Goldman's Hatzius: AI buildout growth peaks late 2026 or early 2027

The two notes land within 48 hours of each other, and both come from economists at firms financing the buildout — Goldman Sachs and Apollo. That timing matters because the premise they challenge is the one that has underwritten three years of spending: build infrastructure now, and a historic wave of corporate productivity will soon pay the tab. Goldman still believes in the long-term payoff, but its path runs through near-term labor-market disruption that weighs on growth before lifting it.

The gap Slok identifies is between the Magnificent Seven and the rest of the S&P 500. His conclusion that the boom's fate will be decided in the S&P 493, not the Mag 7, sits alongside a striking divergence: margins and revenue per worker are rising among the seven, flat among the 493. Outside tech, token spending is being recast as a cost center rather than a growth source, which helps explain why the productivity handoff Goldman assumes is missing from the data.

Both camps are right — the CEOs about the destination, the economists about the distance — and the tension is framed as a classic financial mismatch: short-term capital on an impatient timetable financing a transition whose payoff runs on a multi-year horizon. The supporting evidence points the same way. Only about 6% of companies report a significant bottom-line AI impact, barely 2% of US households pay for an AI service as of April, and among daily AI users, 44% report friction from extra processes, verification, and rework. The authors' own read is that a pullback to reassess is the likely next step and a healthy one.