1. Why This Sector Exists
Utilities deliver the three things modern life cannot pause: electricity, gas, and water. Customers pay every month because the alternative is darkness, cold, or thirst. Regulators grant local monopolies in exchange for price caps and reliability standards. Portfolios own utilities for the same reason landlords own apartment buildings — boring cash flows that show up whether GDP grows or shrinks.
2. What's Happening Right Now
What happened: The S&P 500 Utilities Select Sector has fallen roughly 5.7% over the last three months across its 31 constituents.
Schwab notes utilities has started to underperform after a strong run that drove valuations and earnings expectations higher. The blockbuster catalyst: NextEra agreed to buy Dominion Energy in May 2026, and AES is going private in a deal expected to close late 2026 or early 2027 at $17.28-area pricing.
Why it happened: AI-driven power demand pulled 2025 multiples to ~21x forward — too rich. As 10-year yields stayed sticky in the 4.3-4.5% zone, long-duration cash flows de-rated. M&A is the tell: strategics see asset values regulators won't pay for organically.
What it sets up: Next 4-8 weeks — Q2 prints (mid-July), summer cooling-load data, and FERC interconnection-queue updates determine whether the AI thesis re-rates the group or consolidates further.
3. How the Money Works
Revenue = rate base × allowed ROE. Translation: regulators let utilities earn ~9.5-10.5% on invested capital. Spend more capex on poles, wires, and turbines → rate base grows → earnings grow. The two costs that matter: fuel (passed through) and interest expense (not fully passed through — this is where rates kill you). Scale helps because fixed regulatory/legal overhead spreads across more customers. Great utilities sit in growing, constructive states. NextEra exemplifies it — top credit ratings, low payout ratio, financial flexibility to invest in renewables. Think toll road, not factory.
4. The 4 Macro Drivers
Driver 1: Long-End Interest Rates
Mechanism: Utilities are bond proxies — 4% dividend yields compete directly with Treasuries, and rate base growth is debt-financed. Higher 10-year yields hit both the multiple and the cost of capital.
Now: 10-year hovering 4.3-4.5%. That's why the group is down ~6% in three months despite great fundamentals.
2nd-order effect: Juniors watch the headline rate move; the money is in real rates. When inflation falls faster than nominal yields, real yields rise — and utilities get hit twice (multiple + rate-case lag on recovery).
Threshold: 10-year sustainably below 4.0% re-rates the group; above 4.75% breaks it.
Driver 2: AI / Hyperscaler Power Demand
Mechanism: Data centers are 24/7 baseload customers willing to sign 15-year PPAs at premium prices. With energy demand rising due to AI, Fidelity predicts utilities have potential for significant growth. This converts utilities from GDP-growth (1%) to load-growth (3-5%) businesses.
Now: Visible in NextEra's industry-leading $295-325 billion capex plan from 2025-2032 driving 8%+ annual earnings growth.
2nd-order effect: Everyone owns the obvious winners (CEG, VST, NEE). The hidden play is transmission — you can't ship Texas wind to a Virginia data center without wires. Watch the T&D-heavy names.
Threshold: A hyperscaler cap-ex cut announcement breaks the thesis fast.
Driver 3: Regulatory Constructiveness
Mechanism: State PUCs decide allowed ROE, capex recovery timing, and fuel pass-throughs. A 50bp ROE cut wipes 8-10% off EPS. Friendly states (FL, GA, NC) trade premium; hostile (CA pre-2020, IL) trade discount.
Now: Most commissions are accommodating because they need data center tax revenue and reliability investment.
2nd-order effect: When customer bills rise 15%+ due to AI capex, politics flips. Watch residential rate-shock backlash — that's the bear case nobody is modeling.
Threshold: Any major state opening a "data-center cost allocation" docket is your warning shot.
Driver 4: Commodity / Fuel Cost Backdrop
Mechanism: Natural gas sets marginal power prices in most US markets. Stable gas = stable rates = no political pressure. Spiking gas = bill shock + regulatory lag.
Now: Henry Hub well-behaved in the $3-4 range; LNG export growth a 2027 risk.
