Energy
A 90%-regulated New Jersey wires utility wearing a merchant-nuclear data-center costume — you are paid ~17x for a 6–8% regulated compounder, and the AI-power optionality is real but unpriced *and* unproven; own it as a rate-base bond with a free nuclear call option, not as the next Constellation.
Research
The verdict
A 90%-regulated New Jersey wires utility wearing a merchant-nuclear data-center costume — you are paid ~17x for a 6–8% regulated compounder, and the AI-power optionality is real but unpriced *and* unproven; own it as a rate-base bond with a free nuclear call option, not as the next Constellation.
PSEG is a New Jersey public-utility holding company with two reportable segments and a structurally simple story: a big regulated wires-and-pipes utility, plus a merchant nuclear fleet that has been progressively de-risked into a quasi-contracted cash engine.
. Distribution is rate-regulated by the **NJ Board of Public Utilities (BPU)**; transmission earns a **FERC formula rate** (annual true-up, revenue *not* tied to sales volume). FY2025 segment NI **$1,745M** (FY24 $1,547M, FY23 $1,515M) .. Also a full-requirements gas supply (BGSS) contract with PSE&G, the **PSEG Long Island** O&M contract with LIPA (cost-pass-through + a fixed/variable fee; ~$644M of Servco pass-through revenue in 2025), and legacy lease investments (Energy Holdings). FY2025 segment NI **$366M** (FY24 $225M, FY23 $1,048M — but 2023 was inflated by a **$959M after-tax non-trading MTM gain**, vs −$54M in 2025) .Contract structure / key terms. This is the part bulls under-appreciate. PSE&G's distribution carries a CIP/decoupling mechanism, so "minimal impact from sales volumes on most distribution delivery revenues" — weather and conservation don't dent the regulated earnings . Commodity (BGS/BGSS) and clause revenues are **pass-through, zero-margin** — PSE&G "earns no margin." So PSE&G's economics are pure rate-base × allowed ROE, not a volumetric merchant. The nuclear fleet carries a **federal Production Tax Credit (PTC) floor through 2032** (up to **$15/MWh**, inflation-adjusted) — downside price protection that converts merchant nuclear into something closer to a floored-revenue asset .
Plain terms: ~83–90% of earnings are a regulated bond that grows with rate base; the rest is a PTC-floored nuclear call option on PJM power + the data-center demand wave. Headquartered Newark, NJ; CEO Ralph LaRossa; ~$57.6B total assets ``.
A utility's "supply chain" is really its input fuel chain + capital/equipment chain + regulatory-recovery chain + load (customer) chain. Named stakeholders along each:
Upstream — fuel & inputs:
. The Salem 1/2 units are **jointly owned** (PSEG operates Salem; Hope Creek is wholly PSEG; Peach Bottom interest also held) — co-owners include **Constellation/Exelon legacy interests** in the Salem/Peach Bottom complexes .. The 2026–2030 program is **$22.5–25.5B regulated** ($24–28B total) — a multi-year demand on a supply-chain-constrained market for grid hardware . Management explicitly flags tariffs on imports as a supply-chain risk to the capital program ``.Midstream — the company: PSE&G (wires) + PSEG Power (nuclear) + PSEG Services (shared services at cost).
The recovery chain (the real "supplier" of returns): NJ BPU (distribution base rates, clause programs, GSMP, CEF-EE) and FERC (transmission formula rate, PJM ROE adder). These two bodies are the mechanism by which capex converts to earnings. Chokepoint: a hostile BPU or a FERC ROE-adder repeal directly compresses the return. ``
Downstream — load / customers: 2.4M electric + 1.9M gas retail customers (the captive base) plus the new vector: hyperscaler data-center load. The large-load interconnection pipeline at PSE&G has gone from 400 MW (early 2025) → 4.7 GW (mid-2025) → 9.4 GW (Q1-2026), ~90% data centers . End buyers behind that pipeline are the usual names — **Amazon/AWS, Microsoft, Google, Meta** — though PSEG cautions in-state commitments have lagged without state tax incentives, pushing some load to neighboring zones .
