The long-standing narrative positioning the Utilities Select Sector SPDR Fund (XLU) as a prime investment for AI power infrastructure, independent of direct AI stock purchases, is now facing scrutiny. This investment thesis heavily relies on the substantial gigawatt-scale interconnection requests utilities have cited to justify ambitious capital expenditure plans.
However, recent actions in Texas have introduced a significant challenge to this outlook. Regulators in the state, through ERCOT's new Batch Zero review, have paused all new data center connections. This directive mandates a thorough assessment of ownership, financing, water consumption, and on-site generation capabilities of these facilities. Simultaneously, a report from Reuters indicated that AEP Ohio's anticipated data center pipeline has been more than halved following the implementation of new application fees. This development is particularly impactful for XLU, given that its largest holdings, including NextEra, Southern, Duke, Constellation, and American Electric Power, have increasingly linked their earnings projections to hyperscaler contracts. Should a considerable portion of these reported pipelines prove to be speculative, the growth prospects for XLU might diminish, potentially reclassifying it closer to a bond proxy, with returns aligning with a 10-year Treasury yield.
The issue of inflated projected demand stems from a practice where a single data center developer may submit interconnection requests to multiple utility providers across various states for the same project, without a firm commitment to commence construction. This leads to an aggregation of requested capacity that far exceeds realistic development timelines. A key differentiator between genuine and phantom demand is financial commitment. Exelon's recent reclassification of its data center pipeline, significantly reducing its high-probability capacity and increasing collateral requirements, illustrates this point. Similarly, AEP's experience in Texas, where it has secured substantial cash or collateral for its Batch Zero load, demonstrates that requiring financial backing effectively filters out less committed projects. Furthermore, for regulated utilities, building infrastructure for projects that ultimately don't materialize poses a considerable risk, as costs could be unfairly passed on to existing customers, potentially leading to regulatory penalties. This prudent approach explains why utilities like Exelon maintain their capital plans based on audited demand, even when overall pipeline figures are reduced. While XLU comprises a diverse portfolio of utility companies, offering broad exposure, the fund's overall performance has lagged behind broader market indices and even alternative utility ETFs like Vanguard's VPU, which offers a similar basket of utilities at a lower expense ratio. For investors specifically targeting AI-driven power demand, direct investment in companies like Constellation or NextEra, which exhibit more concentrated and audited exposure to this growth, might offer a more effective strategy than relying on the generalized exposure of XLU.
The current landscape presents a cautious yet balanced outlook for XLU. While the fund continues to represent critical infrastructure for future power demands, its valuation appears to reflect an optimistic view of projected growth rather than the more rigorously audited figures. For investors prioritizing stable income, alternative options offering comparable utility baskets at reduced costs may be more appealing. Conversely, those seeking direct exposure to the AI power narrative may find greater concentration and potential through individual company investments rather than the diversified approach of XLU.