2nd-order effect: Juniors think low gas helps utilities. Wrong — it compresses the spread for merchant generators (CEG, VST) and reduces the value of nuclear/renewable PPAs.
Threshold: Gas sustainably above $5 changes the merchant-generator math and stresses gas-LDC affordability metrics.
5. Sector Map
| Sub-Industry | What It Does | Key Driver | Main Risk |
|---|---|---|---|
| Regulated Electric | Wires, poles, rate base | AI load growth | Rate-case lag |
| Independent Power Producers | Merchant generation | Power prices | Demand shock |
| Regulated Gas (LDC) | Distribute natural gas | Heating degree days | Electrification |
| Water Utilities | Treat, distribute water | Capex recovery | Drought, PFAS rules |
| Renewables/Diversified | Wind, solar developers | Tax credits, ITC | Policy reversal |
6. Company Case Studies
Case Study 1: NextEra Energy (NEE) — The franchise compounder buying scale
Business: Two engines — Florida Power & Light (regulated monopoly serving 12M Floridians) plus NextEra Energy Resources (largest US renewables developer). FPL throws off predictable rate-base growth; NEER monetizes IRA tax credits and signs long-dated hyperscaler PPAs. Unit economics: ~11% allowed ROE in Florida, mid-teens unlevered IRRs on renewables.
Moat: Scale in renewables development (land, interconnection queue position, supply-chain relationships). Florida's constructive regulator. One of the highest credit ratings among large rate-regulated electric utilities = cheapest capital in the sector. Widening.
Macro Linkage: Driver 2 (AI load) is the rocket fuel. The May 2026 Dominion acquisition doubles down — Dominion's Virginia footprint is the data center capital of the world. Driver 1 (rates) is the headwind compressing the multiple right now.
Watch: (1) Renewables backlog — proxy for 5-year EPS. (2) Dominion deal regulatory approvals across multiple states. 8%+ EPS growth through 2032 and a 10% dividend hike planned for 2026.
Risk: Bear case is a Republican Congress gutting IRA tax credits, slashing renewables IRRs. Early warning: any reconciliation bill touching 45Y/48E credits.
Valuation: Trading ~19x forward EPS vs. sector ~17x. Fair given growth, but priced for flawless deal execution.
Case Study 2: Constellation Energy (CEG) — The merchant nuclear AI play
Business: Owns the largest US nuclear fleet — ~22 GW of zero-carbon baseload. Sells power into competitive markets (PJM, NYISO, ERCOT) and increasingly via long-dated PPAs directly to hyperscalers. Marginal cost ~$25/MWh; selling at $60-80/MWh peaks. Operating leverage is enormous.
Moat: You cannot build new nuclear in <10 years. Existing licensed sites with grid interconnection are irreplaceable. Identified as one of the standout AI-trend utility beneficiaries. Moat widening as data centers demand 24/7 carbon-free power.
Macro Linkage: Driver 2 is everything here. CEG is the purest hyperscaler-PPA story. Driver 4 (gas prices) matters because gas sets the marginal price CEG receives in PJM — low gas = compressed margins on uncontracted MW.
Watch: (1) % of fleet contracted under long-term PPAs (target: >70% by 2027). (2) Average PPA price per MWh — current data-center deals printing well above merchant strip.
Risk: Bear case — hyperscaler capex pause + warm winter + cheap gas compresses spark spreads simultaneously. Early warning: PJM forward curve flattening.
Valuation: ~25x forward EPS — priciest in the sector. Justified only if AI-PPA pipeline keeps converting.
Case Study 3: PG&E (PCG) — Deep-value California rebuild
Business: Holding company whose main subsidiary Pacific Gas and Electric serves 5.3 million electricity customers and 4.6 million gas customers across 47 of California's 58 counties. Revenue is pure rate-base growth — undergrounding 10,000 miles of wire to mitigate wildfire risk.
Moat: Monopoly franchise across Northern California — irreplaceable. Moat was eroded by 2019 bankruptcy; now being rebuilt through AB1054 wildfire fund and improved operational track record.
Macro Linkage: Driver 3 (regulation) is the whole story. California's policy to eliminate economywide carbon emissions by 2045 requires substantial electrification investment, and PG&E plans $14 billion+ annual capex in 2026-30, leading to at least 9% annual earnings growth — one of the highest in US utilities.
