Apollo’s Chief Economist Sees No A.I. Payoff Outside Big Tech

Hundreds of billions are flowing into A.I. infrastructure, but Torsten Slok says the wider economy still looks strangely untouched. The Apollo Global Management chief economist argues that the real test is not the headline spending spree, but whether the other 493 companies in the S&P 500 can turn it into better margins, faster growth, and higher productivity. So far, he says, they haven’t.

The numbers he pulls from Bloomberg and Macrobond show a sharp split. Inflation-adjusted revenue per employee at the Magnificent 7 has climbed to about $780,000, after dipping to around $565,000 in 2022 and 2023. At the rest of the S&P 500, revenue per employee has been stuck in a narrow band for years and now sits around $527,000. Profit margins tell the same story: nearly 30 percent for the Magnificent 7, versus roughly 9 to 10 percent for everyone else.

Slok also says the broader economy does not yet show an A.I.-driven productivity jolt. A Federal Reserve Bank of San Francisco measure of utilization-adjusted total factor productivity has fallen to around zero and recently slipped a bit below it. Output per hour is still rising, but only by about 2 to 2.5 percent a year. That is not exactly the stuff of a revolution.

Why the disconnect? One answer is concentration. Using Ramp data, Slok says the top 10 percent of firms account for 95.6 percent of spending on OpenAI and Anthropic, and 99.5 percent of spending on model serving and inference. Ramp’s chief economist, Ara Kharazian, says about 80 percent of OpenAI and Anthropic spending comes from just 1 percent of companies. If the money stays that concentrated, the macro effects stay small too.

Then there is the circularity problem. Among the biggest A.I. spenders, 80.4 percent are venture-backed, compared with 13.1 percent for the median company. That means investor money often ends up cycling through start-ups and back into A.I. vendors. Slok also argues that mainstream companies are treating token bills as a cost item, not a growth engine, especially as they shy away from the newest and priciest models.