Apollo's chief economist: AI has paid off for the Mag 7 and nobody else in the S&P 500
Apollo's Torsten Slok split the S&P 500 in two. Revenue per worker and profit margins are climbing for the Magnificent 7 and flat for the other 493 companies.
Slok's chart set also checks total factor productivity from a San Francisco Fed series and finds no AI lift. His read: mainstream companies now treat token spend as a margin cost, not a growth engine.
The thread's counter was a mill with a water wheel: you can't drop an electric motor into it, you rebuild the building.
Slok's split: Mag 7 on one side, the S&P 493 on the other
In his Daily Spark note, Apollo chief economist Torsten Slok writes that AI "has delivered zero financial payoff to corporate America outside of tech." His evidence is five charts. Revenue per worker (inflation-adjusted to 2019 Q4) and profit margins both rise for the Magnificent 7 and stay flat for everyone else in the index. A San Francisco Fed total factor productivity series shows no sign of AI lifting productivity growth. Ramp data shows AI spending concentrated in a small number of firms, and the Ramp AI Index shows token spending that is partly circular: AI-funded startups paying AI labs.
Slok's explanation is that venture-backed companies keep raising their token budgets, while mainstream firms neither need the newest models nor want to pay for them. He argues that cheaper intelligence may not convert into revenue because business decisions hinge on context: "AI does not understand accountability, trust, culture, unwritten rules, ethical trade-offs or fairness."
The mill that needed a new building
The most upvoted line of defense in the thread of 487 comments was historical.
"You can't just take an electric motor and drop it in in place of the water wheel. It doesn't fit, the wheel is outside, there isn't room between the mill and the wall" — a commenter reconstructing the electrification of grain mills
His conclusion: new factories and companies designed around AI, not retrofits, will move the needle. A second commenter put a date on the lag.
"You can see the computer age everywhere but in the productivity statistics." — Robert Solow's 1987 line, quoted by a commenter who noted electricity and steam took about four decades
Software may simply be the easy case
Zigurd, who runs a tech consultancy argued the tech sector is the outlier for a structural reason: decades of compilers, linters and static analysis already fence in fallible human coders, so agents inherit a safety net other fields lack. Another commenter pushed from the opposite side.
"AIs don't just make code they make words, images, and sounds. Hasn't worked in those markets either." — a commenter replying to the 'most software-defined sector' argument
And one engineer reported the arithmetic from the inside.
"Every time I see a task that LLMs helped me complete in 30% less time, I see 5 more tasks that took me 30-50% longer to complete because of errors, lies, mistakes, hallucinations, context limits." — apercu, a developer who uses LLMs daily
What to check in your own numbers
Slok's charts compare averages across 493 firms, which hides any single company that rebuilt itself around AI. If your margins are flat while your token bill rises, you are in his S&P 493 column; the electrification analogy says the fix is restructuring the workflow, not buying a bigger model.
Get breakdowns like this every day
The top Hacker News stories, unpacked — by email, Telegram, or RSS.
After Musk's Starshield limits on Ukraine, Europe asks which American vendor holds the switch
Steve Blank spent a month in London and heard one question from European policymakers: if one CEO can throttle a military satellite service, who else can?