S&P 500’s Record +29% EPS Surprise Was Cyclicals-Driven—So the “AI Capex = Future Earnings” Trade No Longer Works the Same

The S&P 500 just logged a record-size earnings beat: FactSet’s Earnings Insight reported companies’ blended EPS came in 29.2% above analysts’ estimates, described as more than four times the five-year average and the largest aggregate earnings surprise since 2008.

But the key “AI trade” check is compositional. When you exclude Alphabet and Amazon, the same earnings-surprise measure falls from 29.2% to 10.9%. In other words: the record outcome is not solely a hyperscaler-related reflex—it’s broad enough that cyclicals/equal-weight effects matter.

A record EPS surprise with high breadth implies profit conversion (cost discipline + demand timing + pricing/mix), not merely higher revenue. In the same FactSet-based excerpt, blended S&P 500 net profit margin is cited as 16.9% (highest in FactSet back to 2009).

That matters for the “AI capex = future earnings” narrative. If margins are expanding broadly at the index level, markets can start paying for earnings quality and near-term free cash flow—especially in sectors where AI is an indirect tailwind rather than the whole earnings model.

The investable implication is a regime shift: if the index-level beat survives removal of the two largest AI beneficiaries in the excerpt, then the next question is whether cyclicals can keep converting demand into earnings.

To anchor the idea in fundamentals, consider Caterpillar, a cash-generative cyclicals proxy. In its latest trailing-twelve-months snapshot, Caterpillar shows net income of ~$11.4B and free cash flow of ~$12.4B, with operating cash flow materially positive. The precise link from the S&P 500 index beat to any single cyclical isn’t disclosed in our accessible primary excerpt—but the direction is consistent: earnings surprises that coincide with healthy cash generation tend to be more durable.

Investor takeaway: instead of asking only whether AI capex “creates” future earnings, you should start asking whether cyclicals’ margin and cash conversion remain stable through the next demand inflection.

Short-term (days to a few quarters): because the surprise is described as record-sized and broad (86% EPS beats in the excerpt), the market is likely to keep re-pricing upward earnings revisions for non-Mag 7 exposure—at least until guidance contradicts.

Long-term (1–3 years): if profit margins stay elevated and beats continue without hyperscaler dominance, the “AI capex is the earnings floor” framework will weaken. The market will revert to fundamentals that explain cyclicals’ operating leverage and cash generation—demand, pricing/mix, and cost discipline.

In practice, this means investors should screen cyclicals for persistent cash flow and margin stability rather than assuming their earnings path is purely a second-order AI outcome.

From the accessible primary excerpt, we verified the core quantitative claims: 29.2% EPS surprise, 7.0% five-year average, the post-exclusion surprise drop to 10.9%, 86% EPS beat breadth, and a cited blended margin of 16.9%.

However, the excerpt we could open did not include a sector-by-sector “cyclicals vs equal-weight” composition breakdown. So this article argues compositionally (AI not sufficient) and mechanistically (margins/cash quality matter), rather than claiming a specific sector count or ranking beyond what the opened source provides.