JPMorgan Says Investors Adapt to `Everywhere' AI Trade

JPMorgan Asset Management’s David Lebovitz says AI has become an “everywhere” trade, emphasizing that the key is how investors structure exposure rather than whether to participate. He argues investors are improving at distinguishing opportunities within AI. Overall, the commentary is more directional than market-moving, with limited near-term quantified impact.

Analysis

The investable shift is not "AI exposure" but AI monetization quality. That usually widens dispersion across the ecosystem: cash-generative enablers with visible capex linkage tend to outperform, while application-layer names with vague ROI promises are vulnerable to multiple compression once investors start asking for payback periods instead of demos.

JPM is a modest relative winner if the next leg of AI spending drives corporate financing, advisory, and capital-markets activity. The bigger second-order benefit is in firms that sit on the supply chain for compute buildout—semis, power equipment, cooling, fiber, and data-center infrastructure—because those revenue streams are easier to underwrite than software seat expansion. If AI spend shifts from "pilot" to "production," the market should reward backlog, margin leverage, and free-cash-flow conversion, not just narrative.

The contrarian risk is that "AI everywhere" is already consensus, so beta to the theme may be less attractive than stock selection within it. Over the next 1-3 months, the key falsifier is evidence that enterprise budgets are not translating into measurable productivity or revenue uplift; over 6-18 months, a capex slowdown or lower cloud growth would punish the crowded winners first. In other words, this is increasingly a dispersion trade, not a thematic index trade.

Key Decisions for Investors

• Relative-value: long JPM / short IGV for 1-3 months. Thesis: banks capture monetizable AI activity through financing, M&A, and trading volatility while the software basket faces valuation risk if ROI evidence lags; target 8-12% spread with a stop if IGV outperforms JPM by ~5% on sustained earnings upgrades.
• Favor infrastructure exposure over application-layer AI: long SMH or a semicap basket, funded by trimming high-multiple unprofitable software names. This is the cleaner way to express AI spend with better near-term revenue visibility and lower multiple risk.
• Use XLU/PWR-type power and electrical-infrastructure names as a 6-18 month AI capex proxy. The bottleneck is increasingly electricity and grid buildout, not model hype; upside persists if data-center load estimates keep rising, but the trade should be cut if hyperscaler capex guides down.
• Set a watch item around upcoming bank and cloud earnings: if JPM or other large-cap financials do not show AI-related fee/transaction uplift, or if cloud vendors slow capex growth, fade the 'AI everywhere' basket quickly. That would invalidate the broad-theme trade within one quarter.