Apollo Global Management (Apollo (APO)) chief economist Torsten Sløk dropped a reality check on Tuesday for anyone betting that the AI boom will keep lifting stocks forever. His message: if the profits from all that AI spending don't show up soon, the market could be in for a rough adjustment.

Sløk said in a post that investors might be getting a little too excited about how fast AI investments will turn into earnings growth. This comes even as U.S. stocks keep rallying on AI enthusiasm, with the NASDAQ 100 on track for one of its best quarters in years.

So what's a "repricing"? It's when investors suddenly change their minds about what stocks are worth, usually leading to big price drops. The current AI bull case assumes that all the massive spending on data centers, chips, and infrastructure will quickly lead to productivity gains, fatter margins, and higher corporate earnings.

"This creates a dangerous divergence between aggressive, front-loaded valuations today and a much slower cash flow reality, since equity markets priced for instant earnings growth will face a painful repricing if the productivity hockey-stick takes five years rather than five months," Sløk wrote.

The AI Spending Spree Is Real—and Huge

Sløk's warning comes as AI infrastructure spending continues to surge across the tech sector. Major hyperscalers including Amazon.com Inc. (Amazon (AMZN)), Microsoft Corp. (Microsoft (MSFT)), Alphabet Inc. (Alphabet (GOOGL)), Meta Platforms Inc. (Meta (META)), and Oracle Corp. (Oracle (ORCL)) are collectively expected to spend about $805 billion in capital expenditures in 2026, according to Morgan Stanley estimates.

Goldman Sachs, meanwhile, estimates AI-related spending is running at an annualized $650 billion and could exceed $800 billion by the end of 2026, with money flowing into chips, servers, memory, power infrastructure, and data centers.

Some bulls argue the spending is justified because demand is already here. CNBC's Jim Cramer recently defended the spending spree, arguing that cloud providers aren't building AI infrastructure on spec—they're racing to meet existing demand.

Not Every Industry Moves at Tech Speed

Sløk, however, argues the biggest risk isn't AI demand itself but how fast AI will actually make money outside of tech. He says there's little evidence that AI is meaningfully improving margins in slower-moving sectors like healthcare, banking, insurance, energy, manufacturing, transportation, real estate, education, and legal services.

"The list of slow-moving sectors is long," Sløk said, warning that AI adoption in these industries may take years to generate meaningful financial returns.

That caution comes as policymakers also watch AI's macroeconomic effects. Cleveland Federal Reserve President Beth Hammack said Tuesday that surging AI infrastructure demand could keep inflation elevated, noting that hyperscalers appear willing to pay "almost any price" for critical inputs.

Sløk also flagged rising AI operating costs as an emerging concern. Companies are increasingly focused on token optimization—reducing AI compute usage to lower costs. "Companies will slow their AI spending if they don't see ROI quickly, and the current focus on token optimization is an early warning that AI implementation could be a bumpier, slower road than expected," Sløk wrote.

In other words, the AI boom might be real, but the payoff could take a lot longer than the market is betting on. And when expectations and reality don't line up, the market has a way of making that gap painfully clear.