NRG Energy

Phase A — Understand the business

Lens 1 · Company Overview

What it is. NRG Energy (Houston, TX; NYSE/NYSE Texas: NRG; CIK 0001013871; Delaware incorp.; founded as a competitive power company, auditor KPMG LLP ) is one of the largest competitive (non-regulated) energy retailers in the U.S., fused with an owned generation fleet and a smart-home platform. It serves ~8 million residential customers (6M retail energy + 2M Vivint smart home) plus large C&I, data-center, and wholesale customers, under brands NRG, Reliant, Direct Energy, Green Mountain Energy, and Vivint.

How it makes money — the integrated model. The economic engine is a retail load book hedged by owned generation, run by geography:

• Texas — fully integrated: NRG's own plants supply a portion of its own retail customers, which "reduces the need to sell electricity to, and buy electricity from, other institutions… resulting in more stable earnings and cash flows, lower transaction costs and less credit exposure". This natural hedge is the structural advantage — NRG owns both the customer and the megawatt in ERCOT.
• East and West/Other — retail + wholesale + capacity revenue (East benefits from PJM/ISO capacity payments).
• Vivint Smart Home — hardware + subscription monitoring, ~9-year average customer life, 37M+ connected devices (~16 per household).

Scale (FY2025). Sold 154 TWh of electricity and 1,857 MMDth of natural gas; recurring sales in 25 U.S. states + DC + 8 Canadian provinces; Vivint in all 50 states. Largest share of competitively served residential electric customers in Texas; leading B2B power/gas provider in North America.

Contract structure / payment terms. Home customers contract 1 month to 5 years (fixed, indexed, or month-to-month); Business customers 1–5 years. Critically for concentration risk: "The Company had no customer that comprised more than 10% of the Company's consolidated revenues" in 2025, 2024, or 2023. Revenue is granular millions-of-meters retail, not a few mega-contracts — though that is changing as data-center electric-service agreements scale (see Lens 3).

The two transformational moves that define today's NRG:

1. LS Power ("LSP Portfolio") — closed Jan 30 2026. 18 natural-gas-fired and dual-fuel facilities, ~13 GW across 9 states, plus CPower, a leading commercial-&-industrial demand-response / VPP platform. $12.0B enterprise value; "expands NRG's generation footprint by more than 50%".
2. GE Vernova + Kiewit (TIC) Project Development Agreement — Feb 13 2025. Develop and construct up to 5.4 GW of new gas-fired combined-cycle generation, owned and operated by NRG, with reserved turbine slots — the organic new-build arm of the data-center strategy.

Verdict (L1): A real, cash-generative, two-sided business (customer + megawatt) — not a speculative power-pure-play. The integrated retail hedge is the durable part; the LSP/GEV buildout is the growth option strapped on top.

Lens 2 · Supply Chain

Map: fuel & equipment upstream → NRG generation + supply desk → retail/wholesale load → end customer. Named stakeholders along the chain (names or it didn't happen):

Upstream — fuel & equipment

• Natural gas (primary fuel): mid-merit + peaking fleet, sourced on a spot basis ("not prudent to forward-purchase… as dispatch is highly unpredictable"), with storage + transportation contracts to dampen daily volatility. NRG runs a producer-services business reaching "from the wellhead to end-use customers".
• Coal (declining): ~15 million tons purchased in 2025, all Powder River Basin; forward-contracted for 2026; rail transport + rail-car leases secure PRB transport "for the next four years". Chokepoint: rail logistics for PRB coal.
• Turbines / EPC — the binding constraint of the era: GE Vernova (turbines) + Kiewit / TIC (engineering, procurement, construction) under the 5.4 GW development agreement; the Texas Energy Fund (TEF) finances the three in-state new-build projects. In an AI-power buildout, reserved GE Vernova turbine slots are themselves a moat — gas-turbine lead times now run multiple years.
• LS Power — counterparty/seller of the 13 GW LSP Portfolio and, post-deal, a stockholder (received $2.8B in NRG stock; subsequently sold part of the stake into a March 2026 offering — see L5).

Midstream — NRG itself

• ~25.8 GW generation post-LSP (was ~12 GW competitive / ~13 GW operated pre-deal) at plants across Texas, the East, and West/Canada; plus ~6,200 MW operated on behalf of third parties at 13 plants.
• Market Operations (power + gas commercial desks) hedges via forwards, futures, options, swaps, fuel-supply and transport contracts, and capacity arrangements.

