US Power Markets: Re-rating
The S&P 500 Utilities sector (S5UTIL) has delivered ~25% total return in 2025 and ~32% cumulative over 18 months — its strongest run since the early 2000s. Forward P/E has expanded from the 16–17× range that defined the 2015–22 era to roughly 20–21× today. The merchant nuclear names have re-rated most violently — Constellation Energy now trades at ~26× forward earnings versus a five-year average around 14×. This is not a bubble. It is the market repricing utility earnings growth from 4–5% per year (the bond-proxy era) to 8–10% per year (the AI-infrastructure era).
Our base case from Part II (Hybrid Resilience, ~45% probability) and the AI Abundance scenario (~40%) collectively make up 85% of probability-weighted outcomes — both supportive of the trade. We construct portfolio exposure across five tracks: nuclear-anchored merchants, regulated rate-base growth, equipment manufacturers, climate resilience compounders, and SMR optionality. The thesis is not subtle and it is not over.
The single most important framing: utilities are no longer a defensive bond proxy. They are a growth trade. The portfolio construction, conviction-weighted allocation, and risk framework follow.
| 01 | S5UTIL — the sector has already started running |
| 02 | The investment thesis shift — bond proxy to growth engine |
| 03 | Five positioning tracks — anchor stocks & the rationale |
| 04 | Scenario sensitivity — how each track performs across A/B/C |
| 05 | Portfolio blueprint — conviction-weighted allocation |
| 06 | The five risks that could compress the trade |
| 07 | Catalysts watch & what to monitor |
| G | Glossary additions (cumulative) |
01 · S5UTIL — the sector has already started running
The S&P 500 Utilities sector index (S5UTIL) is the cleanest read on how the market is pricing the structural shift we’ve documented. The trajectory has been violent. After fifteen years of compounding at roughly the rate of inflation — appropriate for a defensive bond-proxy sector — the index re-rated sharply from late 2024 onwards.
Three observations matter. First, the sector tracked roughly at the broad market through 2019–24 — appropriate for its historic role as a bond proxy with low beta. Second, beginning in late 2024 the sector decoupled materially, outperforming the S&P 500 over an extended period. This is not a momentary flash. It is the market re-rating utility earnings growth in response to the data centre demand shock, hyperscaler PPAs, and capacity market reset documented in Part II. Third, the dispersion within the sector has widened sharply: merchant generators (Constellation, Vistra, Talen) have outperformed regulated names by 2–3× — a pattern that informs portfolio construction.
Forward P/E — the multiple expansion story
The S5UTIL forward P/E has expanded from roughly 16–17× during the 2015–22 era to 20–21× today. That move alone represents roughly 25% of the index appreciation; the balance comes from earnings upgrades. Selected name-level forward multiples illustrate the dispersion:
| Name | Type | Fwd P/E now | 5-yr avg | Premium / discount |
|---|---|---|---|---|
| Constellation Energy (CEG) | Merchant nuclear | ~26× | ~14× | +86% |
| Vistra (VST) | Merchant gas + nuclear | ~22× | ~10× | +120% |
| Talen Energy (TLN) | Merchant nuclear | ~24× | n/a (recent IPO) | premium re-listed |
| NextEra Energy (NEE) | Hybrid (regulated + merchant) | ~22× | ~22× | in line |
| Dominion Energy (D) | Regulated, data centre | ~19× | ~17× | +12% |
| Southern Company (SO) | Regulated, Atlanta load | ~20× | ~18× | +11% |
| GE Vernova (GEV) | Gas turbine OEM | ~38× | n/a (2024 spin) | growth multiple |
| PG&E Corp (PCG) | Regulated, wildfire risk | ~14× | ~13× | in line |
| S5UTIL (sector avg) | Sector | ~20–21× | ~16–17× | +25% |
Note: Forward P/E figures are approximate as of May 2026 and based on analyst consensus EPS estimates. Multiples shown for illustrative purposes; investors should verify current valuations before any position.
02 · The investment thesis shift — bond proxy to growth engine
For three decades, US utilities were owned for three reasons: 3–5% dividend yield, defensive characteristics in market drawdowns, and inflation-linked rate-base growth of 4–6% per year. The total return narrative was bond-like: low single-digit EPS growth plus the dividend. This framing is now obsolete.
