U.S. Power Demand Could Set a Record in 2026. 1 Vanguard ETF to Buy to Cash in on the AI Power Surge

Vanguard Utilities ETF (VPU), a $9 billion fund, is positioned as a modest AI-linked play because data-center and commercial power demand are rising with AI adoption. The ETF holds 62.4% in electric utilities, offers a 2.7% dividend yield, and charges a low 0.09% expense ratio. The article argues this could support utility-sector growth, though it remains a conservative income-oriented fund rather than a pure AI investment.

Analysis

The market is underestimating how AI changes utilities from a defensive yield sleeve into a regulated volume-growth proxy. The key second-order effect is not just higher load, but a re-pricing of rate-base growth as utilities need to fund generation, transmission, and grid hardening ahead of demand; that can support earnings expansion even if power prices are politically constrained. In that sense, the cleaner trade is not "utilities as defense" but "utilities as infrastructure capex beneficiaries" with lagged but visible multi-year cash flow visibility.

The biggest winners are likely the vertically integrated regulated names with large service territories and credible capex pipelines, while pure commodity-sensitive power suppliers face a more mixed setup because new supply eventually compresses scarcity rents. AI load growth also pulls forward demand for transformers, switchgear, gas peakers, nuclear uprates, and data-center adjacent infrastructure, so the broader second-order basket may outperform the ETF itself. If commercial demand overtakes residential on the timeline implied, the market will likely move from debating demand existence to debating who gets the allowed return on invested capital.

The main risk is that the current enthusiasm becomes crowded before the actual earnings bridge shows up. Utilities rerate quickly on rate-cut expectations, but if long rates back up or regulators deny faster rate-base recovery, the multiple can compress even as fundamentals improve. The other reversal trigger is load forecast slippage: if hyperscalers defer capex or improve power efficiency faster than expected, the narrative can overshoot by 12-24 months.

Contrarianly, the ETF may be the wrong expression of the theme because it mixes AI-linked utilities with low-beta incumbents whose returns are still dominated by regulation and financing costs. The better risk/reward is to own the nodes where incremental AI demand directly hits earnings: grid equipment, electrical contractors, and select regulated utilities with visible capex acceleration. Income investors may still like the ETF, but for total return the market may be paying up for a broad basket when the alpha is in the bottlenecks.

Key Decisions for Investors

• Long XLU or VPU only as a 6-12 month rate-cut + AI load hedge; prefer using pullbacks rather than chasing strength, since the upside is likely capped by bond-yield sensitivity.
• Overweight grid bottleneck beneficiaries over the ETF: initiate longs in ETN/GEV/EMR on any 3-5% pullback for a 6-18 month trade; these names should capture the earliest capex cycle and have better operating leverage than regulated utilities.
• Build a pair trade: long regulated utility capex beneficiaries vs short a bond proxy basket (e.g., XLU vs TLT overlay) to isolate load-growth upside from duration risk; risk/reward improves if 10-year yields stay range-bound or drift lower.
• For higher-conviction alpha, focus on select utilities with visible data-center load exposure and constructive rate cases; size for a 1-2 year horizon because the fundamental inflection is slow but durable once embedded in rate base.
• Avoid chasing pure power-price beneficiaries unless you have a view on tighter supply; if new generation enters faster than expected, scarcity premiums unwind and the trade becomes a mean-reversion short.