Downstream — load & end customers

• Retail: ~8M residential (NRG Home / Reliant / Direct Energy / Green Mountain) + C&I (NRG Business).
• Wholesale / capacity buyers: ERCOT, PJM, ISO-NE, CAISO, AESO (Alberta), MISO.
• New end-customer class — data centers / hyperscalers: via Bring-Your-Own-Power (BYOP) and electric-service agreements; ~10 GW already contracted under electric service agreements out of a >36 GW pipeline (L3).

Chokepoints / single-source dependencies: (1) Gas-turbine + EPC capacity (GE Vernova / Kiewit) — the gating item for organic GW adds; (2) PRB coal rail transport; (3) ERCOT market design + interconnection queue — both the opportunity and the regulatory chokepoint (Lens 10); (4) fuel-price volatility flows straight through an under-hedged gas peaker fleet (by design — they don't forward-buy gas for peakers).

Verdict (L2): Concrete and named. The supply-chain risk that matters most is no longer fuel — it's turbine/EPC availability for the buildout, where NRG has front-run peers by locking GE Vernova slots early.

Lens 3 · Competitive Advantages (moats)

The moat stack, strongest → weakest:

1. Integrated retail-generation hedge in ERCOT (durable). Owning both the Texas customer and the Texas megawatt is the structural edge — it converts merchant-power volatility into "more stable earnings and cash flows" and slashes collateral/credit exposure. Pure-play IPPs (Talen) and pure-play retailers each carry one leg of this risk; NRG carries the natural offset.
2. #1 competitively-served residential electric share in Texas + scale brands. "Largest share of competitively served residential electric customers in Texas," multi-brand (Reliant, Green Mountain, Direct Energy). Retail energy is a switching-cost + brand + CAC moat: ~8M customers, average Home electricity+gas customer base ~5.6M, sticky billing relationships.
3. The new moat — large-load / data-center positioning. Management frames NRG's edge as the combination of retail scale + flexible demand programs (CPower VPP) + dispatchable gas generation. The prize: a >36 GW large-load request pipeline by 2033 in NRG's served markets, ~10 GW already contracted under electric service agreements, against an ERCOT large-load interconnection queue that has ballooned to >230 GW (≈77% data centers). Whoever can offer firm, fast, dispatchable power to a hyperscaler wins multi-year contracted cash flow — and NRG owns the gas fleet + the turbine slots to deliver it.
4. Reserved turbine + EPC capacity (time-limited but real today). The GE Vernova/Kiewit 5.4 GW agreement is a moat while turbine lead times are long — a competitor can't simply order 5 GW of gas turbines and catch up this cycle.
5. Vivint smart-home cross-sell (weakest / contested). Pitched as the intersection of energy + home (VPP enrollment, 9-yr customer life). In practice it is the asset Elliott called the worst deal in the sector (Lens 9). The strategic logic — bundle energy + home to extend customer life — is plausible; the price paid was not.

Bargaining power (who needs whom more):

• vs. customers (retail): moderate — millions of switchable small customers; pricing power comes from brand + service, not lock-in. But vs. data centers, the balance flips toward NRG: firm dispatchable power is scarce, hyperscalers are time-desperate, and NRG can demand long-duration take-or-pay-style electric-service agreements.
• vs. suppliers: strong on gas (multi-supplier, spot); constrained on turbines/EPC (GE Vernova/Kiewit are scarce — but NRG holds reserved slots, so it sits ahead of the queue).

Verdict (L3): The real, defensible moat is integrated retail + dispatchable gas in a structurally short ERCOT, now levered to data-center demand. It is a real moat, but it is a gas + switching-cost + scarcity-timing moat — not the scarcity-value-of-nuclear moat that lets Constellation command a premium multiple (Lens 7). That distinction is the entire valuation debate.

Lens 4 · Segments

FY2025 vs FY2024, by reportable segment — every figure ``:

Revenue mix (FY2025, ): Total retail revenue $29,543M (Home $12,967M, Business $16,576M) dwarfs Energy ($590M) + Capacity ($280M) + Other ($294M). NRG is ~96% a retail revenue company by the top line — generation shows up as a cost-of-supply hedge, not a revenue line. Geographically the revenue is Texas + East ≈ 83% of the total.