What changed — five structural drivers
- Demand growth is back, structurally. The flat-demand consensus that defined 2005–22 has broken. EIA forecasts 3.1% YoY growth in 2027. The vector mix — data centres, electrification, manufacturing reshoring — is durable across multiple administrations. We covered this in detail in Part II.
- Rate-base growth has accelerated. Regulated utility capex is running 60–80% above the pre-2023 baseline. Climate adaptation (covered conductors, undergrounding), data centre transmission build-out, and aging grid replacement combine to produce 8–10% rate-base growth at the most exposed names — versus 4–5% historic norm.
- Merchant generator economics have repriced. PJM capacity prices at $329/MW-day. Hyperscaler PPAs at premium prices for 15–20 years. Coal-plant retirements deferred. Nuclear restart economics transformed. Constellation guidance: 20%+ annual EPS growth 2026–29.
- Earnings revisions have turned positive. Q3 2025 Utilities sector posted +14% YoY EPS growth — the fastest among defensive sectors and ahead of materials, energy, and consumer staples. This is the empirical evidence of the structural shift.
- Inflation Reduction Act tailwind, OBBBA selective preservation. Nuclear PTC (§45U), battery storage ITC (§48E), geothermal credits, and CCUS §45Q all preserved or enhanced under OBBBA. Wind and solar compressed but operational projects retained credits. As we mapped in Part I, the OBBBA framework is brutal for new wind/solar but supportive of nuclear and storage operators.
The new mental model — three earnings drivers
For regulated utility names, the earnings algorithm is now:
For merchant generators, the algorithm runs differently:
Both algorithms now produce 8–20% annual EPS growth at the most exposed names. That is fundamentally not a bond proxy any longer. The required return investors should expect from the sector — and the multiples the market is willing to pay — have re-rated accordingly.
The most consequential analytical error in US power equities right now is using historic valuation frameworks to assess current multiples. Constellation at 26× looks expensive against the 14× five-year average; it looks reasonable against 20%+ EPS growth guidance. The same logic applies to the regulated names. Anchoring to 2015–22 multiples is the single largest mistake on this trade.
03 · Five positioning tracks — anchor stocks & the rationale
We construct exposure across five distinct tracks. Each captures a different aspect of the structural thesis. Together they provide diversified exposure to the convergent outcomes we identified in Part II’s three-scenario framework — firm zero-carbon capacity, multi-horizon storage, transmission expansion, climate resilience, and demand-side flexibility.
Track 1 — Nuclear-anchored merchant IPPs (the pure-play growth trade)
Names: Constellation Energy (CEG), Vistra (VST), Talen Energy (TLN), Public Service Enterprise Group (PEG).
These are the cleanest hyperscaler beneficiaries. Long-dated PPAs convert merchant exposure into infrastructure-grade contracted cash flows. The §45U Zero-Emission Nuclear PTC, preserved under OBBBA, provides a price floor; data centre PPAs provide an explicit price ceiling well above legacy wholesale economics.
Constellation owns the largest US nuclear fleet (~23 GW), signed the Microsoft–Three Mile Island restart, has 1.1 GW Meta PPA at Clinton, and is guiding 20%+ annual EPS growth 2026–29 after the Calpine acquisition. Vistra closed a $4.7 billion Cogentrix gas deal and has signed 3.8 GW of 20-year hyperscaler PPAs across Comanche Peak, Perry, and Beaver Valley. Q1 2026 revenue at Constellation was up 64% YoY. Talen Energy is the Susquehanna nuclear story — Amazon’s 1.92 GW behind-the-meter deal. PSEG operates regulated PSEG New Jersey alongside merchant nuclear at PSEG Power; the hybrid structure provides a more conservative entry into the same thesis.
The risk: valuations already reflect a meaningful portion of the AI thesis. The opportunity: if 20% earnings growth proves durable, multiples can hold or re-rate further. Conviction: very high. Position size: 25–30% of sector sleeve.
Track 2 — Regulated utilities with data centre footprint (the defensive growth trade)
Names: Dominion Energy (D), Southern Company (SO), NextEra Energy (NEE), Duke Energy (DUK), Xcel Energy (XEL), Entergy (ETR).
