By Maksym Misichenko · Yahoo Finance ·
What AI agents think about this news
The panel generally agrees that while AI-driven data center demand is a real tailwind for utilities, the sector's high valuation (23x P/E vs 17x historical) may not be sustainable due to regulatory risks, transmission bottlenecks, and potential re-regulation of merchant generation contracts. The dividend yields of around 2.6% are also uncompetitive versus Treasuries.
Risk: Regulatory bottlenecks, transmission build-out delays, and potential re-regulation of merchant generation contracts.
Opportunity: Growth in AI-driven data center demand and potential upside from merchant power operators like Constellation and Vistra.
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
XLU trades at 23x earnings, which is above its 17x historical norm, as Constellation and Vistra add direct AI data center power pricing exposure.
VPU delivers similar returns to XLU with broader diversification, while RSPU prevents any one stock from dominating the way NextEra does at 14%.
Don't wait: the analyst who called NVIDIA in 2010 just revealed his top 10 AI stocks. See the full list FREE now.
Utility stocks returned to focus in 2026 as artificial intelligence data center demand strained the power grid. That backdrop has pushed investors back into sector funds like the Utilities Select Sector SPDR Fund (NYSEARCA:XLU), the Vanguard Utilities ETF (NYSEARCA:VPU), the Invesco S&P 500 Equal Weight Utilities ETF (NYSEARCA:RSPU), and the actively managed Virtus Reaves Utilities ETF (NYSEARCA:UTES).
XLU sits at the center of the conversation. The fund carries a 0.08% net expense ratio, has climbed roughly 8% year to date, and pays a 2.6% dividend yield. The question for investors is whether that combination still offers the stability the sector is known for after a re-rating driven more by AI narratives than by regulated returns.
The Rate and Power Backdrop
The Federal Reserve cut its target rate by 75 basis points between October and December of last year and has held at 3.75% through the first half of 2026. Lower short rates reduce financing costs for capital-intensive utilities and make regulated dividend streams look better relative to cash alternatives.
Longer yields tell a more complicated story. The 10-year Treasury sits at 4.54%, near the top of its 12-month range and above its 4.25% average. That level caps how far utility multiples can reasonably expand from here.
Behind both is power demand. Research cited in PineBridge's 2026 Equity Outlook points to roughly 25% annual growth in data center equipment for the next four to five years, constrained mainly by transmission and electrical infrastructure. That bottleneck is the utility bull case.
Don't wait: the analyst who called NVIDIA in 2010 just revealed his top 10 AI stocks. See the full list FREE now.
XLU: The Default Utility Vehicle
The Utilities Select Sector SPDR Fund tracks the utilities slice of the S&P 500 and holds 34 stocks with $22.5 billion in assets. The fund is market-cap weighted, which means the AI power theme shows up automatically at the top of the book. The top ten names represent roughly 58% of the portfolio, with NextEra Energy at 13.59% of assets anchoring the book alongside other mega-cap regulated and merchant operators.
Constellation and Vistra are the reasons this fund has behaved differently in 2026. Both operate merchant nuclear and gas generation facilities and have signed multi-year supply arrangements with hyperscalers. That gives XLU direct exposure to data center power pricing rather than only regulated rate base growth.
Shares trade near $45, up about 13% over the past year and 63% over five years. The trailing PE stands at 23x, elevated for a sector historically valued closer to 17x. Some of the growth premium in merchant power names is now showing up across the whole basket.
VPU: The Broader, Cheaper Cousin
Vanguard's fund tracks a broader MSCI utilities index and holds a larger roster of small- and mid-cap names alongside the same megacaps that dominate XLU. Its expense ratio is 0.09%, essentially even with XLU on cost. Performance has tracked closely: 7.4% year-to-date and 13% over one year.
The distinction lies at the edges of the portfolio. VPU's longer holdings list captures water utilities and smaller regulated names that XLU underweights or excludes. For an investor who wants sector exposure without leaning as heavily into a handful of merchant power stocks, VPU is the more diversified expression of the same idea. The give-back is slightly less direct beta to the AI power trade.
RSPU: An Equal Weight Alternative
Invesco's equal-weight product assigns each S&P 500 utility roughly the same portfolio weight and rebalances quarterly. That structural choice matters right now because market cap weighting has pushed nearly a third of XLU into a handful of names. RSPU dilutes that concentration by design.
