Apollo economist: AI profit gains outside tech may lag market

Apollo chief economist Torsten Slok warned on Jul 7, 2026 that profit margin gains from AI outside the tech sector could take "well beyond" what Wall Street expects. He said AI company valuations rest on the promise of rising margins at S&P 493 companies, and if productivity improvements take five years instead of five months many AI stocks face a painful repricing.

What exactly did Torsten Slok warn?

Slok argues that, outside the tech sector, there is no clear sign AI is boosting profit margins and gains could be delayed "well beyond" market expectations. He singled out regulated industries — healthcare, banking, energy, pharma, and manufacturing — where process overhauls and privacy requirements could slow productivity, and he warned that falling token costs could cap hyperscaler revenue.

He wrote that markets are pricing in fast earnings growth tied to AI, but that real cash flows could trail far behind. Slok framed valuations for AI companies as depending on margin expansion at S&P 493 companies, the S&P index excluding the so-called "magnificent seven."

How does Apollo model slower AI adoption and its impact?

Apollo illustrates a scenario where market-expected earnings and actual earnings diverge through 2029 if AI adoption is slower than expected. In that illustrative scenario the published chart contrasts market-expected earnings (green) and actual earnings (blue) diverging through 2029, showing the timing gap a slower rollout produces.

Slok gives a concrete framing: if the productivity bump takes five years instead of five months, many AI stocks would face what he calls a painful repricing. The scenario points to two sources of mismatch: over-optimistic market timing for earnings growth, and measurement problems in knowledge work where productivity gains are hard to capture on the balance sheet and get absorbed into daily operations.

Why it matters

Valuations built on rapid, economy-wide margin expansion are vulnerable if real cash flows lag. That gap would force investors to reprice AI-exposed equities, especially among the S&P 493 companies Slok highlights. Corporate managers who cannot measure knowledge-work productivity will struggle to convert AI-driven improvements into reported earnings, keeping those gains off balance sheets and delaying the expected uplift.

Apollo’s warning also shifts attention to hyperscalers: falling token costs, Slok notes, could cap their revenue even as adoption grows, limiting one obvious pathway for investors to capture AI-driven topline expansion.

What to watch

Track corporate earnings relative to market expectations and the pace of measurable productivity improvements in regulated sectors named by Slok: healthcare, banking, energy, pharma, and manufacturing. Watch whether firms report near-term margin lifts within months, or whether the timeline stretches toward years, matching Slok’s five-years-versus-five-months contrast and producing the earnings divergence through 2029 shown in Apollo’s scenario.