These are the rate-base growth stories. Dominion serves Northern Virginia, where roughly 70% of global internet traffic flows; it is in contract talks for 40–47 GW of new data centre capacity. Southern has 50 GW of potential large-load growth in its Georgia pipeline, with ~40 GW concentrated around Atlanta. NextEra has a 33 GW renewables backlog plus a 60 GW data centre hub in development with 10 GW of approved new gas. Duke, Xcel, and Entergy cover similar territory in the Carolinas, Upper Midwest, and Gulf Coast respectively.
These names trade at lower multiples than the merchant IPPs (typically 18–22× forward EPS), pay higher dividends (3–4% yields), and offer regulated rate-base growth in the 6–10% range. The earnings outcome is more predictable; the catalyst path is slower. This is the sweet spot for a balanced equity portfolio. Conviction: high. Position size: 30–35% of sector sleeve.
Track 3 — Equipment manufacturers (the picks-and-shovels)
Names: GE Vernova (GEV), Eaton (ETN), Quanta Services (PWR), MasTec (MTZ), Vertiv (VRT), Generac (GNRC).
The gas turbine, transformer, and transmission build-out has a finite list of beneficiaries. GE Vernova reported $2.4 billion in Q1 2026 data centre-related orders — exceeding all of 2025. Management has guided 16–18% organic revenue growth for the power division in 2026, with electrification revenues up 20%. Free cash flow is targeted at $4.5–5 billion. The risk: cyclical execution and gas turbine supply chain bottlenecks could swing margin guidance.
Eaton, Quanta, and MasTec are the transmission and substation buildout proxies — Quanta’s backlog stretches well into 2028, and the firm has built deep capacity in HVDC engineering relevant to all three Part II scenarios. Vertiv (data centre cooling and power management) is the most direct data centre play but trades at the most expensive multiple in the basket (35× forward). Generac provides residential and small-commercial backup power — secondary play with elevated cyclical sensitivity.
Conviction: high. Position size: 15–20% of sector sleeve. Important caveat: this track has more cyclical risk than Track 1 or 2 — order books can compress quickly if construction slows.
Track 4 — Climate resilience and grid hardening (the rate-base compounders)
Names: PG&E Corporation (PCG), Edison International (EIX), CenterPoint Energy (CNP), Avangrid (AGR).
PG&E remains the most contentious name in US utilities — the wildfire liability tail is genuinely difficult to size, and the political risk in California is non-trivial. But the rate-base growth trajectory is exceptional: ~10% annual compounding on the back of $12+ billion of undergrounding and covered-conductor deployment, all rate-base eligible. SCE’s parent EIX is similar with somewhat lower volatility. CenterPoint is the Houston-focused version of the same thesis — post-Hurricane Beryl, the company committed over $5 billion to distribution hardening across the Houston territory through 2030. Avangrid (Iberdrola’s US subsidiary, recently re-taken private) offers similar exposure across Northeast.
These names trade at meaningful discounts to peers (PCG at 14× forward vs sector 20×) and offer the highest risk-adjusted rate-base growth in the sector. The market is pricing the wildfire liability tail; we believe the political consensus around resilience and the ongoing capex programmes have materially reduced that tail risk versus 2019. Conviction: medium-to-high (asymmetric). Position size: 10–15% of sector sleeve.
Track 5 — SMRs and advanced nuclear (the option value)
Names: NuScale Power (SMR), BWX Technologies (BWXT), Cameco (CCJ), Centrus Energy (LEU), Oklo (OKLO).
Position size matters here. SMRs are a 2030+ deployment story with significant execution risk. NuScale’s 6 GW TVA partnership and ENTRA1 relationship represent the largest committed pipeline but no module has yet been built. BWX Technologies is the most defensive way to play the theme — it has the existing fabrication capacity for SMR pressure vessels and a $1.5 billion NNSA defence enrichment contract. Cameco owns 49% of Westinghouse, making it the cleanest broad nuclear exposure. Centrus Energy is the most direct play on HALEU (High-Assay Low-Enriched Uranium) supply for advanced reactors. Oklo is the highest-beta SMR story.
Conviction: medium (binary outcome). Position size: 5% of sector sleeve. Treat the segment as venture-style optionality within an income-oriented utility book. If SMRs deploy at scale by 2030, this segment 3–5×; if they slip to 2033+, the names compress materially.
