valuations—measured by the price-to-earnings ratio—remain within historical ranges, while earnings themselves are well above their long-term trend. This raises concerns that some of this profitability may not be sustainable.
The AI boom has accounted for almost half of S&P 500 earnings growth this year. At the same time, signs are growing that this tailwind may weaken: tighter regulation of AI models, increasing competition—including from China—and higher capital costs. Spending on data centers, chips, and infrastructure is likely to keep rising, but investment growth will probably slow. Higher depreciation could also weigh more heavily on earnings in the coming years. A strong tailwind could therefore become a moderate headwind by 2028.
The exceptional cycle is especially evident in the semiconductor sector: Profit margins are at historically high levels of around 70%. If competition and pricing pressure increase, margins could fall toward their longer-term average—with noticeable consequences for earnings. Valuation gains on equity stakes have also inflated the reported earnings of large technology companies—an effect that is also likely to fade in the coming years.
Does this mean the AI boom is nearing an end? Hardly. The key question is how well markets can navigate the transition from the current phase, in which investment is the main driver of earnings, to a phase in which productivity-enhancing AI applications spread across the broader economy. On average, global corporate earnings most recently rose around 35% year over year—not only in the technology sector, but across countries and industries. This broad earnings base creates capacity for further investment, as well as higher employment and wages.
Our conclusion? The pace of AI investment will probably slow at some point. Given the importance of this spending to the global economy, markets are likely to react sensitively over the next few quarters to news about AI investment, margins, and regulation. Even so, we continue to see attractive opportunities, increasingly beyond the technology sector. Companies using AI to lower costs, automate processes, and develop new business models could form the next cohort of potential beneficiaries.
For investors, there is a strong case for actively managing their AI exposure, diversifying broadly, and looking beyond established “AI leaders.” This may improve the prospects of benefiting from the next phase of the AI cycle—even if the path becomes bumpier at times.
