Wall Street may be overestimating how quickly artificial intelligence will boost profits outside the technology sector, according to a new analysis from Apollo Global Management.
Torsten Slok, chief economist at the US financial firm, wrote that there is no evidence so far that AI is lifting profit margins in non-tech industries. The high valuations of many AI companies, he argues, depend entirely on the assumption that AI will eventually improve margins at the S&P 493 companies, the index minus the so-called "magnificent seven" tech giants.
Regulated industries face major delays
Slok singled out sectors where regulation and compliance requirements could slow AI adoption. Healthcare, banking, energy, pharmaceuticals, and manufacturing all face what he called "process overhauls and privacy requirements" that could push productivity improvements "well beyond what the market currently projects."
These industries require significant changes to existing workflows before AI tools can be deployed at scale. Privacy laws and regulatory approvals add further friction. Slok's analysis suggests that the timeline for measurable gains could stretch far past the optimistic forecasts built into current stock prices.
Market expectations versus reality
Apollo's economist noted that financial markets are now pricing in rapid earnings growth from AI adoption across the broader economy. But if the productivity bump takes five years instead of the five months some investors seem to expect, many AI-related stocks could face a significant repricing.
Stay ahead of the AI curve
The most important updates, news, and content — delivered weekly.
No spam. Unsubscribe anytime.
A chart accompanying Slok's analysis illustrates two diverging scenarios. One line shows expected earnings that climb steeply through 2029. The other, reflecting slower AI adoption, shows actual earnings trailing well behind.
The analysis also flagged another risk: falling token costs could cap the revenue of hyperscale cloud providers, adding pressure to the AI investment thesis.
Productivity gains that are hard to measure
Even in cases where individual employees are becoming more productive with AI tools, the benefits are difficult to quantify in knowledge work, Slok pointed out. Without clear metrics, management cannot act on the improvements, and the gains never show up on the balance sheet. Instead, they get absorbed into daily operations.
The Decoder previously covered this measurement problem in its Frontier Radar #3 report.
Apollo is one of the world's largest alternative asset managers with over $500 billion in assets under management. Torsten Slok is a well-known economist whose research is widely followed on Wall Street for its focus on macro trends and market risks.
The warning comes at a time when investor enthusiasm for AI has driven valuations to elevated levels across the technology sector and beyond. If Slok's scenario plays out, the gap between market expectations and real cash flows could widen, leading to a correction.