The Utilities Select Sector SPDR Fund (NYSEARCA:XLU) was the consensus “AI power” trade heading into 2026. Data centers need electricity, regulated utilities sell it, and XLU packages the largest names in the sector at a rock-bottom 8-basis-point expense ratio. The logic was straightforward, though results have diverged.
Year-to-date through August 7, XLU is up 3.52%, while a narrower fund focused on the equipment and services powering that same buildout has returned roughly 8 times that amount. Investors holding XLU specifically for AI exposure may be getting a different exposure than they originally intended.
What XLU Actually Owns
A sector tracker is what XLU amounts to. The top ten holdings account for 57.92% of assets and are dominated by regulated utilities, with NextEra Energy at 13.59%, Southern Company at 7.47%, and Duke Energy at 7.15%. These are businesses that earn allowed returns on rate base, all set by state regulators.
Rising demand from hyperscalers eventually translates into approved capex and slow earnings growth, but the pass-through takes years. NextEra, the fund’s anchor, is up 6.91% YTD, running slightly ahead of the ETF. The one most directly tied to the AI-power narrative, Constellation Energy at 6.11%, has moved the other way: down 23.38% YTD after last year’s nuclear PPA enthusiasm reset.
For XLU holders, the fund captures utility earnings, which are regulated and slow. It does not capture the manufacturers actually building substations, switchgear, transformers, and gas turbines for data centers.
The Alternative
The Tema Electrification ETF (NASDAQ:VOLT) is up 31.68% YTD and 39.78% over the past year. The gap comes from what it owns. VOLT’s largest positions are electrical equipment makers: Powell Industries at 8.05%, Bel Fuse at 7.39%, Quanta Services at 5.58%, and Eaton at 5.38%. Add GE Vernova at 4.04%, and Vertiv at 3.33%, and roughly a third of the portfolio sits in companies that sell the physical hardware data centers require before a single kilowatt-hour flows.
When a hyperscaler signs a 500-megawatt data center lease, the utility books a future service obligation. The equipment vendor books an order this quarter. That timing gap is why VOLT’s holdings have repriced faster than XLU’s. VOLT still carries some utility exposure, NextEra at 4.09%, AEP at 4.20%, but they are secondary rather than the entire fund. VOLT holds 33 positions with roughly $724 million in net assets as of May 31.
The Tradeoffs
A higher cost profile is what VOLT carries. The 0.75% expense ratio is nearly 100 times XLU’s 0.08%, and on a $10,000 position, that works out to roughly $67 in additional annual fees. The fund also delivers far less income, since XLU pays a 2.74% dividend yield with quarterly distributions while VOLT’s yield stays minimal because industrial equipment makers tend to reinvest cash flow rather than pay it out.
Volatility is higher as well. Powell Industries and Bel Fuse trade more like semiconductor equipment names than utilities, meaning a slowdown in data center orders would cut both ways. VOLT launched in December 2024, so there is no long-cycle track record through a recession.
A straightforward instrument for defensive income and low volatility is what XLU remains. For positions held with that purpose in mind, the fund continues to serve that role just fine. A swap becomes relevant primarily for capital allocated specifically to the AI power theme.
Making the Switch
In a taxable account, selling XLU may trigger capital gains, particularly for holders who bought between 2022 and 2023. XLU has returned 50.2% over five years on a total-return basis. A partial rotation, trimming XLU and adding VOLT rather than a full swap, preserves the income base while adding the equipment exposure. In a tax-advantaged account, the calculus is simpler.
One structure some investors have used when the XLU position is specifically an AI thesis is to reallocate the thematic slice to VOLT while leaving any defensive-income allocation in XLU. That approach preserves the yield and the low fee while adding exposure to the electrification build-out that regulated utilities cannot reach in the next 12 to 18 months.
What to Weigh
What XLU was designed to do is exactly what it continues to do: track regulated utilities and deliver dividend income. Investors who bought it expecting an AI infrastructure vehicle ended up with a different exposure altogether. The roughly 28-point year-to-date gap between VOLT and XLU reflects two distinct exposures to the same underlying trend. Whether the equipment-side lead persists depends on order flow through 2026 and 2027, and if data center capex slows, VOLT will give back more than XLU. Position sizing tends to be weighed against that risk rather than the trailing return.
Contact [email protected] for any questions or corrections.