'There's A Lot More To Come' In AI Says Aliaga

Stephanie Aliaga of JPMorgan Asset Management said the AI boom is still in its early stages and that markets may see additional volatility as they find their footing around the technology. The comments are broadly constructive on AI's long-term trajectory but emphasize near-term uncertainty. This is analyst commentary rather than a new market event, so the likely market impact is limited.

Analysis

The important setup is not the AI theme itself but the stage of the cycle: early-cycle technology narratives typically reward infrastructure first, then broaden into software and application-layer names only after earnings visibility improves. That means the next leg is likely to be more volatile and more dispersion-driven, with capital rotating toward the picks-and-shovels ecosystem—semis, networking, power, and datacenter real estate—while crowded “AI beta” software names can de-rate if monetization lags expectations.

Second-order, the market’s current assumption set may be too linear on demand. If AI capex keeps accelerating, the bottleneck shifts from model quality to electricity, cooling, packaging, and grid interconnects, which can create winners outside the obvious AI basket. The risk is that investors have already priced in a smooth adoption curve; any slowdown in enterprise spend, tighter cloud budgets, or evidence of overbuild can trigger a sharp factor unwind in high-multiple growth names over days to weeks, even if the secular thesis remains intact.

The contrarian takeaway is that ‘more volatility’ is often bullish for active dispersion trades rather than directionally bullish for the theme. In early AI cycles, fundamentals usually matter less than revisions and capex guidance, so the best risk/reward is often long infrastructure beneficiaries versus short the most crowded, least profitable AI beneficiaries. Over the next 3-12 months, the opportunity is to own the constraint providers and fade names trading purely on narrative premium.

Key Decisions for Investors

• Long SMH or SOXX / short IGV or the least profitable AI software basket for 1-3 months; thesis is capex monetization arrives first in semis while software multiples are vulnerable to revision risk. Use a 1.5-2.0x upside/downside structure via equal-dollar pair sizing.
• Overweight NVDA, AVGO, and ANET versus high-multiple application software for 3-6 months; the risk/reward favors the infrastructure layer because order visibility and pricing power should persist even if AI sentiment cools.
• Add a basket long in datacenter and power-enablement names such as EQIX, DLR, and VRT on pullbacks over the next 4-8 weeks; these names benefit from the physical buildout regardless of which model/provider wins.
• Buy 3-6 month put spreads on a crowded, unprofitable AI software proxy after strength; the trade is a volatility hedge against narrative unwind if adoption timelines slip or guidance disappoints.
• If wanting pure optionality, express the theme via call spreads in semis rather than outright longs; this captures upside from continued AI capex while limiting damage if the market rotates or a capex pause hits.