Goldman and Apollo analysts see AI's macro impact plateauing as broader corporate gains lag

Goldman Sachs economist Jan Hatzius told clients that the growth impulse from the artificial-intelligence boom is likely to peak in late 2026 or early 2027, before recent concerns about the sustainability of AI spending intensify. He cited higher energy prices, rising long-term interest rates and what he called “AI-linked borrowing” as factors that could shave about 0.6 % from U.S. GDP if they persist.

Hatzius noted that AI investment continues to rise, but its share of GDP is expected to slow, reducing the contribution to capital-spending growth. He also said the “AI wealth effect” that added roughly 0.4 percentage points to consumer-spending growth last year is likely to decline, with the pace depending on stock-market performance.

Apollo Global Management’s Torsten Slok argued that, beyond the so-called “Magnificent 7” tech firms, AI has produced no measurable financial payoff for mainstream corporate America. He pointed to flat profit margins and productivity measures for the broader S&P 500 (which he refers to as the S&P 493) and said companies are treating AI expenditures as cost centers rather than growth drivers.

“With no impact on the bottom line, mainstream corporate America is pulling back and starting to view token spending not as a source of growth but as a cost center that weighs on margins,” Slok wrote.

Slok added that venture-backed firms are increasing token-related spending, but most companies do not need the latest models and are unwilling to pay for them. He warned that as long as AI use remains confined to a small segment of the economy, its macroeconomic effects will stay limited.

The two analysts diverge on future outcomes. Hatzius maintains a base-case view that AI will eventually boost productivity and profits, while Slok remains skeptical that commoditized AI can translate into higher revenue and margins for typical firms.

“AI may be a great calculator, but in business, intelligence is mostly judgment,” Slok wrote.

Hatzius also highlighted potential labor-market disruptions, noting a negative correlation between AI adoption and employment growth in non-manufacturing sectors, alongside a positive correlation with unemployment.

Overall, the analysts agree that AI’s contribution to the broader U.S. economy is likely to be modest once the initial hype subsides.