AI Dominates 331 S&P Earnings Calls as Payoff Fears Rattle Market

More than 65% of S&P 500 earnings calls for Q2 cited "AI", with the term appearing on 331 calls in the period analyzed by FactSet. FactSet The figures were published on September 18, 2026. The breadth matters for market structure. Discussion of AI was not confined to a narrow cluster of issuers.

The S&P 500 had gained about 12% in 2026 as of September 8, 2026, while remaining down about 1% from its record-high close on August 13. Reuters That left the benchmark close to highs but already off the peak before the latest round of AI warnings. Narrow leadership can sustain an index near highs even as breadth narrows.

AI-linked stocks slumped worldwide on September 14, 2026. Reuters In September 2026, investors were nervous about an AI spending slowdown after industry warnings. Reuters Citigroup warned that safety calls by leading large-language-model developers could crimp artificial intelligence's contribution to corporate earnings. MarketWatch

Torsten Slok is chief economist at Apollo. Slok warned that a slower-than-expected AI payoff risks tipping the economy into recession. MarketWatch The warning connects capex timing to aggregate demand. A pause would affect orders, hiring and credit before it affected model capabilities.

Big Tech companies spent hundreds of billions of dollars over three years to power the artificial intelligence boom. Reuters AI spending is expected to reach nearly $800 billion in 2026. Reuters AI spending is projected to surpass $1.3 trillion by 2027. Reuters Cash leaves first. Earnings follow later, if at all.

The broader context here is duration mismatch between capital outlay and earnings realization. For an expert reader, the 331 mentions are a disclosure signal, not a cash-flow statement. Capex is capitalized and depreciated over time, while equity value discounts expected incremental return on that capital. When deployment timelines extend, near-term free cash flow absorbs the cost and operating leverage is deferred. That can compress multiples even if long-run productivity assumptions are unchanged.

In my view, the market reaction on September 14 fits that mechanism rather than a simple sentiment swing. Earnings-call frequency tells practitioners how widely management teams believe AI must be addressed to satisfy investors. It does not measure conversion into revenue, margin or return on invested capital. Citigroup's focus on earnings contribution and Slok's focus on recession risk point to the same transmission channel from different ends. One is corporate profit, the other is aggregate spending. Both depend on payoff speed.

Looking at what this means for positioning, the variables to monitor are capex guidance revisions, order visibility across the AI supply chain, and evidence of monetization in software and services margins. A slower payoff does not require a cut in absolute spending to tighten financial conditions. A slower growth rate of spending can do it. With the S&P 500 still up about 12% for the year on September 8 and only about 1% below its August 13 high, prior gains leave little cushion if expected earnings growth is pushed further into the future.