Trend & cause:

• Texas — steady, high-margin, the anchor. Operating income $1,113M on $11.1B rev (~10% op margin) vs East's ~5%. The integrated model shows up here.
• East — fastest revenue growth (B2B power/gas), but margin-light; capacity revenue concentrated here.
• Vivint — the segment that looks unprofitable on a GAAP basis purely because of $810M of acquisition-driven D&A and $3.5B of goodwill. On a cash/EBITDA basis it contributes; on reported earnings it's a drag and a perennial impairment-watch item (Lens 10).
• West/Other — managed decline; capital is rotating east and into Texas new-build.

Pro-forma note: FY2025 segment figures are pre-LSP (closed Jan 30 2026). From 2026, the Texas + East generation lines step up materially — Q1-2026 total revenue already $10,256M with 11 months of LSP ahead. The reported segment mix will tilt further toward owned-generation EBITDA.

Verdict (L4): Texas is the profit engine; East is the growth-but-thin-margin volume; Vivint is a GAAP-depressed, goodwill-heavy bolt-on. Segment data fully ``-sourced — no paraphrase.

---

Phase B — Measure performance

Lens 5 · Earnings Result (latest print: Q1-2026, reported 2026-05-05)

The headline tension: a revenue blowout next to a GAAP/adjusted earnings drop — and the stock went UP.

What drove it. The GAAP EPS collapse is mostly mark-to-market noise on a now-much-larger hedge book, not operational deterioration. On the adjusted line, the miss was real and explained by two weather/market items: (1) milder Q1 weather in Texas (lower retail load/margin) and (2) elevated supply costs in the East from Winter Storm Fern — both partially offset by the newly consolidated LSP fleet. Adjusted net income ~$308M.

Guidance — reaffirmed, and it's the real story. Alongside a weak print, NRG reaffirmed full-year 2026 guidance:

• Adjusted EBITDA $5.325–5.825B — raised from the pre-LSP $3.925–4.175B standalone range (the ~$1.4–1.65B step-up ≈ 11 months of LSP).
• Adjusted EPS $7.90–9.90 (initiated with LSP, ~11 months of ownership).
• Free Cash Flow before Growth $2,800–3,300M.
• Long-term: ≥14% annual growth in adjusted EPS and FCF/share over five years, excluding any upside from large-load contracts.

Balance-sheet flags (FY2025 actuals, ):

• Total long-term debt (incl. current) $16,565M at YE2025 vs ~$10,880M at YE2024 — up ~$5.7B as NRG pre-funded LSP ($6,676M of long-term debt issued in FY2025).
• Operating cash flow $1,913M (FY25) vs $2,306M (FY24) — down $393M.
• Cash + counterparty deposits + restricted at end of FY25: $4,998M (vs $1,173M) — the pre-funding war chest.
• Series A Preferred: 650,000 shares of 10.25% Fixed-Rate Reset Cumulative Redeemable Perpetual Preferred, $1,000 liquidation preference = $650M — a pricey 10.25% coupon, senior to common, used in the financing stack.
• Interest expense FY25 $741M vs FY24 $651M (and rising with the LSP debt).

Market reaction: Shares rose on the revenue beat despite the adjusted-EPS miss — a tell that the market is trading NRG on the LSP/guidance/data-center narrative and the buyback, not the quarterly weather print. That is both the bull's confirmation and the bear's warning (expectations have decoupled from current earnings).

Verdict (L5): A messy, weather-hit print papered over by a reaffirmed, LSP-supercharged full-year guide. The number that matters is the FY2026 EBITDA midpoint ~$5.575B and the ≥14% per-share growth commitment — both ``-sourced from management, both unproven on the new fleet.

Lens 6 · Earnings Calls (sentiment trend)

Transcripts are absent from the shelf (transcripts/ empty); sentiment below is ``-sourced from the Q1-2026 Benzinga transcript and call summaries.

Where management is focused (Q1-2026 call):

• Contracted, long-duration cash flows. Direct quote in substance: "definitely putting more focus around contracted cash flows, looking for duration of cash flows with counterparties," which "leads to things like data center deals and new build generation". This is the strategic pivot in one sentence — from merchant volatility to contracted infrastructure cash flow.
• Large-load pipeline as the growth narrative: >36 GW of large-load requests by 2033; ~10 GW contracted; 1 GW VPP target in Texas; TEF new-build (3 projects ~1.5 GW, ~300,000 homes at peak) on schedule.
• Capital return discipline: $1B/yr buybacks until <3.0x leverage, then 80/20 framework (Lens 9).

