Apollo Economist Says AI Gains Remain Concentrated in Tech as S&P 493 Sees Little Financial Return

AI investment has produced no measurable financial payoff for major US companies outside technology, according to Apollo Global Management chief economist Torsten Slok. His comparison separates the S&P 500’s technology-heavy leaders from the remaining 493 companies, creating a test of whether AI gains have spread beyond the firms building and selling the technology.

The conclusion is stark: the financial benefits remain concentrated among the so-called Magnificent 7, while the broader group has yet to show comparable improvement in revenue per worker or profit margins. Bloomberg AI, TechCrunch, Finance.yahoo, 24/7 Wall St., and a Threads post reproducing Apollo’s Daily Spark all describe the same divide, though the underlying claim comes from Slok’s analysis.

The S&P 493 refers to the S&P 500 after removing Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia, and Tesla, the companies commonly grouped as the Magnificent 7. Those firms dominate the market’s AI buildout through cloud infrastructure, chips, software, advertising systems, and consumer platforms. Their financial results therefore provide a poor proxy for AI adoption across the rest of corporate America.

Slok’s framework focuses on two operating measures: revenue per worker and profit margins. Both are rising for the Magnificent 7, while both remain flat for the S&P 493, according to the Apollo Daily Spark note summarized by TechCrunch and echoed in the Threads post. The comparison does not show that individual nontechnology companies receive no benefit from AI. It shows that those gains have not appeared in aggregate corporate performance.

The largest early financial gains have accrued to companies selling the infrastructure and services required for AI deployment. Chip makers, cloud providers, platform companies, and digital advertising businesses can monetize demand directly, while companies in manufacturing, retail, finance, health care, and other sectors must first redesign workflows, train employees, manage data, and absorb computing costs.

That gap helps explain why heavy spending has not translated into broad productivity gains in the headline measures Slok tracks. AI projects often begin as pilots, customer-service tools, coding assistants, or internal search systems. Those deployments can improve specific tasks without lifting companywide revenue per employee or margins. TechCrunch framed the finding as a challenge to the broad corporate AI narrative, while Bloomberg AI and 24/7 Wall St. emphasized that the payoff remains confined to technology.

The analysis is a snapshot of financial outcomes, not a verdict that AI has no operational value. A company can reduce response times, improve forecasting, or automate routine work before those changes become visible in quarterly margins. The same metrics also reflect pricing, labor costs, capital spending, demand, and other forces unrelated to AI.

The wording matters because the claim is about aggregate payoff outside technology. Finance.yahoo’s headline presents Slok’s broader argument that the AI story will be decided by the S&P 493, while the Apollo framing compares improving Mag 7 results with flat results elsewhere. The comparison raises a measurement question for executives and investors: whether AI spending is producing durable productivity gains rather than demonstrations, pilots, or added software expense.

Companies outside technology now face pressure to show where AI changes their economics. Revenue per worker and profit margins offer a demanding test because they connect adoption to business performance rather than usage counts or announced projects. If those measures remain flat, investors will ask whether AI spending is replacing older tools, supporting growth, or simply adding to costs.

The result also puts more weight on execution. Firms need usable data, redesigned processes, clear ownership, and enough scale for savings to reach the income statement. Apollo’s conclusion gives executives a practical benchmark: compare AI programs with changes in labor productivity, revenue generation, and operating margins. The broader market will judge the technology through those results, not through the size of corporate announcements.

The next phase of the AI investment cycle will be measured by whether the S&P 493 begins to show the gains already visible among the Magnificent 7. A sustained rise in revenue per worker or profit margins across the broader index would indicate that AI is spreading from infrastructure providers into ordinary business operations.

Flat results would keep the burden of proof on adopters and intensify scrutiny of capital spending. It would also reinforce the distinction between selling AI and using AI: the first has generated visible gains, while the second has yet to produce an aggregate financial payoff outside technology, according to Slok’s analysis.