JPMorgan's Chang Says AI Capex Story 'Remains Intact'

JPMorgan's Joyce Chang said the AI capex-beneficiary thesis for technology stocks remains "very much intact," signaling continued support for the group. She also cited retail flows showing investors still have a "buy the dips" mentality, pointing to resilient sentiment rather than a fundamental reversal. The commentary is constructive for tech and AI-linked names, but it is primarily market color and unlikely to move the broad market materially.

Analysis

The message for tech is not about another leg of multiple expansion; it is about the durability of earnings revisions in the AI supply chain. If capital spending stays elevated, the first beneficiaries are still the picks-and-shovels: semiconductor manufacturing equipment, power infrastructure, cooling, and networking. The second-order effect is that this broadens leadership beyond the few hyperscalers funding spend, which tends to support index-level tech while compressing dispersion inside the sector.

The flow signal matters because it changes the path dependency of drawdowns. When retail is still buying dips, downside tends to be shallower but more frequent mean-reversion rallies can also delay capitulation, keeping elevated implied volatility in place for weeks rather than days. That favors selling panic spikes in quality AI names rather than chasing breakouts after strong tape days.

The main risk is a capex pause from the hyperscalers, which would hit the entire ecosystem with a 1-2 quarter lag. A softer ad market, tighter cloud optimization, or a board-level shift toward free-cash-flow discipline would be enough to re-rate the trade from "infrastructure scarcity" to "over-earning cycle." The consensus may be underestimating how quickly sentiment can flip if the market starts to believe the AI buildout is no longer accelerating, only sustaining.

For JPM specifically, the relevance is less directional beta and more the trading-adjacent read-through: stronger tech flows keep equity issuance, M&A, and prime brokerage activity healthier, but a crowded dip-buying base can turn violent if volatility spikes. The best setup is to stay long the parts of tech with visible supply constraints and avoid the high-duration software names that need multiple expansion to work.

Key Decisions for Investors

• Long SMH vs short IGV for the next 1-3 months: own AI hardware/equipment over software. Risk/reward favors the trade if capex remains firm and software multiple compression resumes.
• Add on pullbacks in AMAT/LRCX/ANET rather than chasing momentum. Use 4-8 week timeframes; these names should outperform if the AI capex cycle is intact, but trim if hyperscaler commentary turns cautious.
• Sell short-dated put spreads on NVDA into selloffs only when IV spikes above recent ranges. This monetizes the retail-dip-buying bid while limiting downside if the market re-prices AI growth slower than expected.
• Reduce exposure to long-duration software baskets on any sign of capex digestion. Pair long XLK against short IGV if the market starts rewarding cash-flow visibility over narrative-driven growth.
• Set a catalyst watch on next hyperscaler capex guides and cloud utilization commentary. If guidance inflects down, exit cyclically sensitive AI beneficiaries quickly; the reversal can happen within one earnings season.