Tone shift over time (the multi-call arc):

• 2023 (Gutierrez era / Elliott fight): defensive, on the back foot over Vivint.
• 2024 (Coben): "refine strategy + execution," capital return, balance-sheet repair — credibility-rebuilding tone.
• 2025–Q1 2026 (Coben → Gaudette transition): confident, growth-forward, data-center-and-LSP-centric. The recurring new phrases: "contracted cash flow," "duration," "large load," "data center," "VPP." What they stopped saying: the apologetic Vivint-integration framing of 2023.

Caveat: sentiment is web-derived, not transcript-grounded — and it spans a CEO handoff (Gaudette took over April 30 2026), so the next call is the first under new leadership. Flag to re-ground with the actual transcript on refresh.

Verdict (L6): Tone has traveled from defensive (2023) → disciplined repair (2024) → confident growth (2025–26), with language converging hard on "contracted data-center cash flow." Conviction-positive, but unverified against primary transcripts and now carrying handoff risk.

Lens 7 · Comps

Peer set: the listed competitive-power / IPP complex (NRG's index had no peers registered, so peers + multiples are ``).

Sources: NRG ~$150.59, mkt cap ~$35.1B, EV ~$47.1B, fwd P/E ~18x, TTM EV/EBITDA ~15.9x, TTM P/E ~40.9x. CEG fwd ~26x P/E / ~13x EV/EBITDA; VST fwd ~18x P/E / ~10x EV/EBITDA, net debt ~$19.6B. TLN mkt cap ~$19.3B, EV ~$25.1B, TTM EV/EBITDA ~30.7x, 2026 EBITDA guide $1.75–2.05B, <3.5x net-lev target. Dividend yield: NRG $1.90/sh on ~$150 = ~1.3%; 5-yr avg ROE n/a (distorted by the FY2023 GAAP loss and Vivint goodwill — not meaningful without normalization).

Read. On forward EV/EBITDA, the right denominator is the FY2026 guide midpoint ~$5.575B against EV ~$47.1B → ~8.4x forward EV/EBITDA. That puts NRG below Vistra (~10x) and well below Constellation (~13x) on forward EBITDA — i.e., NRG is the cheapest large-cap way to play data-center power, which is exactly why analysts skew bullish (consensus ~Strong Buy, avg target ~$199–202, range $96–354). The discount is earned: NRG has no nuclear (no zero-carbon scarcity premium), the highest leverage trajectory of the group post-LSP, and integration risk on a doubled fleet.

Verdict (L7): NRG screens as the value name in the AI-power basket — ~8.4x forward EBITDA vs CEG ~13x / VST ~10x. The gap is the market pricing gas-not-nuclear + integration + leverage. If LSP delivers and large-load contracts convert, the multiple gap is the upside; if integration slips, the gap is justified.

Lens 8 · Stock-Price Catalysts (what actually moves NRG)

Pattern of >5% moves over the cycle, ``-sourced:

• Dec 2022 — Vivint announcement: −15% on the day, ~−27% peak-to-trough. The market hated the smart-home pivot; it read as empire-building away from the core.
• Jun–Nov 2023 — Elliott activist campaign + board/CEO change: rally from ~$31 → ~$38. Activism = catalyst; the market rewarded the prospect of a refocus + capital return.
• 2024–2025 — the data-center re-rating: ~$38 → ~$150+. The dominant move of the cycle. Driven by (a) ERCOT large-load queue exploding past 230 GW, (b) NRG's repositioning as a power supplier to AI demand, (c) the GE Vernova deal (Feb 2025) and (d) the LS Power deal (announced May 2025). NRG was among 2025's best-performing dividend/utility-adjacent names.
• May 2025 — LS Power deal announced: positive re-rate (transformational scale + accretion framing).
• Jan 7 2026 — CEO succession (Coben→Gaudette) announced: orderly/planned — muted reaction.
• Mar 2026 — $2.3B upsized equity offering (LS Power exits part of stake): typical equity-overhang pressure; absorbed.
• May 5 2026 — Q1 print: shares ROSE on revenue beat despite adjusted-EPS miss.

