Apollo’s Slok says AI’s financial payoff has yet to spread beyond tech

AI investment has not translated into a clear financial payoff across most of corporate America, Apollo chief economist Torsten Slok argued in an October 24 analysis. His test is not whether technology companies are growing, but whether the other 493 companies in the S&P 500 are generating more revenue per worker and higher profit margins.

The divide behind Slok’s warning

The Daily Spark note compares the Magnificent 7 with the remaining companies in the index, which Slok calls the S&P 493. Revenue per worker and profit margins are rising for the first group but remain flat for the second. He uses that divergence to challenge the idea that heavy AI investment is already delivering broad business gains.

Stocktwits’ October 25 account puts inflation-adjusted revenue per employee at roughly $780,000 for the Magnificent 7, against about $525,000 for the other 493 companies. Profit margins are nearly 30% for the first group and around 10% for the second. Those figures describe the performance gap; Slok’s claim about AI’s payoff is his interpretation of the broader indicators.

He also finds no sign of AI lifting productivity growth. The productivity chart uses a Federal Reserve Bank of San Francisco series adjusted for changes in how intensively businesses use their resources. Apollo credits Bloomberg and Macrobond for the company-performance charts, and Federal Reserve and Bureau of Labor Statistics sources for the productivity evidence.

A cost center, not a growth engine

Slok describes a corporate response to that missing payoff: mainstream businesses are pulling back and treating token spending as a cost that weighs on margins, rather than a source of growth. He contrasts them with venture-backed companies spending more on tokens. In his assessment, mainstream buyers neither need the latest model nor want to pay for it.

The spending figures show how narrow the biggest buying activity remains. According to Slok’s data, as reported by Stocktwits, the top 10% of firms account for 99.5% of spending on serving and running AI models, and more than 95% of spending tied to OpenAI and Anthropic. Cloud-computing spending is also concentrated: the top tenth accounts for 95.7%.

The buyers differ, too. Among the top 1% of AI spenders, 80.4% are venture-backed, compared with 13.1% among median AI spenders. Apollo’s spending charts credit Ramp-related sources. Slok’s argument is that intense purchasing within a small slice of businesses does not yet amount to an economy-wide transformation.

As long as agentic AI and token spending are confined to a small part of the economy, the macroeconomic effects of AI will be very limited

Torsten Slok, Apollo chief economist, October 24 analysis

Cheaper intelligence still needs a business result

Looking ahead, Slok asks whether increasingly cheap intelligence can produce higher revenue growth and profit margins as models become more interchangeable. Lower prices alone are not his answer. He argues that business intelligence depends heavily on judgment, which in turn depends on context and can vary with the person or model being asked.

His skepticism extends beyond model prices or technical ability. Slok argues that AI does not understand accountability, trust, culture, unwritten rules, ethical trade-offs or fairness. Those considerations, he says, shape reputational risk and ultimately business results. The unresolved step in his assessment is turning access to intelligence into judgment that improves a company’s bottom line.