Watch: (1) Wildfire season ignition count — every clean season de-risks the equity. (2) CPUC rate-case outcomes on cost-of-capital.
Risk: Catastrophic wildfire event reopening inverse-condemnation liability. Early warning: PSPS (public safety power shutoff) frequency spiking in dry windy conditions.
Valuation: Trading 19% below Morningstar's $20.50 fair value estimate. ~13.5x forward — cheap for 9% grower. Cheap for a reason; that's the trade.
7. How to Value These Companies
Use P/E on forward EPS for regulated names — earnings track rate base, which is highly predictable. Use EV/EBITDA for merchant generators (CEG, VST) because tax-credit monetization distorts EPS. Dividend yield vs. 10-year Treasury spread (target: 100-200bp) tells you cyclical positioning. Typical range: 16-20x forward P/E; merchants 8-12x EV/EBITDA. Junior mistake: anchoring on dividend yield without checking payout ratio — a 5% yield with 95% payout is a dividend cut waiting to happen.
8. KPIs That Actually Matter
| KPI | What It Signals | Why It Beats EPS | Benchmark |
|---|---|---|---|
| Rate base growth % | Future earnings power | EPS lags by 12-18 months | 6-9% target |
| Authorized ROE | Regulator constructiveness | Sets entire cap structure return | 9.5-10.5% |
| FFO/Debt | Credit quality | Drives cost of capital directly | >14% (BBB+) |
| Equity issuance need | Dilution risk | EPS hides share count creep | <$2B/yr healthy |
| % load from data centers | AI exposure | Tomorrow's growth, not today's | >5% = premium |
| Regulatory lag (months) | Cash flow timing | Earnings recognition vs cash | <12 months ideal |
9. Risk Map
Risk 1: Wildfire Catastrophic Liability
California's inverse condemnation makes utilities strictly liable for equipment-caused fires regardless of negligence. Transmission: one bad season → multi-billion liability → bankruptcy → equity wipe → rate base reset at lower allowed ROE. Precedent: PG&E's 2019 Chapter 11 filing destroyed ~$25B equity value after Camp Fire. Early warning: Drought monitor showing extreme conditions in service territory + low reservoir levels + sustained Diablo wind events. Watch CAL FIRE incident reports through October.
Risk 2: Interest Rate Whipsaw / Refinancing Wall
Utilities carry 55-60% debt-to-cap. A 100bp rise in long rates lifts annual interest expense materially and compresses the dividend-yield-vs-Treasury spread that defines fair multiple. Transmission: higher coupon on refis → EPS drag → multiple compression simultaneously. Precedent: 2022-23 sector underperformance — XLU fell ~20% while S&P rose. Early warning: TIPS breakevens diverging from nominals; Fed dot plot pushing terminal rate higher. Watch August Jackson Hole positioning closely.
Risk 3: Hostile Rate Case / Political Backlash
When residential bills jump because utilities are spending on data-center infrastructure, voters revolt. Regulators respond by allocating costs to hyperscalers (good) or denying recovery (bad). Transmission: disallowance → write-down → ROE cut → multiple de-rate. Precedent: Illinois 2023 ComEd rate case slashed Exelon assumed ROE. Early warning: Local TV coverage of bill shock, attorney general intervention petitions, gubernatorial campaigns featuring utility-bashing rhetoric. Watch Virginia and Ohio dockets specifically.
Risk 4: AI Demand Disappointment
The bull case assumes hyperscaler load grows 15%+ annually for a decade. If model efficiency gains (smaller, cheaper inference) reduce GPU demand, or if hyperscalers overbuild and pause, the merchant and renewables-developer thesis cracks. Transmission: PPA repricing → backlog impairment → growth multiple collapses to value multiple. Precedent: Telecom 2001 — dark fiber overbuild. Early warning: Hyperscaler capex guidance cuts, cancelled colocation contracts, falling PJM capacity auction prices.