What the tape reveals: NRG trades on (1) the data-center/ERCOT-scarcity narrative, (2) M&A and capital-return news, and (3) activist/governance events — far more than on the quarterly weather-driven earnings line. Macro power-price and ERCOT-summer-scarcity headlines move it; a single quarterly miss does not. This is a narrative + capital-allocation stock right now.

Verdict (L8): The market reacts to the data-center story and the buyback, not the print. That's bullish while the narrative holds — and a defined risk the day the narrative (AI-capex, ERCOT scarcity) cracks.

---

Phase C — Judge people & books

Lens 9 · Management

The governance arc is the single most important context for this name.

1. Track record — the activist turnaround (quantified). NRG is a textbook Elliott Management value-creation case. Dec 2022: prior CEO Mauricio Gutierrez announced the Vivint deal ($5.2B, closed Mar 2023); the stock fell 15% on the day. Jun 2023: Elliott ($1B, >13% economic interest) publicly called Vivint "the single worst deal in the power and utilities sector in the past decade," demanded a new CEO + board refresh. Nov 2023: Gutierrez out; Elliott seated 4 new directors; Lawrence (Larry) Coben (board Chair since 2017) became interim, then permanent CEO (Jul 2024). Result: strategy refocused on capital return + balance-sheet repair + the data-center/LSP pivot, and the stock ran ~$31 → ~$150. Whatever one thinks of Vivint, the post-activist team delivered the re-rating.

2. Tenure & skin in the game — and a fresh CEO handoff. Coben steps down as CEO effective April 30 2026; Robert J. Gaudette — EVP and President of NRG Business & Wholesale Operations (i.e., the executive who ran the commercial/large-load engine that is now the growth story) — becomes CEO; Antonio Carrillo (Lead Independent Director) becomes Chair. Insider ownership is incorporated by reference into the proxy and not disclosed in the 10-K body — n/a in-filing; not separately sourced. The handoff is planned and orderly (a positive), but it means the entire LSP-integration + data-center-conversion phase executes under a CEO in his first months in the seat (a risk).

3. Capital-allocation history — disciplined and shareholder-first.

• Dividend: raised 8% to $1.76/sh in 2025, then 8% again to $1.90/sh for 2026; long-term target 7–9% annual dividend growth.
• Buybacks: $3.7B program (since 2023) actively executing; additional $3.0B authorized Oct 16 2025 (through 2028). Committed to $1B/yr buybacks until leverage <3.0x, then revert to an 80/20 capital-allocation framework. Total ~$6.9B of capital-return commitment.
• M&A: the LSP deal at 7.5x 2026 EV/EBITDA / ~50% of new-build replacement cost is, on its face, a disciplined price for scarce dispatchable capacity — the anti-Vivint, structurally.
• Leverage target raised to <3.0x Net Debt/Adj EBITDA (from 2.50–2.75x), to be hit 24–36 months post-close, while protecting investment-grade metrics.

1. Red flags. (a) Vivint remains the strategic-credibility scar — $3.5B of goodwill and $810M/yr of D&A that the activist case wanted gone; it is still on the books. (b) The 10.25% Series A Preferred is expensive capital that flatters common-equity optics while ranking ahead of it. (c) Leverage is rising into a doubled, partly-merchant gas fleet — the balance-sheet repair narrative is now in tension with the de-levering-after-LSP reality. (d) CEO transition mid-integration.

2. Founder vs. professional manager. Purely professional/activist-installed management — Coben (academic/industry chair, archetype: disciplined capital allocator) → Gaudette (operator who built the commercial book). For this stage — integrate 13 GW, convert a 36 GW pipeline into contracts — an operator-CEO who came up through wholesale/large-load is arguably the right archetype. No founder; no founder-control risk; the discipline is externally enforced (Elliott-seated board).

Verdict (L9): A credible, shareholder-aligned, activist-disciplined team with a real turnaround on the scoreboard — handing the keys, mid-transformation, to the operator who built the growth engine. Net positive, with the asterisk that the riskiest execution phase (LSP integration + large-load conversion) runs under a first-year CEO.

Lens 10 · Forensic Red Flags

Acting as a forensic equity analyst. Where could the accounting bite?