04 · Scenario sensitivity — how each track performs across A/B/C
The five tracks have different sensitivities to the three scenarios we constructed in Part II. This is what makes the portfolio construction robust — the convergent enabling systems means most tracks perform across most scenarios. The table below scores each track across the three scenarios with a directional rating.
| Track | A · AI Abundance (40%) | B · Climate-Led (15%) | C · Hybrid Resilience (45%) | Prob-weighted |
|---|---|---|---|---|
| 1 · Nuclear merchants | Strong | Strong | Strong | Very strong |
| 2 · Regulated rate-base | Strong | Strong | Strong | Strong |
| 3 · Equipment OEMs | Strong | Mixed* | Strong | Strong |
| 4 · Climate resilience | Strong | Strong | Very strong | Strong |
| 5 · SMR optionality | Medium | Strong | Medium | Medium |
* Track 3 equipment OEMs see mixed outcome in Climate-Led Scenario B because gas turbine demand compresses while transmission/HVDC build accelerates — net depends on company mix. GE Vernova is more balanced; pure-play gas turbine names would suffer.
The portfolio implication: Tracks 1, 2, and 4 perform across all three scenarios. Track 3 is robust on the dominant A and C scenarios. Track 5 has optionality biased toward Scenario B but with sufficient base-case demand in A and C to sustain the names. The scenario-agnostic nature of Tracks 1, 2, and 4 is what justifies the heavier weights in those buckets.
05 · Portfolio blueprint — conviction-weighted allocation
The portfolio blueprint below assumes a dedicated utility / energy infrastructure sleeve within a broader equity book. For investors managing the sector as a single line item within a diversified equity portfolio, sleeve weight should be 8–12% of total equity — meaningfully above the S&P 500 utilities weighting (~2.5%) to reflect the structural overweight thesis.
| Track | Weight | Profile | Anchor names |
|---|---|---|---|
| 1 · Nuclear merchants | 25–30% | Growth, AI tail, premium multiples | CEG, VST, TLN, PEG |
| 2 · Regulated rate-base | 30–35% | Defensive growth, dividends, lower volatility | NEE, D, SO, DUK, XEL, ETR |
| 3 · Equipment OEMs | 15–20% | Cyclical growth, picks-and-shovels | GEV, ETN, PWR, MTZ, VRT |
| 4 · Climate resilience | 10–15% | Asymmetric value, rate-base compounders | PCG, EIX, CNP, AGR |
| 5 · SMR optionality | 5% | Venture-style optionality, binary outcome | BWXT, CCJ, SMR, OKLO, LEU |
| Cash / dry powder | 5% | For drawdowns and adding to convictions | Treasury bills, short duration |
Within-track guidance
Within each track, position-level sizing should follow the within-track diversification rule — no single name above 40% of track weight. For Track 1, that implies CEG and VST at 30–40% each, TLN at 15–20%, PEG as the conservative hedge at 10–15%. For Track 2, the natural anchor weights are NEE (largest, most diversified) at 20–25%, with the others equally split. For Track 3, GE Vernova is the dominant exposure (35–40%) given its leveraged play on the gas turbine bottleneck.
The portfolio is constructed for the central tendency of our Part II scenario framework — predominantly Hybrid Resilience (45%) and AI Abundance (40%) outcomes. It is robust to Climate-Led Decarbonisation (15%) through Track 5’s optionality and Track 2’s regulated diversification. The most acute risk is a fast capacity-market reform in PJM that compresses Track 1 economics; we discuss this and four other risks in Section 06.
06 · Five risks that could compress the trade
No thesis survives complete certainty. The five risks below could materially compress the trade. We assign probabilities and direction of impact to each.
| Risk | Probability | Most affected tracks |
|---|---|---|
| Capacity market reform | High | Track 1 (merchant nuclear most exposed) |
| Hyperscaler capex slowdown | Medium | Tracks 1, 3 directly; 2 indirectly |
| Cost of capital shock | Medium | Tracks 2, 4 most sensitive; 1 partially insulated |
| Regulatory backlash on rates | Medium | All tracks; Track 4 most resilient |
| Gas turbine supply chain | Medium | Track 3 (binary upside or downside) |
Risk 1 — Capacity market reform
State-level political pressure on PJM is intense. A re-engineered auction with lower price caps, demand-response priority, or load-shifting mandates could compress the wholesale tail meaningfully. Watch Pennsylvania, Maryland, and Virginia legislatures through 2026–27. The most likely outcome is incremental tightening — for example, a graduated price cap reduction over 3–5 years — rather than a fundamental redesign. But the political risk premium is real. A 30–40% capacity price reduction would compress Track 1 EPS by ~10–15% versus current consensus.
