The AI payoff outside Big Tech is zero so far, according to Apollo Global Management chief economist Torsten Slok. In a note covered by TechCrunch on 25 October 2026, he said AI has delivered “zero financial payoff to corporate America outside of tech.”
That is a bold claim after years of heavy spending. Here is what he actually measured, what the other data says, and where the argument is weakest.
Key takeaways
- Slok looks at the “S&P 493”: the S&P 500 minus Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia and Tesla.
- In that group, profit margins and revenue per worker look flat, and he sees no evidence of a productivity lift.
- Surveys show many workers feel faster, but only a minority of firms report a profit gain.
- Defenders say the delay is about timing and process change, not failure.
What did Apollo’s economist actually say?
Slok argues that AI spending has not shown up in the results of companies that buy it. He wrote that AI “may be a great calculator, but in business, intelligence is mostly judgment,” and that it “does not understand accountability, trust, culture, unwritten rules, ethical trade-offs or fairness.”
This is not a new line for him. Apollo’s own Daily Spark research notes have pushed the same theme for months: the AI capex boom shows up in the sellers’ margins, not the buyers’. TechCrunch adds a disclosure worth knowing: Apollo used to own TechCrunch.
https://images.unsplash.com/photo-1542744173-8e7e53415bb0?ixid=M3wxMTA3MjYzfDB8MXxzZWFyY2h8MXx8b2ZmaWNlJTIwd29ya2VycyUyMGxhcHRvcHMlMjBtZWV0aW5nfGVufDF8MHx8fDE3OTI5NjkyODZ8MA&ixlib=rb-4.1.0&w=1200&q=80&fm=jpg&fit=cropWhat is the S&P 493, and why does it matter?
The S&P 500 is dominated by a handful of giants that build or sell AI. Strip out seven of them and you get 493 companies that mostly use AI: banks, retailers, insurers, factories, utilities. If AI really lifted productivity, you would expect their margins to rise.
Slok says they have not. Think of it like a gold rush: the shovel sellers are getting rich, while most of the miners are still digging. Earlier Apollo work also found that earnings expectations for the biggest tech names rose sharply, while expectations for the other 493 were basically unchanged.
Does other data agree?
Partly. A 2026 McKinsey survey, summarised by MarketScale, found about 80% of respondents say AI boosts their personal productivity, yet only 37% say it adds to their organisation’s EBIT (earnings before interest and taxes). Roughly 6% qualified as “high performers” with at least 5% of EBIT from AI.
A survey of about 750 executives, covered by Fortune, found CFOs reported average AI productivity gains of 1.8% in 2025. Researchers checking real revenue and employment figures found much smaller gains.
| Source | What it found |
|---|---|
| Apollo (Slok) | Flat margins and revenue per worker in the S&P 493 |
| McKinsey 2026 | 80% feel more productive, 37% see an EBIT effect |
| CFO survey (NBER) | Reported gains far larger than measured gains |
Is “zero” too strong? The counter-view
In my opinion, “zero” is a headline, not a law of nature. Aggregate margins are a blunt tool: rising wages, tariffs, interest costs and weak demand can all hide a real gain from AI. Flat margins do not prove AI did nothing; they show nothing visible survived everything else.
Slok himself concedes the timing point. The Decoder reports he expects gains in regulated sectors such as healthcare, banking, energy and manufacturing to arrive “well beyond” what markets project, because of process overhauls and privacy rules. Electricity took decades to show up in factory output for similar reasons: firms had to redesign the work, not just plug in the machine.
Why this matters for the AI trade
Big Tech’s AI spending only makes sense if customers eventually earn more by using the tools. If they do not, the revenue behind the build-out is shakier than it looks. Our look at Nvidia’s role as AI financier shows how much of the boom runs on expected future demand.
It also fits a pattern we covered in the AI graveyard of 2026: many projects stall between demo and real business results. And a study on vibe coding hints that feeling productive and being productive are different things.
https://images.unsplash.com/photo-1554224154-26032ffc0d07?ixid=M3wxMTA3MjYzfDB8MXxzZWFyY2h8M3x8Y2FsY3VsYXRvciUyMGZpbmFuY2lhbCUyMGRvY3VtZW50c3xlbnwxfDB8fHwxNzkyOTY5Mjg5fDA&ixlib=rb-4.1.0&w=1200&q=80&fm=jpg&fit=cropWhat happens next
Watch three things. First, margins and revenue per worker in the S&P 493 over the coming quarters. Second, whether more firms join the small group that credit AI with a meaningful share of earnings. Third, whether the price of AI usage keeps falling, since Slok has flagged that near-zero token costs could leave too little revenue for every large provider.
If those numbers move, the “zero payoff” claim will age badly. If they stay flat, expect louder questions about the spending.
FAQ
Who is Torsten Slok?
He is the chief economist at Apollo Global Management, a private equity firm, and writes the widely read Daily Spark research notes.
Does this mean AI does not work?
No. Many workers report time savings. The claim is that those savings have not yet appeared in company-wide profit margins outside the largest tech firms.
What is the S&P 493?
It is the S&P 500 without Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia and Tesla, used to isolate companies that mostly buy AI rather than sell it.
Next step: check your own team’s numbers. Pick one AI tool you pay for and measure whether time, cost or revenue actually changed.
This article is for information only and is not investment advice.
Leave a comment