• Mark-to-market / derivative volatility (highest-attention item). NRG runs a large commodity hedge book under ASC 815; unrealized MTM swings dominate GAAP earnings quarter to quarter — Q1-2026 GAAP EPS $0.52 vs $3.61 is largely MTM, not operations. The risk is not fraud — it is that adjusted (non-GAAP) numbers do the heavy lifting, and the gap between GAAP and adjusted is wide and growing with the larger book. Scrutinize the adjustments each quarter.
• Goodwill & intangibles — Vivint impairment watch. $5,017M total goodwill, of which $3,523M sits in Vivint; Vivint carries $810M/yr of D&A and reported only $53M segment operating income. A demand or churn shock at Vivint, or a higher discount rate, is a live impairment trigger. The activist thesis literally argued this asset was overpaid — the carrying value is the place to watch.
• Capitalized contract costs / customer-acquisition accounting. Amortization of capitalized contract costs rose to $510M (FY25) from $332M (FY24); in a retail + smart-home model, the timing of capitalizing vs. expensing CAC flatters near-term margins. Watch the capitalized-cost balance vs. customer additions.
• Provision for credit losses $272M (FY25) — retail energy carries real bad-debt exposure (mass-market customers, weather-spike bills); 67% of counterparty exposure is investment-grade, 33% non-IG/non-rated.
• Leverage & off-the-headline obligations. Gross debt + finance leases $16,600M (with $6,038M of scheduled interest payments ahead), plus the $650M 10.25% preferred ranking ahead of common, plus pension minimums (~$96M/5yr). Acquisition accounting for LSP (purchase-price allocation, new goodwill) lands in 2026 financials — a future PPA to scrutinize.
• Cash flow vs. earnings divergence. CFO fell to $1,913M (FY25) even as the company guides to big EBITDA growth — partly working-capital/collateral on the bigger hedge book, partly pre-funding. Watch FCF-before-growth conversion against the $2.8–3.3B 2026 guide.

Regulatory findings (required sub-section).

• SEC Litigation Releases / AAERs: None. Verified via SEC EDGAR EFTS (LR + AAER) for "NRG Energy," 2021-06-29 → 2026-06-29 — 0 findings.
• 10-K Item 3 (Legal Proceedings): NRG points to Note 22 (Commitments & Contingencies) and Note 23 (Regulatory Matters) for material proceedings — i.e., no standalone material litigation called out in Item 3 itself; the company discloses ordinary-course commitments/contingencies and ERCOT/PUCT regulatory matters in the notes. (Standard for a multi-state competitive power operator — ERCOT market-design / Winter Storm Uri-legacy / regulatory-rate matters are the live category, not securities fraud.)
• Non-SEC enforcement (web): No material FTC/DOJ/FDA/CFPB enforcement action against NRG Energy surfaced in search. The relevant regulatory exposure is state utility-commission / ERCOT market-design risk (PUCT, ERCOT rules), which is operational-regulatory, not enforcement.
• Conclusion: No material regulatory or legal enforcement findings — verified via SEC EDGAR EFTS (LR, AAER), web search, and 10-K Item 3 as of 2026-06-29. The forensic risk here is accounting-optics (GAAP-vs-adjusted gap + Vivint goodwill + leverage), not fraud or enforcement.

Verdict (L10): Clean on enforcement; the real diligence is (1) the widening GAAP-vs-adjusted gap on a bigger hedge book, (2) Vivint goodwill impairment risk, and (3) rising leverage into the LSP integration. All three are disclosed and quantifiable, not hidden — but they are the levers a skeptic should press every quarter.

---

Phase D — Project & stress-test

Lens 11 · Forward Projection (FY2026 → FY2028 adjusted EPS)

Built bottom-up from the latest actuals + management's own initiated guidance. Anchor: management FY2026 adjusted EPS guide $7.90–9.90 (midpoint ~$8.90), FY2026 Adj EBITDA $5.325–5.825B (midpoint ~$5.575B), FCFbG $2.8–3.3B. Management's structural commitment: ≥14% annual adjusted-EPS & FCF/share growth over 5 years, excluding large-load upside.

Input lines (base):

• Revenue/EBITDA driver: FY2026 EBITDA ~$5.575B (LSP full-year from 2027 adds ~1 more month + ramp). Owned-generation EBITDA now a larger, more contracted share.
• Interest: rising — FY25 $741M + LSP debt (~$3.2B assumed + new issuance); a real drag on EPS conversion.
• Preferred: $650M × 10.25% ≈ $67M/yr of preferred dividends ahead of common.
• Share count: ~208M and falling via $1B/yr buyback (~6–7M shares/yr at ~$150), partly offset by the LSP stock issuance already in the base.
• Tax: ~mid-20s% effective + CAMT exposure (15% corporate AMT).