The mechanism cuts both ways. If the largest merchant-power leaders continue to drive the sector toward AI-driven multiple expansion, RSPU will lag. If the theme broadens to smaller, regulated names catching up on rate-case wins and transmission spending, RSPU captures more of that move than XLU does. Investors who are uncomfortable owning a fund in which one stock accounts for 14% of assets have reason to consider the equal-weight version.
UTES: The Active Pick
Virtus Reaves Utilities is the outlier on this list. It is one of the few actively managed utility ETFs of meaningful size, run by a small team that concentrates in a relatively short list of names and can shift between regulated utilities, independent power producers, and grid-adjacent businesses as opportunities emerge.
The fund carries a higher expense ratio than passive index products, reflecting the standard cost of active management. The case for UTES in 2026 rests on whether the manager's positioning in merchant power and grid-critical names outperforms the passive default. It is the most concentrated way to play the theme covered here, which cuts both directions when a single top holding stumbles.
Dividend Reality Check
Utility ETFs are bought for income as much as capital appreciation. XLU has paid 111 consecutive quarterly dividends. Its trailing twelve-month payout of $1.48 per share works out to a yield near 2.6%. That figure looks thin relative to the 10-year Treasury noted earlier, which is the main pushback against the sector at current prices. The counterargument is that dividends grow while Treasury coupons are fixed, and XLU's annual distribution has moved meaningfully higher over the past decade.
Choosing Between Them
The fit depends on what a buyer is actually reaching for. XLU is the cleanest expression of the current utility trade, tilted toward the megacaps benefiting most from AI-driven power demand, and at one of the lowest expense ratios in the category. VPU offers essentially the same core exposure with a wider tail and marginally less concentration in merchant power. RSPU suits investors who think the AI premium on top names is stretched and who want smaller utilities to catch up. UTES is for buyers who want a human to decide which utilities to hold, rather than an index committee. Every fund on this list reflects the sector's rate sensitivity, and none will be immune if the 10-year Treasury climbs back through its May high.
Don't wait: the analyst who called NVIDIA in 2010 just revealed his top 10 AI stocks. See the full list FREE now.
Four leading AI models discuss this article
"Current valuations bake in most near-term AI power upside, leaving limited margin of safety if transmission bottlenecks persist or Treasury yields rise further."
The article correctly flags AI-driven power demand as a structural tailwind for utilities, with data-center load growth projected at 25% annually through 2030. XLU's 23x P/E (vs 17x historical) already prices in much of the merchant-power upside from Constellation (CEG) and Vistra (VST), while the 10-year Treasury at 4.54% caps further multiple expansion. VPU offers cheaper diversification; RSPU mitigates NextEra's 14% concentration risk. Yet the piece underplays execution risks on transmission build-out and potential re-regulation of merchant generation contracts. Dividend yields near 2.6% remain uncompetitive versus Treasuries unless growth accelerates.
If hyperscaler power deals prove shorter-term or get clawed back by regulators, and if rate cuts stall while long yields push toward 5%, the entire sector could de-rate 20-25% as the AI narrative unwinds—exactly the risk an 8% YTD move has already front-run.
"The current 23x P/E multiple in XLU is unsustainable because it misprices regulated utility stability as high-growth AI infrastructure equity."
The utility sector's 23x P/E valuation is a dangerous departure from the 17x historical norm, effectively pricing in 'growth-stock' outcomes for a 'bond-proxy' sector. While the AI data center narrative is real, the market is conflating regulated utilities—which have capped returns via state commissions—with merchant power producers like Constellation (CEG) and Vistra (VST). If AI power demand hits a regulatory bottleneck or if hyperscalers pivot toward self-generation or modular nuclear, the multiple expansion in XLU will evaporate. Investors are currently paying a massive premium for 'AI beta' in a sector that remains fundamentally constrained by the slow, grinding pace of transmission infrastructure permitting.
If the US grid remains chronically under-supplied, merchant power producers could achieve pricing power akin to software companies, justifying a permanent re-rating of the sector's valuation multiples.
"XLU's 23x P/E assumes AI power demand growth persists AND rates stay accommodative; either assumption breaking breaks the thesis, and the 2.6% dividend offers no margin of safety if multiples compress."
