AI Spending Reshapes Tech Winners

JPMorgan Asset Management’s Gabriela Santos argues the latest CPI data could keep the Fed on hold, while noting AI infrastructure spending is pressuring software companies. She also cautions investors to stay selective as AI stocks grow more volatile and highlights that upcoming tech earnings may influence the next market direction.

Analysis

The main market mechanism here is not the CPI print itself; it is the extension of the “higher-for-longer” regime into a market already crowded in long-duration software and AI-beta names. If real rates stay sticky, the first-order hit is multiple compression in SaaS and enterprise software, while the second-order hit is budget reallocation: enterprise CIOs keep funding AI projects, but they do it by deferring broad-based app spend. That tends to favor the picks-and-shovels layer — NVDA, AVGO, ANET, MSFT, AMZN — over names whose pitch depends on rapid seat expansion or cross-sell acceleration.

For JPM specifically, the setup is more nuanced. A delayed easing cycle supports net interest income near term, but the growth mix matters more than the level of rates over the next 1-3 months: if tech earnings show softer corporate capex or weaker hiring, loan growth and underwriting activity can decelerate even as deposits remain stable. In other words, the bank can benefit from sticky rates while the broader economy transmits the same inflation impulse into slower credit demand later.

The contrarian risk is that consensus is treating AI as one trade when it is really two: infrastructure is still early-cycle, but software monetization is already facing margin pressure from AI-enabled cost bases and longer payback scrutiny. That makes the next earnings round a dispersion event, not a broad index event. If hyperscaler capex guide is cut, or if software companies start showing better AI-driven retention and net expansion, this thesis needs to be reassessed quickly.

Key Decisions for Investors

• Relative value: long JPM / short IGV for the next 1-3 months. Thesis is that sticky rates plus AI budget reallocation favor financials over software. Falsify if software earnings surprise on net retention and billings acceleration while JPM guides down NII.
• Pair trade: long SMH / short IGV into tech earnings. The cleanest expression of AI spend concentration is semis and networking versus application software. Best entry is on any post-CPI pullback in QQQ when software lags the tape.
• Event-driven: buy NVDA or AVGO downside-protected upside exposure into earnings only if implied move remains below the stock’s recent realized move; otherwise wait. This is a dispersion trade, not a blanket AI bullish call.
• Watch item, not a trade yet: if two consecutive large software names cite AI as a margin headwind rather than a revenue tailwind, increase short interest in the software basket and reduce exposure to the broad software ETF.