This Boring Utility ETF Just Became a Stealth AI Play, and It's Up 20%

XLU is up about 20% over the past year and 9% year-to-date, trading around $46, while still paying a recent quarterly dividend of $0.31. The fund’s utility-heavy portfolio is benefiting from AI-driven power demand growth, with the IEA calling it a "power demand super cycle" and AEP citing 63 GW of new load by 2030. Offsetting that are rate sensitivity and concentration risk, especially with 10-year Treasury yields near 4% and the sector already re-rated on the AI thesis.

Analysis

The market is starting to re-rate regulated utilities from pure bond proxies into embedded picks-and-shovels on grid scarcity. That re-rating is uneven: names with the cleanest path to data-center interconnection and rate-base expansion should keep compounding, while the sector average may get capped by balance-sheet pressure and long-duration yield competition. In other words, this is less a broad utility bull market than a dispersion trade inside a formerly sleepy sector.

AEP looks best positioned because the market is still underestimating how much incremental load translates into multi-year capex, not just near-term EPS optics. The second-order winner is the utility supply chain: transformers, switchgear, gas turbines, and transmission contractors should see a longer backlog than the utilities themselves, with pricing power persisting for several quarters as utilities scramble to secure equipment. CEG is more levered to the idea that reliable baseload generation becomes strategically scarce, but that also makes it more sensitive to any reversal in power-demand expectations or policy scrutiny.

The main risk is that investors are paying today for load that may monetize slower than expected. Data-center announcements are not the same as completed interconnections, and utility returns are governed by regulatory lag that can easily stretch 12-24 months; if long rates stay sticky or rise another 50-75 bps, the dividend discount argument weakens faster than the AI narrative strengthens. For XLU specifically, the issue is not whether utilities grow, but whether the sector has already pulled forward too much of that growth into valuation.

Consensus is missing how defensive this trade becomes if the AI capex cycle disappoints: utilities still have the dividend floor, while the hyperscaler narrative can be repriced abruptly. That asymmetry argues for being selective rather than owning the basket outright. The cleaner expression is to own the utilities with explicit load backlog and avoid names whose rerating is mostly sentiment-driven rather than tied to contracted capex.

Key Decisions for Investors

• Long AEP vs short XLU for 3-6 months: express the view that load-backed capex names will outperform the sector ETF as the AI grid theme broadens beyond headline sentiment; target 8-12% relative outperformance, stop if 10Y yields break materially lower.
• Buy CEG on pullbacks over the next 4-8 weeks: this is the higher-beta way to play power scarcity and baseload value creation, but size smaller given regulatory and policy risk; look for a 15-20% upside window if demand data keeps firming.
• Use XLU only as an income sleeve, not a momentum chase: accumulate on a 3-5% drawdown rather than after strength, since the sector’s rerating is vulnerable to another 50 bps move higher in long rates.
• Pair long utility supply-chain winners vs short low-quality utility proxies for 6-12 months: favor transmission/equipment beneficiaries over pure yield names, as equipment bottlenecks should support margins before utility EPS growth shows up.
• If using options, sell 1-2 month covered calls on XLU against existing holdings to monetize elevated implied volatility while preserving dividend exposure; this is the better risk/reward than initiating fresh upside calls at current valuations.