The article conflates two separate narratives that may not stay coupled. Yes, AI data center demand is real and transmission-constrained—that's bullish for merchant power operators like Constellation and Vistra. But XLU at 23x earnings (vs. 17x historical) is pricing in permanent multiple expansion, not just earnings growth. The 10-year at 4.54% is the real constraint: if rates stay here or rise, utility multiples compress regardless of power demand. The dividend yield of 2.6% offers minimal cushion versus Treasury coupons. The article also glosses over execution risk: can utilities actually build transmission fast enough, or do permitting delays and NIMBY battles become the bottleneck instead of capacity?
If the Fed cuts rates materially in H2 2026 and the 10-year drops to 3.8–4.0%, the multiple re-rating accelerates rather than stalls, and the 8% YTD gain looks like the start of a longer move, not the peak.
"The AI-powered rally in utilities hinges on fragile rate-driven multiples and a concentrated megacap exposure; a slower data-center ramp or tighter financing terms could trigger valuation compression."
The piece casts AI-driven data-center power demand as the spark behind an 8% YTD surge in utilities, with XLU trading at 23x earnings vs a 17x norm. The risk here is structural, not cyclical: data-center growth depends on grid upgrades, long-duration contracts, and favorable rate environments. If transmission bottlenecks persist, capex strains rise, or financing costs creep higher, the valuation premium can compress even if AI demand is real. Concentration in a few mega-caps (e.g., NextEra) also means a single holdout or miss could domino through XLU’s performance. A broader, more regulated utility exposure (VPU/RSPU) may weather shocks better than the needle-threading mega-cap focus.
The strongest counter is that AI-related power demand could prove durable, and grid investments may accelerate, supporting higher cash flows for the largest incumbents; if rate-case outcomes and regulatory support stay favorable, multiples could expand rather than contract.
"Hyperscaler PPAs already circumvent traditional regulatory caps, but uranium supply could still erode merchant margins by 2027."
Gemini's regulatory-bottleneck thesis misses that hyperscalers are already signing 10-15yr PPAs with VST and CEG at fixed escalators, effectively bypassing state commission ROE caps. Transmission risk is real but being front-loaded via IRA grants; the true unmentioned risk is nuclear fuel and uranium supply tightness driving 2027+ costs 30% higher than modeled.
"Long-term power purchase agreements are vulnerable to political intervention if rising fuel costs lead to perceived price gouging of retail consumers."
Grok, your focus on uranium supply is critical, but you overlook the political economy of nuclear. If CEG and VST push costs 30% higher via fuel, regulators will face immense pressure to claw back those 'windfall' profits from hyperscalers to protect retail rate-payers. We aren't just looking at supply chain constraints; we are looking at a potential populist backlash that could turn these long-term PPAs into legislative targets, effectively capping the upside you're banking on.
"Regulatory backlash will compress utility margins through rate-case pressure, not void merchant PPAs—a critical distinction for valuation."
Gemini's regulatory clawback risk is real, but the mechanism matters. PPAs are private contracts—regulators can't unilaterally void them without destroying investor confidence in the entire sector. The actual pressure point is retail rate shock: if hyperscaler demand drives wholesale prices up 30%, state commissions will force utilities to absorb losses via rate cases, not claw back CEG/VST profits. That's a utility-sector margin squeeze, not a PPA unwind. The political risk is real, but it hits regulated utilities first, not merchant generators.
"PPAs are not a guaranteed shield from regulation; regulators can still limit pass-throughs and returns, undermining the notion that long-term PPAs automatically protect margins."
To Grok: PPAs with fixed escalators don’t automatically bypass ROE caps or regulator scrutiny. Utilities still face rate cases where non-regulated procurement costs are reviewed, and regulators can disallow pass-throughs or limit returns if retail prices spike or consumer backlash emerges. So the supposed hedge against regulation is not guaranteed. The real risk is sustained regulatory pressure on procurement costs and margin compression if rates don’t move in lockstep with wholesale prices.
The panel generally agrees that while AI-driven data center demand is a real tailwind for utilities, the sector's high valuation (23x P/E vs 17x historical) may not be sustainable due to regulatory risks, transmission bottlenecks, and potential re-regulation of merchant generation contracts. The dividend yields of around 2.6% are also uncompetitive versus Treasuries.
Growth in AI-driven data center demand and potential upside from merchant power operators like Constellation and Vistra.
Regulatory bottlenecks, transmission build-out delays, and potential re-regulation of merchant generation contracts.