Single-source / chokepoints: (1) the NJ BPU is the single regulatory counterparty for ~83% of earnings — concentration risk by design; (2) PJM capacity-market design (auction price formation, co-location rules) gates the nuclear upside; (3) the U.S. Treasury PTC "gross receipts" guidance is an unresolved single point that could change recorded nuclear revenue. Names or it didn't happen — those three are the chokepoints.
This is a regulated-monopoly + irreplaceable-asset moat, not a product moat.
Moat verdict: Wide and durable on the regulated wires (franchise + rate base). The nuclear "moat" is really scarcity optionality — high value, but its monetization depends on PJM rules and a counterparty signing, neither of which PSEG fully controls.
segments.csv is header-only, so all figures sourced to the filing tables. PSEG reports NI by segment (its CODM metric), not segment EBITDA — so I show NI and revenue.
Net income by segment ($M) ``:
| Segment | FY2023 | FY2024 | FY2025 | FY25 share | Trend |
|---|---|---|---|---|---|
| PSE&G (regulated) | 1,515 | 1,547 | 1,745 | 83% | Steady accel — rate-case + clause programs |
| PSEG Power & Other | 1,048 | 225 | 366 | 17% | Noisy — driven by MTM/NDT swings |
| PSEG total NI | 2,563 | 1,772 | 2,111 | 100% | — |
| Diluted EPS | $5.13 | $3.54 | $4.22 | — | 2023 flattered by MTM |
The PSE&G line is the signal; the Power line is the noise. The $959M after-tax non-trading MTM gain in 2023 (vs −$151M in 2024 and −$54M in 2025) is why 2023 GAAP EPS looks artificially high — strip MTM/NDT and the underlying trend is up and to the right . This is exactly why management and the Street steer on **non-GAAP operating EPS** ($3.68 FY24 → $4.05 FY25 → $4.28–4.40 guide FY26) .
Revenue by source ($M, consolidated) ``:
Geography: ~single-jurisdiction (New Jersey) for the regulated business; nuclear sells into PJM; the one out-of-state regulated lever is the $424M PJM-awarded MD/N.VA transmission project (2027 in-service now doubted) ``. Geographic concentration in NJ is a feature (deep, constructive history) and a risk (one BPU, one state's politics).
. ~**95% of 2026 nuclear output hedged** .Balance-sheet flags (FY2025 full-year, the cleaner read) ``:
Market reaction: the stock "rose"/"reacted modestly" on the Q1 beat `` — i.e. a clean print but no re-rate, consistent with a name where the narrative (data centers) is doing the heavy lifting, not the quarter.
Unusual vs own history: nothing alarming. The only oddity is the GAAP-to-non-GAAP gap driven by nuclear MTM/NDT — recurring and well-flagged.
No transcripts/ on disk — sentiment is ``, triangulated across the last ~4 prints (Q4-24 → Q1-25 → Q4-25 → Q1-26).
. Through 2025 the data-center pipeline became the headline — by Q1-2026 management is guiding **7% growth, raising long-term to 6–8%, and leading with the 9.4 GW large-load pipeline** .Peer set: large regulated/utility-holding names, split into (a) regulated-T&D compounders and (b) merchant-nuclear "AI power" pure-plays — because PSEG straddles both and the multiple gap between the two camps is the entire valuation debate. Multiples are `` with date or n/a.