Risk 2 — Hyperscaler capex slowdown
The AI capex cycle is correlated to hyperscaler earnings. A meaningful enterprise-software demand pause would cascade to power. The 2030 deployment forecasts assume continuous compute scaling; this is not guaranteed. The 2025 enterprise IT spending environment has been resilient but recall the cloud capex pause in 2019–20. A 12-month pause in hyperscaler builds would not break the thesis but would substantially compress Track 1 and 3 valuations — both rerate to historic ranges in such a scenario.
Risk 3 — Cost-of-capital sensitivity
Utility names with 60–65% debt-to-cap ratios are highly sensitive to long-end yields. A 100 bps move in 10-year Treasuries compresses regulated utility multiples by ~10–15%. The Federal Reserve’s path through 2026–27 is the dominant macro input here. Our base case anticipates the curve roughly stable through 2026 with measured cuts; a hawkish surprise could compress Tracks 2 and 4 materially. Track 1 is partially insulated because hyperscaler PPAs are largely contracted at fixed prices and capacity payments are not directly rate-sensitive.
Risk 4 — Regulatory backlash on rate increases
Average residential bills are rising. Multiple states (Virginia, Oregon, New Jersey) have introduced or are considering data-centre-specific rate classes that shift costs from households to large-load customers. This is rational and arguably overdue, but it could compress merchant generator margins on new contracts and slow rate-base growth at affected utilities. Specifically, Virginia legislation could materially affect Dominion’s data centre tariff structure. Track 2 names with the largest data centre concentration are most exposed.
Risk 5 — Gas turbine supply chain
The single largest execution risk. If GE Vernova, Siemens Energy, or Mitsubishi Power cannot scale production, gas-heavy build plans slip and merchant nuclear holds pricing power longer — but the broader sector earnings trajectory disappoints. This is genuinely binary for Track 3. If supply chain resolves on schedule, GE Vernova is a 2–3× return story by 2028. If it slips further (gas turbine lead times rising from 60 months to 72+ months), the names compress materially. Currently rated as 50/50 in our base case.
07 · Catalysts watch & what to monitor
The trade is now well-mapped. From here, the question is execution. Below is the calendar of catalysts we monitor through 2026–27.
| Timing | Catalyst | Impact |
|---|---|---|
| Q3 2026 | PJM 2028/29 Base Residual Auction | Confirms (or breaks) the $329 capacity price plateau |
| Q4 2026 | Three Mile Island restart commercial date | First major restart proof point; CEG validates |
| 2027 | First NuScale module commercial operation | SMR thesis validation; TVA/ENTRA1 pipeline |
| 2027–28 | PJM capacity market reform decisions | Track 1 multiple expansion/compression |
| 2028 | US presidential election | Scenario B trigger; Climate-Led Decarbonisation |
| 2029 | Duane Arnold restart commercial date | Second major restart proof point |
| Ongoing | Hyperscaler PPA flow; quarterly earnings | Track all five tracks for execution |
Bottom line · what Part III concluded
The American utility sector is undergoing the most violent re-rating it has experienced in two decades. Part I established the foundations — how the machine works, what choke points constrain expansion. Part II documented the inflection — demand shock, capacity repricing, climate vulnerability, narrative shift, and the three scenarios for 2035. The bonus post on Other Grids placed the US system in international context. Part III translates everything into actionable portfolio positioning.
