AI Capex And Earnings: What Happens When Growth Slows

Goldman Sachs estimates that almost half of 2026 S&P 500 earnings growth will come from AI investment, but forecasts AI-related capex growth slowing from 94% in 2026 to about 17% by 2028. Supplier profits are tied to the year-over-year increase in spending, not its level. Goldman also cautions that imports, record chip margins and roughly $150 billion of unrealized gains mean less of the boom reaches the broad economy than headlines suggest.

Analysis

The key exposure is to the change in AI spending, not its absolute level: as capex growth decelerates, suppliers can face a sharp earnings-growth reset even while customer budgets remain large. That makes near-term semiconductor earnings unusually sensitive to spending revisions and order timing. High margins in constrained components also invite capacity additions and customer efforts to diversify suppliers, creating a later-cycle risk of both weaker pricing and lower utilization.

The macro spillover is less certain than the market narrative: imported equipment limits domestic value capture, while unrealized gains are not equivalent to recurring cash income. The more important 6–18 month test is whether AI adopters produce measurable incremental revenue, productivity, or cost savings that justify continued investment. If not, capex scrutiny could spread from suppliers to infrastructure providers and eventually weigh on broad earnings expectations. Conversely, evidence of attractive returns could sustain spending even as its growth rate normalizes.

Near term, watch hyperscaler capex guidance, supplier order trends and cancellations, lead times, and pricing—not just headline spending totals. A material capex re-acceleration or verified downstream returns would invalidate a bearish supplier-growth view. The cautious angle is not that AI investment stops, but that investors may be extrapolating peak incremental supplier earnings into a slower-growth phase.

Key Decisions for Investors

• Avoid treating the AI-capex theme as a broad-market earnings tailwind; size exposure around the concentrated suppliers and infrastructure beneficiaries whose earnings depend most on incremental orders.
• For a 1–3 month tactical expression, consider defined-risk put spreads on a diversified AI semiconductor or equipment basket only after a rally or into an earnings catalyst, conditional on downward capex or order revisions. Keep risk limited: renewed spending acceleration or stronger-than-expected orders would invalidate the setup.
• Prefer a relative-value watchlist pairing capex-sensitive chip suppliers against businesses able to demonstrate recurring AI-related revenue or productivity gains. Do not initiate the long leg solely on AI claims; verify monetization in reported results and guidance.
• Track capex guidance, supplier backlog quality, cancellation rates, component pricing, and customer return-on-investment disclosures over the next several quarters. If spending growth slows without order deterioration and downstream returns improve, reduce or exit supplier-underweight positions.