Investors See AI Slowdown as a Temporary Step

BMO Capital Markets' Jennifer Lee said calls to slow AI development could create near-term market uncertainty, but AI infrastructure spending, investment and demand are unlikely to reverse. Concerns over jobs, data-center buildouts and profitability may moderate the pace of growth, while labor shortages and automation demand should support AI's longer-term economic expansion.

Analysis

The relevant transmission mechanism is not whether AI adoption continues, but where incremental spending migrates if application-layer monetization remains uneven. Hyperscalers can defer discretionary model and software investments while remaining committed to long-lead power, networking and compute capacity; this favors the infrastructure stack—NVDA, AVGO, ANET, VRT, ETN and GEV—over richly valued application beneficiaries whose revenue conversion depends on enterprise ROI. The first-order market risk is multiple compression in AI software, not an immediate collapse in capital expenditure.

Labor scarcity makes automation demand relatively resilient through a softer macro cycle, but the economic surplus may accrue to customers rather than AI vendors. Industrials, logistics and business-process operators with measurable labor substitution potential—ROK, HON, TER, GXO and WNS—could see margin optionality over 6-18 months, whereas software vendors face pricing pressure as models commoditize. The underappreciated constraint is electricity availability: data-center buildouts increasingly depend on grid equipment, turbines and generation rather than semiconductor supply alone, extending the earnings runway for ETN, GEV, PWR and VRT.

Near-term sentiment can still punish the entire AI complex if a major cloud provider signals lower capex or if enterprise software bookings fail to validate AI pricing. That would create a more attractive entry point in power and networking suppliers, whose backlog conversion is generally less sensitive to quarterly AI-product adoption metrics. The thesis is falsified by sustained hyperscaler capex cuts, evidence that data-center utilization is falling rather than merely being reprioritized, or a material easing in power-equipment lead times over the next two quarters.

Key Decisions for Investors

• Maintain a barbell rather than add broad AI-beta exposure: long ETN and GEV versus short IGV over the next 3-6 months. Grid and generation orders have multi-year visibility, while the software basket remains most exposed to AI-monetization disappointment; reassess if hyperscaler capex guidance is cut by more than 10% or if IGV meaningfully underperforms without a booking slowdown.
• Accumulate ANET and VRT only on post-earnings or capex-guidance pullbacks, targeting a 6-12 month horizon. Both capture physical scaling requirements that cannot be solved by lower-cost models; key risk is a customer concentration-driven order pause, so avoid chasing if either gaps materially on unverified AI-demand commentary.
• Express the labor-automation second-order effect through a small long ROK / short WNS pair for 12-18 months. ROK has direct operating-technology exposure to factory automation, while labor-arbitrage business models are vulnerable if AI lowers the value of offshore headcount; exit if industrial production contracts sharply or ROK orders deteriorate for two consecutive quarters.
• Do not initiate directional NVDA exposure solely on this signal. Set an alert around the next hyperscaler reporting cycle: a synchronized capex reaffirmation with stable data-center revenue would support re-risking into NVDA and AVGO, while capex deferrals would favor the power-equipment longs over semiconductor exposure.