| Company | Ticker | Mkt cap | Fwd P/E | Div yield | Camp / note |
|---|---|---|---|---|---|
| Public Service Enterprise | PEG | ~$39.8B | ~17.7x (on ~$4.05 ttm; ~18.2x on FY26 mid $4.34) | ~3.3% | Hybrid: ~83–90% regulated + 3.7 GW nuclear `` |
| Exelon | EXC | n/a | n/a | ~3.65% | Pure regulated T&D (post-Constellation spin) `` |
| WEC Energy | WEC | ~$36.9B | n/a | ~3.0% (≈$0.953/qtr) | Regulated compounder `` |
| Xcel Energy | XEL | n/a | n/a | ~2.96% | Regulated + renewables `` |
| DTE Energy | DTE | ~$30B (EV ~$57B) | ~19.6x | n/a | Regulated `` |
| Duke Energy | DUK | n/a | n/a | ~4%+ | Regulated `` |
| Southern Co | SO | n/a | n/a | ~4%+ | Regulated `` |
| Constellation Energy | CEG | large (≈$100B+); ~$323/sh Mar-26 | n/a (rich; ~20% EPS boost FY26) | low (~0.5%) | Merchant-nuclear pure-play, +$22B Calpine, 22 GW nuclear `` |
Read: PEG at ~17–18x forward sits roughly in line with the regulated cohort (DTE ~19.6x; the group broadly mid-to-high-teens) and yields more (~3.3%) than the pure compounders that yield ~3% — i.e. you are not paying a nuclear-AI premium. The contrast with CEG is the whole story: Constellation re-rated violently on the merchant-nuclear/data-center thesis, while PEG's 3.7 GW of the same scarce asset is trapped inside a regulated holdco and valued at a wires multiple. That is either (a) the market correctly discounting that PSEG's nuclear is PTC-floored, regulated-adjacent, and not yet contracted to a hyperscaler, or (b) an unpriced re-rate option. Both can be true. Precise EV/EBITDA and full forward P/E for several peers are not sourced — do not fabricate; pull from a terminal before underwriting a relative-value trade.
Mostly ``; the pattern, not a tick-by-tick log:
. Conversely, **Jefferies' downgrade to Hold (PT $89) on *reduced confidence in nuclear-DC deal conversion*** is the bearish mirror .What the market actually reacts to: (1) PJM capacity/policy + data-center demand signals (the upside narrative), (2) NJ regulatory/affordability politics (the downside governor), and (3) rate-base/guidance updates (the steady compounding). Pure quarterly EPS beats barely move it — the stock trades on the nuclear-AI option and NJ political risk, bracketing a regulated core.
. The result: PSE&G now drives ~90% of operating earnings and the multiple/predictability improved. This is a CEO who **shrank to quality** rather than empire-built — exactly the behaviour you want from a utility steward. ; PSE&G distribution allowed ROE **9.6%**, transmission **9.90% + 50 bps PJM adder** . Solidly earning its allowed returns.insider-transactions.csv not on disk — insider ownership not sourced (typical low single-digit % for a large-cap utility; do not fabricate). 43,642 registered holders; ~498.7M shares out ``.Archetype: seasoned professional operator-stewards, long-tenured, disciplined capital allocators. For a regulated-utility-plus-nuclear-option business at this stage, that is the right team — you want predictability and clean execution, not a visionary swinging for fences.
Acting as a forensic analyst. For a rate-regulated utility, the accounting risk surface is regulatory assets/liabilities, AROs, pension, MTM/derivatives, and the GAAP-vs-non-GAAP gap — not channel-stuffing or aggressive revenue recognition (revenue is tariff-driven, recognised as service is delivered).
. Collateral call exposure of **~$691–703M if PSEG Power loses IG rating** — a contingent liquidity item, currently well-covered by $3.75B facilities .Regulatory findings (required sub-section):
Net: clean books, clean auditor, no SEC enforcement history. Real items to monitor are the (1) July-2025 nuclear-wage-fixing class action, (2) PTC/Treasury estimation sensitivity, and (3) NJ regulatory/affordability overhang — none of which impugns the accounting.
Built bottom-up from FY2025 actual operating EPS $4.05 and the company's reaffirmed FY2026 guide $4.28–$4.40 (mid $4.34) + the 6–8% long-term growth framework, which is itself underwritten by the 6.0–7.5% rate-base CAGR to 2030 ``. Fiscal year = calendar year.
Input lines (all labelled):
| Scenario | FY2026E | FY2027E | FY2028E | Logic |
|---|---|---|---|---|
| Bull | $4.40 (top of guide) | ~$4.77 | ~$5.18 | 8% growth holds; a nuclear-DC contract lands and lifts Power; top-of-range capex executes. `` |
| Base | $4.34 (guide mid) | ~$4.64 | ~$4.96 | 7% midpoint compounding; regulated rate base does the work; no DC contract yet priced in. `` |
| Bear | $4.28 (bottom) | ~$4.49 | ~$4.71 | 5% growth; ROE adder lost (−$40M ≈ −$0.08), rate-case drag, higher rates, no nuclear monetization. `` |
Valuation cross-check: at ~$80 and base FY26 $4.34, PEG trades ~18.4x — in line with regulated peers; the ~$90 Street median PT implies ~20–21x FY26 / ~10% upside plus a ~3.3% yield ``. The asymmetry is in the un-modelled nuclear-DC option: a signed hyperscaler offtake could add several $/share of Power earnings and a multiple re-rate toward the CEG cohort — but it is not yet contracted, and Jefferies just cut on exactly that doubt.