Sensitivity: the swing factor is large-load contract conversion (management excludes it from the ≥14%, so it is optionality, not base) and ERCOT power-price/scarcity. A 20–30% disappointment in growth (Lens 13) pushes you to the bear ~$7.90–8.7 path and the multiple compresses simultaneously — the classic narrative-stock double-hit.

No forecast.ts create logged — this is an unattended --watchlist run (per SKILL, skip the Brier commit in the loop). Base case to log on a future attended pass: "NRG FY2026 non-GAAP EPS ≥ $8.90, resolves 2026-12-31."

Verdict (L11): Base ~$8.90 FY2026 → ~$11.6 FY2028 on management's ≥14% path. At ~$150, that's ~17x FY2026 / ~13x FY2028 adjusted EPS — undemanding if the per-share growth machine (EBITDA + buyback) delivers and large-load is upside on top.

Lens 12 · Bull vs Bear

Institutional, adversarial.

Bull case. NRG is the cheapest large-cap call option on U.S. data-center power demand. The thesis: (1) ERCOT and PJM are structurally short dispatchable capacity into an AI-driven demand surge (ERCOT large-load queue >230 GW, ~77% data centers); (2) NRG owns the scarce asset — ~25.8 GW of dispatchable gas post-LSP, bought at 7.5x EBITDA / ~50% of replacement cost (you literally cannot build it cheaper), plus reserved GE Vernova turbine slots (5.4 GW) that competitors can't replicate this cycle; (3) the integrated retail hedge turns merchant volatility into stable cash flow; (4) a >36 GW large-load pipeline (~10 GW already contracted) converts that capacity into long-duration contracted cash flow — upside management excludes from its ≥14%/yr per-share growth guide; (5) disciplined, activist-policed capital allocation — $1B/yr buybacks, 7–9% dividend growth, IG balance sheet. On forward EBITDA (~8.4x) it trades at a 3–5 turn discount to CEG/VST for a faster per-share growth rate. Re-rating + compounding = the bull's double.

Bear case (risks that could permanently impair):

1. It's gas, not nuclear — no scarcity premium, and carbon/policy/merchant exposure. The premium multiple in this group accrues to zero-carbon baseload (Constellation). A gas fleet faces carbon-policy risk, fuel-price pass-through volatility, and the possibility that hyperscalers prefer nuclear/renewables-plus-storage for their clean-power commitments — capping NRG's contracted-demand TAM.
2. Leverage into a doubled fleet. Gross debt ~$16.6B pre-LSP balloons with ~$3.2B assumed + new issuance; the leverage target had to be raised to <3.0x. A power-price downturn or integration miss with this balance sheet is a far worse outcome than for a low-levered peer — and it would throttle the buyback that is currently supporting the stock.
3. Expectations have decoupled from earnings. The stock ran ~5x off the data-center narrative while the latest print missed adjusted EPS by 17%. If the AI-capex cycle cools, ERCOT scarcity eases (230 GW of requests is not 230 GW of built load), or large-load contracts convert slower/at thinner terms than hoped, both EPS and the multiple fall together.

Pre-mortem (18 months out, thesis broke — what happened?): Most likely path: large-load contracts under-delivered — the 36 GW "pipeline" proved to be speculative interconnection requests, hyperscalers signed with nuclear/behind-the-meter alternatives or stalled on their own capex, and NRG's contracted GW grew far slower than the narrative implied. Simultaneously a mild-weather/soft-power-price year compressed merchant margins on the bigger fleet, leverage stayed elevated, the buyback slowed, and the multiple re-rated from ~17x to ~12x — a 30–40% drawdown despite "fine" operations. Secondary path: a Vivint goodwill impairment + an integration stumble dents credibility under a first-year CEO.

Are multiples too high? No — forward EBITDA ~8.4x is the cheapest in the basket. The risk is not the multiple; it's the denominator (will the EBITDA/EPS growth and large-load optionality actually show up).

