‘AI Economy’ Driving Earnings Strength, Says JPM’s Caffrey

JPMorgan Asset Management's Jack Caffrey said AI-related spending is broadening the range of companies benefiting in earnings. He cautioned that increased corporate borrowing needed to finance growth could cause "a hiccup or two" in markets over the next few months, balancing constructive AI earnings momentum with near-term financing risk.

Analysis

The relevant inflection is not AI demand but the financing mix behind it. As internally funded capex gives way to debt-funded data-center buildouts, equity dispersion should widen: cash-rich hyperscalers (MSFT, GOOGL, META) can sustain investment without a material WACC shock, while levered infrastructure and data-center developers face a sharper sensitivity to Treasury yields and credit spreads. JPM has limited direct earnings sensitivity; the investable signal is instead a potential rise in underwriting, syndicated-loan and hedging activity if issuance accelerates.

Over the next 1-3 months, monitor investment-grade spreads, high-yield spreads and hyperscaler capex guidance rather than AI narrative indicators. A 25-50 bp widening in IG spreads alongside unchanged capex plans would pressure long-duration AI infrastructure multiples—particularly ORCL, VRT and data-center REITs—before reported earnings deteriorate. Conversely, stable spreads and evidence that incremental capacity is monetizing through cloud backlog or pricing would invalidate the financing concern.

Consensus may be underpricing the distinction between equipment revenue and project economics. Semiconductor and power/cooling suppliers can still report strong orders even as customers’ returns on new capacity weaken; this creates a 6-18 month risk of order deferrals rather than an immediate demand collapse. The near-term setup is therefore a financing-quality trade, not a broad AI short.

Key Decisions for Investors

• Maintain or initiate a 3-6 month pair: long MSFT or META / short ORCL, sized market-neutral. The thesis is balance-sheet asymmetry if funding costs rise; exit if ORCL demonstrates sustained cloud backlog acceleration with capex funded without incremental leverage or if IG spreads tighten materially.
• Do not add broad AI-infrastructure beta after sharp rallies until Q3 earnings establish whether capex is being financed from operating cash flow versus debt. Set a watch trigger at a 50 bp widening in US IG spreads or a capex-guidance cut from two major hyperscalers; either would support reducing VRT and data-center REIT exposure.
• For JPM, treat this as a monitoring catalyst rather than a standalone position. A measurable pickup in debt underwriting and capital-markets fees, without credit-loss deterioration, would be incrementally constructive over the next two earnings reports; widening leveraged-loan defaults or reserve builds would falsify that read.
• If credit conditions remain benign, prefer a selective long basket of cash-generative AI beneficiaries (MSFT, GOOGL, META) over thematic ETFs with greater exposure to levered suppliers. Risk/reward is favorable only if forward capex remains supported by cloud revenue growth, not by successive debt issuance.