Investors Are Nervous About Timing of Potential 'Crash,' BofA Says

Bank of America’s European equity strategy head said the AI capital-expenditure boom is driving the strongest US earnings upgrades seen in 40 years. However, he characterized the AI trade as high stakes, warning that investors are increasingly nervous because prior investment booms have ultimately ended in crashes. The outlook supports continued positioning in AI beneficiaries but highlights elevated valuation and boom-bust risk.

Analysis

The investable question is not whether AI spending remains elevated next quarter, but whether incremental capex converts into cloud/AI revenue before depreciation, power, and networking costs absorb the return. The next 1-3 months favor suppliers with booked demand and constrained capacity—AVGO, VRT, ETN and CEG—over platform owners whose return on invested capital will be judged against a rapidly rising depreciation base. A modest hyperscaler capex miss can therefore de-rate MSFT, AMZN, GOOGL and META even if absolute spending remains high.

The less appreciated bottleneck is electrical infrastructure rather than compute availability. Data-center interconnection delays, transformer shortages, and power procurement can extend the revenue-recognition cycle for AI clusters; that supports VRT/ETN and selected regulated utilities with data-center exposure, but creates execution risk for operators promising near-term capacity additions. Over 6-18 months, the key split will be between companies selling scarce physical inputs and companies competing to monetize broadly available models, where pricing pressure may emerge faster than consensus expects.

Positioning risk is asymmetric: a reduction in capex growth is more damaging than a still-high level of spending is beneficial, because suppliers are priced on operating leverage and elevated utilization. BAC has no direct earnings sensitivity sufficient to justify a standalone response; its relevance is as a sentiment indicator. Falsify the cautious supplier-over-platform view if the next earnings cycle shows accelerating AI revenue or cloud backlog conversion alongside stable capex intensity, rather than merely higher infrastructure commitments.

Key Decisions for Investors

• Maintain a 1-3 month pair: long VRT and ETN versus short an equal-dollar basket of MSFT/GOOGL. The trade captures physical-infrastructure scarcity against monetization/depreciation risk; exit if hyperscaler AI revenue growth accelerates while capex as a share of revenue stabilizes or declines.
• Add AVGO only following evidence that custom-AI networking and accelerator revenue guidance is raised, not simply maintained. Use the next earnings report as the entry gate; a guidance miss or customer concentration disclosure is the thesis break.
• Watch CEG and other power-exposed utilities as a 6-18 month second-order AI allocation, but require contracted data-center load, approved rate-base treatment, and visible funding before initiating. Interconnection delays or regulatory disallowance would impair the expected earnings conversion.
• Do not establish a directional BAC position from this item. Use a sharp underperformance of AI infrastructure versus the Nasdaq over several weeks as a risk-off signal to reduce crowded semiconductor and data-center exposure rather than as a bank-specific catalyst.