Utilities Select Sector SPDR Fund
State Street Global Advisors (SPDR)
Tracks the Utilities Select Sector Index — the S&P 500 utilities (electric, multi-utility, IPPs, gas/water), the cheapest and largest way to own US power generation.
Top holdings
Largest holding; renewables + ~30 GW datacenter backlog
Regulated Southeast electric utility
Regulated electric utility
Largest US nuclear fleet; key 24/7 AI-power play
Large transmission; ~63 GW new load cited by 2030
IPP with nuclear; AI-power re-rating name
Performance
July 2026: approx -2.2% (close $45.34 Jun 30 -> $44.35 Jul 31; no July ex-div) — Providers disagree and SSGA is the laggard: its site still shows Jun 30 quarter-end data (YTD +7.62%, 1Y +14.09%) as of Aug 1 — stale, not wrong. On a Jul 31 basis the 1-year roughly halved (+6.4% to +9.1% by source/as-of day) as a strong Jul 2025 rolled out of the window and July 2026 subtracted ~2.2%. Both ranges given rather than picking one.
Recent moves
Re-cast as a 'stealth AI play' on the power-demand supercycle (AEP citing ~63 GW new load by 2030, NextEra's ~30 GW datacenter backlog); nuclear-heavy producers CEG and VST drove the re-rating, though sharp pullbacks showed how fast sentiment reverses, with ~4% Treasury yields a headwind.
Our take
The lowest-cost, lowest-risk, highest-liquidity way to ride AI power demand — at 0.08% you own NEE/CEG/VST/AEP, the regulated and merchant utilities that physically sell electricity to datacenters, but the defensive sleeve and rate sensitivity cap the upside vs the pure-play funds.
The AI-power narrative is real for a slice of XLU but not the fund. ~75-80% of assets are regulated utilities (NEE, SO, DUK, AEP, D) whose earnings are ROE-capped by regulators — datacenter load grows rate base slowly and lags through multi-year rate cases, so demand does NOT flow to EPS the way marketing implies.
The actual 2026 torque came from the merchant/nuclear sleeve — Constellation (~5.7%) and Vistra (~3.5%) — which trade on power prices and hyperscaler PPA optionality, together barely ~9% of the book. So buying XLU for 'AI electricity' gets you mostly rate-sensitive bond-proxies and a thin merchant kicker. It is also cap-concentrated: a ~13% NextEra weight is the single biggest swing factor. The honest framing: a low-beta defensive utility fund that caught a thematic bid, not a pure-play on the load supercycle. For that, GRID (grid equipment) or the nuclear/uranium funds give cleaner exposure at far higher volatility.
- 2026-07July 2026 split the merchant sleeve: Constellation +5.8% ($248.37 to $262.75) against Vistra -6.6% ($158.63 to $148.19). With CEG positive and the fund still down 2.2%, the drag sat in the regulated core.
Thesis
XLU is the cheapest, most liquid way to own S&P 500 utilities — but its 2026 re-branding as a 'stealth AI power play' oversells what a cap-weighted regulated-utility basket actually delivers to the datacenter-demand thesis.
Passively tracks the Utilities Select Sector Index — all ~31 S&P 500 utilities (electric, multi-utility, IPPs, gas/water), cap-weighted so a few mega-caps dominate. At 0.08% it is a defensive, dividend-yielding sector sleeve, not a thematic fund.
Assessment
- 0.08% expense ratio — among the cheapest sector ETFs; near-free beta to US utilities
- ~$23.3B AUM, deep liquidity, tight spreads — an efficient core building block
- Genuine defensive/income profile: low beta, ~2.7% yield, ballast when growth sells off
- Owns the merchant-nuclear names (CEG, VST) actually signing datacenter PPAs, without single-stock risk
- 'AI power play' framing overstates it — ~80% is regulated utilities where load growth is ROE-capped and lags via rate cases
- Cap concentration: ~13% NextEra plus top-10 ~57% means a few names drive the result
- The 2026 re-rating leaned on CEG/VST (~9% of assets), not the regulated core the fund mostly is
- Bond-proxy rate sensitivity — ~4% Treasury yields are a persistent valuation headwind for regulated utilities
- Reported 1-year returns diverge widely by source (SSGA ~14% vs some aggregators ~23%), so headline numbers are date/methodology-sensitive
- 2026-07July 2026 gave the counter-example: with SOX -20.6% in a growth sell-off, XLU fell about 2.2% while the S&P 500 was -0.1%. The ballast underperformed the index by roughly 2pp in a risk-off month.
- 2026-07Jul 31, 2026: the spread has narrowed and inverted — one-year prints now run +6.42% to +9.14%. The ~23% high side is gone, and SSGA's +14.09% is the high outlier because it still reflects Jun 30.
Record
Per SSGA (NAV, as of Jun 30, 2026): YTD +7.62%, 1-year +14.09%, 3-year +14.86% annualized, 5-year +10.76%, 10-year +9.00%. The 2026 gain modestly beat the broad market's defensive expectation, driven disproportionately by the merchant-nuclear names (Constellation, Vistra) re-rating on datacenter PPAs and power-price strength rather than by the regulated majors that are the bulk of the fund. Treat cross-source 1-year figures cautiously — aggregators showed as high as ~23% vs SSGA's ~14%, a function of differing as-of dates around a volatile utility tape.
- 2026-07Jul 31, 2026: year-to-date is +5.28% and the one-year has roughly halved to +6.42% to +9.14% depending on source. The Jun 30 readings of +7.62% and +14.09% now sit a month behind the tape.
Risks & fit
- Rate risk: a back-up in long yields de-rates regulated utilities (bond-proxy dynamic) and can reverse the thematic bid fast
- Concentration: a NextEra-specific shock (~13% weight) or regulatory setback moves the whole fund
- Narrative unwind: if the datacenter-load story fades or utilities' capex outpaces authorized returns, the thematic premium compresses
- Merchant-power volatility: CEG/VST gains can reverse on power-price or PPA-cadence disappointment
The 'stealth AI-power' thesis for XLU is falsified if datacenter load growth fails to translate into utility earnings — i.e. regulated ROEs stay capped and rate-case lag means EPS growth tracks its historical low-single-digit rate while the stocks trade purely on the 10-year yield. If the fund's returns going forward correlate far more tightly with Treasury moves than with datacenter-capacity announcements, XLU was a bond-proxy that caught a narrative, not a power-demand play.
Suits an investor wanting cheap, liquid, defensive exposure to US utilities with an income tilt and incidental AI-power optionality — not someone seeking a concentrated bet on the datacenter electricity supercycle. Those wanting cleaner thematic torque look to grid-equipment (GRID) or nuclear/uranium funds, accepting far higher volatility. An equal-weight utility fund suits anyone wanting the sector without the NextEra concentration.
0.08% expense ratio — one of the lowest in any sector ETF and effectively tied with Vanguard's VPU (0.09%) and Fidelity's FUTY (0.08%); cost is not a differentiator among utility funds, exposure construction is.