10. Cycle Playbook
| Phase | Sector Behaviour | Why | What to Own |
|---|---|---|---|
| Early Expansion | Underperforms | Cyclicals lead | Underweight |
| Mid Cycle | In-line | Rates stabilize | Quality regulated |
| Late Cycle | Outperforms | Defensive bid begins | High-quality compounders |
| Recession | Strong outperform | Bond proxy, earnings stable | Regulated electric, water |
| Recovery | Lags | Risk-on rotation out | Trim, rotate to merchants |
Now: Late-cycle with sticky inflation and AI-driven capex super-cycle overlaid. Utilities should be a hold-to-overweight here, but the AI-growth premium has stretched valuations — focus on regulated names with hidden data-center upside, avoid the obvious merchant winners at 25x.
11. Structural Themes
Theme 1: The Electrification Super-Cycle
EVs, heat pumps, AI data centers, and industrial reshoring are pushing US electricity demand growth from ~0.5%/yr to ~3%/yr — first sustained acceleration since the 1970s. Why now: IRA subsidies, hyperscaler capex, onshoring policy converged simultaneously. Winners: regulated electrics with growing service territories (NEE, SO, DUK), grid-equipment names. Losers: gas LDCs facing electrification of heating. Position before consensus: Transmission specialists and grid-equipment suppliers — bottleneck is wires, not generation.
Theme 2: Consolidation Wave
NextEra-Dominion (May 2026), Essential Utilities-American Water (announced October 2025, closing Q1 2027, with Essential holders owning ~31% of the combined entity), and AES going private late 2026/early 2027 — the sector is consolidating fast. Why: scale matters for AI-era capex (>$300B over a decade), and rising rates have made mid-caps cheap takeout candidates. Winners: shareholders of subscale BBB-rated utilities trading at discounts. Position before consensus: Screen for sub-$30B-EV regulated names with constructive regulatory environments — they're the next targets.
12. Portfolio Reference
| Factor | Value |
|---|---|
| S&P 500 weight | ~2.4% |
| Typical dividend yield | 3.0-3.5% |
| Beta vs S&P 500 | 0.55-0.70 |
| Overweight when | Late cycle, falling rates |
| Underweight when | Early expansion, rising rates |
| ETF | Focus | Expense Ratio |
|---|---|---|
| XLU | S&P 500 Utilities | 0.09% |
| VPU | Broad US Utilities | 0.09% |
| RSPU | Equal-weight Utilities | 0.40% |
13. Three Questions You Should Be Able to Answer
Q1: Why do utility earnings grow when management spends more money, when in every other sector that destroys returns?
A: Because the regulatory compact allows utilities to earn an authorized return on invested capital (rate base), not just from it. Spend $1B on transmission → rate base grows $1B → at 10% allowed ROE that's $100M of pre-tax earnings. The catch: regulators must approve the spend as "used and useful," and rate-case lag means cash recovery trails the spend by 12-18 months. So capex = growth, but only inside a constructive regulatory framework. Outside it, capex = stranded asset.
Q2: Walk me through how a 50bp move in the 10-year yield actually hits a utility's intrinsic value, in three steps.
A: Step one: discount rate on long-duration dividend stream rises → DCF value falls ~8-10%. Step two: refinancing wall — utilities roll $20-40B of debt annually; 50bp adds materially to interest expense, compressing EPS 2-4%. Step three (the one juniors miss): regulators reset allowed ROE based on prevailing rates, but with an 18-month lag, so the cash flow hit precedes the recovery. Net effect: ~12-15% fair-value compression on a sustained 50bp move, not the 5% the dividend-yield math suggests.
Q3: With AI demand booming but rates sticky and valuations stretched, are you long or short utilities here?
A: Long, but selectively. Bull: AI-driven energy demand gives utilities significant growth potential, M&A is supporting valuations, late-cycle macro favors defensives. Bear: the group has started to underperform after a strong run that drove valuations too high, and 4.5% 10-years cap multiples. Trade: own regulated compounders with hidden data-center exposure (NEE, DUK, SO) and one deep-value rebuild (PCG); avoid merchant generators above 22x. Flips bullish: 10-year breaks 4.0%. Flips bearish: hyperscaler capex cut announced.
Research via live web search | Saturday, June 27, 2026 | GICS Rotation Series