The five-track framework — nuclear merchants, regulated rate-base, equipment OEMs, climate resilience, SMR optionality — provides scenario-robust exposure across the three Part II outcomes. The portfolio blueprint allocates 25–30% to nuclear merchants for AI-tail growth, 30–35% to regulated names for defensive growth, 15–20% to equipment OEMs as picks-and-shovels, 10–15% to climate resilience as asymmetric value, and 5% to SMR optionality. The portfolio is constructed for the central tendency of our Part II framework — predominantly Hybrid Resilience and AI Abundance outcomes — while preserving exposure to a Climate-Led pivot through Track 5 and Track 2 diversification.
The five risks — capacity market reform, hyperscaler capex slowdown, cost-of-capital shock, regulatory backlash on rates, gas turbine supply chain — could compress the trade but do not break the thesis. The catalysts to monitor through 2026–28 include PJM auction outcomes, first restart proof points, the 2028 election, and ongoing hyperscaler PPA flow.
The trade is not subtle. It is not over. Position accordingly. Diversify across the five tracks. Watch the regulatory dockets. Monitor the catalysts. Stay disciplined on entry — utility re-ratings are violent but episodic.
Power & Markets — the full trilogy
The Power & Markets series has covered the American electricity system from physics to portfolio. Part I established the foundations. Part II mapped the inflection. The bonus post placed the US in international context. Part III translated everything into positioning. The series is now complete; ongoing coverage will appear as standalone Learning Series posts and quarterly updates.
Forthcoming standalone primers: UK Power Markets (deep dive); India Power Markets (deep dive); and continued ENSO & Climate Markets coverage through Part III of that series.
G · Glossary additions (cumulative from Parts I, II, and Bonus)
New terms introduced in Part III. Full glossary spans across all four posts.
| EQUITY & PORTFOLIO | |
| S5UTIL | S&P 500 Utilities sector index. Ticker for the GICS Utilities sub-index. |
| Forward P/E | Price-to-earnings ratio using next-twelve-month consensus EPS estimate. |
| GICS | Global Industry Classification Standard. S&P/MSCI sector classification framework. |
| Track (in this post) | Distinct equity exposure type within the portfolio framework; five used here. |
| HALEU | High-Assay Low-Enriched Uranium. 5–20% U-235 enrichment level; required for many advanced reactors. |
| NNSA | National Nuclear Security Administration. US DOE agency overseeing nuclear weapons stockpile and defence nuclear fuel cycles. |
S&P Dow Jones Indices — S&P 500 Utilities sector index data. FactSet — sector EPS estimates Q3 2025; analyst consensus. MacroMicro — sector forward P/E series. Deutsche Bank Wealth Management — S&P 500 EPS Tracker Q3 2025 (November 2025). Constellation Energy Corporation — earnings releases and management guidance FY2026. Vistra Corp — earnings releases and Cogentrix acquisition disclosures. Talen Energy Corporation — Amazon Susquehanna deal disclosures. NextEra Energy — Duane Arnold restart announcement; data centre PPA disclosures. Public Service Enterprise Group — nuclear segment disclosures. Dominion Energy — Northern Virginia data centre pipeline disclosures. Southern Company — Georgia large-load pipeline disclosures. Duke Energy, Xcel Energy, Entergy — investor materials. GE Vernova — Q1 2026 earnings release; data centre order backlog. Eaton, Quanta Services, MasTec, Vertiv, Generac — investor materials. Pacific Gas & Electric Company — 2026–28 Wildfire Mitigation Plan; rate-base growth disclosures. Edison International (SCE), CenterPoint Energy — resilience capex programmes. NuScale Power, BWX Technologies, Cameco, Centrus Energy, Oklo — investor materials. PJM Interconnection — Base Residual Auction results. Monitoring Analytics LLC — independent market monitor reports. Fenrir Research prior publications: Power & Markets Part I — Foundations, Power & Markets Part II — Inflection, Power & Markets Bonus — The Other Grids, ENSO Primer, ENSO Markets & Portfolio, War & Markets.
This analysis is for informational purposes only. Not investment advice. All probability estimates, conviction ratings, and forward-looking judgements are analytical synthesis based on cited sources. Stock-specific references and portfolio allocations are illustrative of the investment framework discussed and are not buy or sell recommendations. Forward P/E figures and other valuation metrics are approximate as of May 2026 and subject to change. The author and Fenrir Research may hold positions in securities mentioned. Investors should conduct their own due diligence and consult qualified professionals before making investment decisions.
Leave a Reply