(No forecast.ts create — per --watchlist rules, only log a Brier forecast on genuine commitment. The scoreable base call would be: "PEG FY26 non-GAAP operating EPS ≥ $4.34, p≈0.62, resolves 2026-12-31" — left for the /thesis pass.)
Bull case. PSEG is a rare hybrid: a constructive-jurisdiction regulated compounder (6–8% EPS, ~3.3% yield, decoupled from volumes) bolted to the single scarcest asset in the AI-power era — existing, licensed, carbon-free baseload nuclear in the load pocket where 9.4 GW of data-center demand is queuing. You're paid a regulated multiple for the wires and handed the nuclear-AI option for free. Management has spent three years de-risking (offshore-wind/fossil/solar exits) into exactly this clean profile, the PTC floors the downside through 2032, and a single hyperscaler offtake on Salem/Hope Creek could re-rate the Power segment toward the Constellation cohort. Rate base compounds regardless of whether the option pays.
Bear case (permanent-impairment risks). (1) NJ political/affordability backlash — capacity prices up ~10x have already forced bill credits, a legislated summer-shutoff moratorium, and a hostile-leaning new administration (took office Jan 2026) with an uncertain Energy Master Plan; a punitive BPU is the one thing that can structurally lower the allowed return on 83% of earnings. (2) The nuclear-DC option may never monetize — PSEG itself says in-state commitments have lagged without tax incentives, FERC co-location rules are unsettled, and Jefferies cut to Hold on reduced deal-conversion confidence; if the option expires worthless, you own a ~17x wires utility, not a re-rate. (3) Funding drag — structurally outspends cash flow; rising rates + ATM dilution nibble EPS growth.
Pre-mortem (18 months out, thesis broke). The NJ governor and BPU, under affordability pressure, claw back at PSE&G's ROE/equity layer in the next rate case; PJM's "reliability backstop auction" allocates costs punitively to NJ; no hyperscaler signs a Salem/Hope Creek offtake (co-location economics or Treasury PTC ambiguity kill it); the AI-power narrative rotates to merchant pure-plays (CEG/VST) that can actually contract; PEG de-rates to a low-teens multiple on a "policy-risk NJ utility" label and dead-money the yield.
Are multiples too high? No — ~17–18x forward is fair-to-slightly-cheap for a 6–8% regulated grower with this yield; you are not overpaying for the option. The risk is fundamental (NJ politics), not valuation.
Contrarian view (what the market refuses to see): The Street is bucketing PEG as either a sleepy regulated utility or a failed-to-launch AI-power wannabe (the Jefferies cut). The non-consensus read is that the regulated core alone justifies the price, so the nuclear-DC option is a genuinely free call — and the market is so fixated on whether a flashy co-location deal lands that it under-credits the boring, near-certain rate-base compounding plus the PTC-floored nuclear cash that shows up with or without a hyperscaler signature.
Dismantling the bull case.
A consolidating monopolist on a leveraged treadmill — RUN's GAAP "profit" is an HLBV mirage, but the OBBBA's asymmetric kill of 25D (not 48E) hands the TPO leader the residential market it can't yet profitably finance; the bet is whether ~$15B of non-recourse debt rolls before rates or a securitization-market hiccup forces a dilutive reset.
A regulated-utility levered call on the Georgia data-center build-out — the cleanest large-cap way to own AI power demand, but priced as if the affordability politics and equity dilution won't bite; own the growth, respect the ~24x multiple.
A real turnaround that has already been paid for — six straight quarters of margin repair and the return to positive operating cash flow are genuine, but at ~$60 the stock prices in a clean, AMPTC-independent recovery the filings explicitly say does not yet exist (ex-45X credits, SolarEdge is still gross-loss-making), so the asymmetry from here is poor.