Gains in AI company stakes juice second-quarter earnings for S&P 500

S&P 500 firms are on track for 52% aggregate Q2 earnings growth y/y, led by a 74% profit jump in technology, but LSEG notes the “mark-to-market” component is substantial (Alphabet’s $77.1B and Amazon’s $53.4B unrealized gains tied to AI investments). Excluding those gains, S&P 500 profit growth estimate is 33%—still the strongest since 2021—while strategists warn mark-to-market gains can reverse quickly and that AI infrastructure growth is partly supported by circular financing. Goldman estimates AI infrastructure stocks contribute about one-third of S&P 500 EPS growth in Q2, yet investors are increasingly nervous about valuation and the AI buildout’s cost, with weakness also linked to stress in the U.S. bond market.

Analysis

The market is increasingly rewarding reported EPS that is not cash-operating EPS, which is a subtle but important quality deterioration for the index. That usually matters most when multiples are already rich: once investors start discounting valuation marks and financing-supported demand, the AI complex can go from “earnings leadership” to “earnings illusion” quickly, even if headline growth stays strong. The immediate winners are the platforms with diversified monetization and real optionality; the more fragile beneficiaries are pure-play AI hardware names whose demand is partly being pulled forward by customer financing structures.

Over the next 1-3 months, the bond market is the key transmission mechanism. If rates remain elevated, the market will scrutinize the cost of capital behind AI buildouts, which can compress multiples across NVDA-adjacent beneficiaries before any actual revenue slowdown shows up. Second-order, that should pressure adjacent infrastructure names and any supplier chain that has been repriced on the assumption of uninterrupted hyperscaler capex; the risk is not an earnings miss today, but a reset in forward estimates and order visibility into the next two quarters.

Contrarian view: consensus is treating AI as a clean secular growth story, but the near-term P&L composition suggests some of the strength is balance-sheet-driven and reversible. The trade is not that AI is fake; it is that the market may be overpaying for the timing of monetization versus the timing of spend. Watch 10Y yields, NVDA customer financing disclosures, and next-quarter capex guidance: if yields stay high and capex keeps rising without a commensurate acceleration in operating cash flow, the AI basket likely underperforms broader tech on a 1-3 month horizon.

Key Decisions for Investors

• Short NVDA on rallies or via 1-3 month put spreads; use this as a valuation-and-financing-risk hedge rather than a fundamental call on AI adoption. Falsify if NVDA guides higher on demand visibility and the 10Y yield trend rolls over.
• Pair trade: long AMZN/GOOGL vs short NVDA for a relative-value expression of ‘real platform monetization’ over ‘pure duration AI exposure.’ This works best if rates remain sticky and the market starts penalizing circular financing.
• Set a watch alert on the 10Y Treasury and AI capex commentary from hyperscalers over the next earnings cycle; if yields fall meaningfully while capex stays firm, cover any NVDA underweights because multiple compression risk should ease.