Why the AI payoff has not appeared
The AI payoff has become a key test of whether the artificial-intelligence boom can extend beyond technology companies. Apollo chief economist Torsten Slok says that, outside the “Magnificent Seven,” the S&P 493 shows no clear improvement in revenue, profit margins, or productivity so far. By contrast, revenue per worker and margins are rising for the Magnificent Seven.
The implication is that large investments and intense publicity have not yet produced visible financial results for mainstream businesses. Venture-backed companies continue to increase token spending, while many established firms do not need the newest model and are unwilling to absorb the cost. They are beginning to treat such spending as a margin burden rather than a growth engine. If agentic AI and token consumption remain concentrated in a small part of the economy, the broader macroeconomic effect will stay limited.
The central constraint on AI productivity
Cheaper or increasingly commoditized models will not automatically create higher revenue or margins. The source argues that AI may be good at calculation, but business judgment depends on context, accountability, trust, culture, unwritten rules, ethical trade-offs, and fairness. Those factors shape reputational risk and influence whether customers are willing to buy.
The S&P 493 is the real test
This assessment does not prove that AI can never generate returns, and a small set of macro indicators cannot invalidate every enterprise experiment. The open question is whether non-tech companies can embed cheaper intelligence into their operations and turn it into sustained revenue growth, stronger margins, and higher productivity. The next test will concern organizational execution and accountability as much as model capability.