AI's missing payoff outside tech is the adoption gap the hype skipped

If AI is the biggest productivity story of the decade, why can't the companies buying it find it in their margins? That is the uncomfortable question behind Apollo's latest note, and we think it deserves more attention than the next model launch.

Apollo Global Management's newest Daily Spark says that, despite massive investment and hype, AI has delivered zero financial payoff to corporate America outside the tech sector. It isn't a one-off. Apollo's team has been returning to the same theme since at least February, and in August it argued that the S&P 493 is spending heavily on AI without any visible lift in profit margins.

The hype measured adoption; the P&L measures results

The reassuring numbers are about activity. Forbes reported that about 69% of S&P 500 companies actively deploy AI, citing an Apollo analysis, and the headline says few track metrics over time. That second half is the story. Deploying something you don't measure isn't a strategy. It's a press release.

A July arXiv paper on S&P 500 adoption helps explain the gap. Using a measure based on SEC filings, it finds only 11% of S&P 500 firms had AI deeply integrated into their business processes in 2025. Technology accounts for two-thirds of that deep adoption. It also finds a "J-curve" in profitability as firms move toward deep adoption, with no differences in capex or productivity. A J-curve means things get worse before they get better. So "no payoff yet" may be partly a timing problem rather than proof of failure.

We think that nuance matters, and Apollo's own framing, an "ROI runway" that "could be long" outside tech, leaves room for it. But a long runway is not an excuse for a blank dashboard.

Value is accruing to suppliers, not users

Apollo's earlier chart work found that over 20 years all of the increase in the S&P 500's operating margin came from tech-related sectors. Non-tech margins stayed near 9%. AI looks like it is repeating the pattern: the people selling the picks and shovels are showing the gains, and the buyers are showing the invoices.

That's our reading, not Apollo's wording, and it's an inference from margins rather than a proven cause. Still, it fits the arXiv finding that tech firms account for two-thirds of deep AI adoption. The question for a retailer or a manufacturer is simple: are you a customer funding someone else's growth, or are you capturing value yourself?

The market stakes are real. As Bloomberg summarized, Apollo chief economist Torsten Slok said Wall Street needs to see positive results on earnings and margins from AI adoption among what he called the S&P 493. In our view, if those results don't arrive, the big-tech valuations that assume they will become a risk.

What boards should do now

Our recommendation is plain. Boards should stop approving AI budgets on the strength of adoption statistics and ask for three things: a baseline, a named metric, and a date by which the project is judged. Cost per ticket, cycle time, margin per product line: anything a CFO can audit.

Caution also has its defenders. JPMorgan's Jamie Dimon, per The Times of India, made clear at an earnings call that the bank will not let external AI agents access its data. We read that as a reminder that the hard part is organizational and risk-related, not just buying licenses.

Our verdict: the skeptics are right about today and probably too early about tomorrow. The evidence doesn't show that AI can't pay off outside tech. It shows that, so far, the proof of payoff is thin. Until companies can show it, treat any claim of AI-driven transformation as a hypothesis, not a result.