In his market note published this morning, Apollo's chief economist Torsten Slok delivers a scathing review of the failure of AI to boost profit margins outside of tech... which of course is what AI is supposed to do since it is meant to boost productivity across the entire economy, not just a select group of chipmakers.
As Slok shows in the chart below, so far there are no signs of profit margins rising outside the tech sector. He notes that "this is ultimately what we are waiting for, because the value of AI companies today rests entirely on the promise that margins in the S&P 493 will eventually climb."
As Slok notes, the promise of higher margins for all is the link to current (soaring) market prices, since implicit in the valuations of AI companies are assumptions about future earnings. That's why the current debate about token costs, model routing and token marketplaces is important. If token costs converge toward zero for most AI use cases, then there is not enough revenue for all hyperscalers even in a situation where compute demand surges higher, Slok cautions stomping all over the now traditional "but Jevon's paradox" counterargument. (for more discussion, Slok recommends reading this piece from his colleagues in Apollo Thematic Investing).
Going back to the matter at hand, the key issue is the length of the ROI runway outside the tech sector. In a handful of sectors, software and tech above all, implementation is nearly immediate, since these firms can fold AI into their own products and processes overnight (ironically, it is the same software sector that has been crushed in 2026 due to doubts over the terminal values of ventures which may well be made obsolete by the same AI that is meant to boost their margins).
But that is the exception. Across most of the economy, and especially in capital-intensive, heavily regulated sectors, deep process re-engineering and data governance requirements could delay structural productivity gains well beyond what the market currently projects. The list of slow-moving sectors is long, spanning health care, banking and insurance, energy and utilities, defense and aerospace, pharma and life sciences, manufacturing, transportation and logistics, construction and real estate, education, legal and the public sector.
This, according to Slok, 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.
Put differently, 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.
Slok's bottom line is that a mismatch between current earnings expectations and the actual time firms need to generate ROI on AI investments could have significant implications for many AI company valuations today