Utility ETFs Soar 8% YTD on AI Power Demand. Here's Which One to Buy

Utilities are being re-priced less as defensives and more as claims on scarce grid capacity. The real winners are the names with either merchant exposure to power pricing or enough scale to monetize interconnection bottlenecks; passive baskets dilute that optionality, so the AI premium accrues unevenly. The second-order effect is that capital will keep migrating toward a small set of balance sheets that can finance generation, transmission, and long-dated supply contracts, while smaller regulated names risk becoming slow-growth funding vehicles rather than scarcity assets.

The key near-term risk is rates, not demand. If the 10-year stays near the top of its range, utility multiples can compress even with intact earnings because the sector loses relative appeal versus Treasuries; that is especially dangerous for the higher-multiple, long-duration names. Over 1-3 months, the main catalyst is whether earnings and guidance confirm that data-center load is translating into contracted cash flow rather than just capex; over 6-18 months, the thesis breaks if grid constraints delay load additions or regulators force more of the investment burden onto shareholders.

Consensus is assuming AI load is broadly additive, but the more likely second-order outcome is bifurcation: direct merchant-power beneficiaries outperform while broad ETFs look crowded and less differentiated. XLU is the default expression, but that also makes it the least clean; the alpha is likely in relative value, not outright beta. If the market starts treating utilities like growth assets with rate sensitivity, dispersion inside the group should widen materially.