Apollo Global Management's chief economist has put a number on the suspicion many finance chiefs have been voicing privately. In a note published October 24, Torsten Slok argued that artificial intelligence has delivered, in his words, zero financial payoff to corporate America outside of tech. TechCrunch reported the note on October 25. His test is deliberately plain: strip the Magnificent Seven out of the S&P 500, and the remaining 493 companies show flat profit margins, flat revenue per worker, and no sign in the productivity data that AI is moving the needle yet.
The gap in his charts is stark. Revenue per employee at the Magnificent Seven has climbed to roughly 780,000 dollars on an inflation-adjusted basis, against about 525,000 dollars for the S&P 493, and margins sit near 30 percent for the big tech firms against roughly 10 percent for everyone else. Summaries of the note also cite Ramp spending data showing how concentrated the buying still is: the top 10 percent of firms account for 99.5 percent of model serving and inference spend. In other words, the companies selling AI are getting richer from it, a small club of heavy spenders is experimenting at scale, and the broad economy is still waiting.
Slok is skeptical that lag alone explains the gap. He argues it will be hard for ordinary companies to turn increasingly cheap intelligence into higher revenue growth and higher profit margins, because, as he put it, AI may be a great calculator, but in business, intelligence is mostly judgment. He lists what models do not supply on their own: accountability, trust, culture, unwritten rules, ethical trade-offs and fairness, all of which shape reputational risk and, ultimately, revenue. That is a harder critique than the usual deployment complaint, because it questions the product, not just the rollout.
The same weekend brought a useful counterpoint from the vendor side. The Information reported that OpenAI is winning coding work from Anthropic customers with steep discounts, including Replit, on a contract worth more than 100 million dollars a year, and Rippling, where OpenAI's share of engineers' coding work reportedly rose from 16 percent in May to 52 percent while Anthropic's fell from 57 percent to 18 percent. Vendors would not be cutting price that hard if buyers treated frontier capability as priceless. It fits Slok's other observation: mainstream companies are starting to view token spending as a cost center, and many do not need, or will not pay for, the very latest model.
For people building with AI, the lesson is not to stop building. It is to stop selling tokens and start selling measured outcomes. The projects that survive a chief financial officer reading Slok's note are the ones wired into a specific workflow, with a baseline, a human who owns the judgment calls, and a line in the budget that the work visibly moves. Pilots that cannot name the margin, revenue or risk number they are supposed to change are exactly the spending now being questioned. Cheap intelligence is abundant. Deployments that turn it into accountable results are still rare, and that rarity is the business opportunity.