Contrarian view (what the market refuses to see): The bears fixate on "gas, not nuclear" as a permanent discount. But for data centers that need power NOW, dispatchable gas + an existing interconnection + reserved turbines beats a nuclear plant that can't add a megawatt for a decade. NRG's "inferior" gas fleet may be the only asset class that can actually serve near-term AI load — which would mean the scarcity premium is mispriced toward gas, not away from it, for this specific demand wave. If even a third of the 36 GW contracts at infrastructure-grade terms, the ≥14% guide (which excludes it) is a floor, not a target.

Lens 13 · Devil's Advocate (short-seller)

Dismantling the bull case.

• The "36 GW pipeline" is a queue, not a contract book. ERCOT's 230 GW of large-load requests is notoriously speculative — developers file at multiple sites, most never build. NRG's 36 GW "pipeline by 2033" and even the ~10 GW "contracted under electric service agreements" deserve hard scrutiny: contracted at what price, what duration, what take-or-pay floor, what counterparty credit? If these are interruptible or thin-margin, the entire growth-optionality narrative is air.
• Revenue/earnings concentration is shifting toward a few mega-counterparties — the opposite of today's diversification. Today NRG's prized feature is "no customer >10% of revenue." The data-center strategy deliberately concentrates future cash flow into a handful of hyperscaler counterparties whose own AI-capex plans could reverse. NRG is trading granular retail diversification for lumpy tech-cycle exposure.
• The moat is weaker than bulls think. Vistra (50 GW, gas+nuclear+retail) and Constellation (nuclear) are larger and better-capitalized competitors for the same data-center load; the integrated-retail hedge is real but replicable (Vistra has it too). NRG's edge is timing (early turbine slots, cheap LSP) — timing edges decay.
• Worst capital-allocation history in the group is recent. This is the company that did Vivint ("worst deal in the sector in a decade") and Direct Energy (which an activist argued raised the cost structure). The current discipline is externally imposed by Elliott — if activist attention drifts, does the old empire-building instinct return? And $3.5B of Vivint goodwill still sits on the books unresolved.
• Leverage + 10.25% preferred = a financing stack that screams "stretched." Paying 10.25% for perpetual preferred to help fund LSP is expensive capital — a tell that the balance sheet was pushed. In a downturn, the preferred and the debt eat the common's cash flow and freeze the buyback.
• What must hold for $150: (1) ERCOT/PJM scarcity persists; (2) AI-capex cycle keeps hyperscalers signing long-duration power deals; (3) LSP integrates without a margin or impairment surprise; (4) leverage de-rates to <3.0x on schedule so the buyback continues; (5) a first-year CEO executes the most complex phase in NRG's history. Break any two and the stock is a 30%+ drawdown.
• Growth disappoints by 20–30%: EPS slides to the ~$7.90–8.3 bear path and the multiple compresses from ~17x toward ~12x → a double de-rate, ~35–40% downside.
• Single permanent-impairment scenario (most plausible): A broad AI-capex / data-center demand pause (hyperscaler digestion, model-efficiency gains cutting power intensity, or financing tightening) that simultaneously (a) stalls large-load contracting and (b) softens merchant power prices on the now-doubled fleet — with elevated leverage amplifying the equity hit. Plausibility: moderate — not a base case, but the defining risk and entirely outside NRG's control.

Lens 14 · Management Questions (ordered by information value)

1. Of the >36 GW large-load pipeline and ~10 GW "contracted under electric service agreements," how many GW are under binding, long-duration take-or-pay (or capacity-payment) contracts today — and what is the weighted-average term, floor price, and counterparty credit quality?
2. What is the realistic path and timeline to <3.0x Net Debt/Adjusted EBITDA, and under what power-price/integration scenario would the $1B/yr buyback be paused?
3. Walk us through LSP integration to date — synergy capture vs. the deal model, any asset/performance surprises, and the largest remaining execution risk in 2026–27.
4. The GAAP-to-adjusted EPS gap is now very wide (Q1-26 $0.52 vs $1.48 adjusted). As the hedge book doubles, what is the right way for investors to think about cash earnings through the MTM noise?
5. With the strategy concentrating cash flow into a few hyperscaler counterparties, how do you protect against losing the "no customer >10% of revenue" diversification that has been a core strength?
6. CEO transition: Mr. Gaudette, you built the commercial/large-load book — what changes (and what stays) versus the Coben strategy, and what is your single highest priority for the first 12 months?
7. On the 5.4 GW GE Vernova / Kiewit build — confirmed turbine delivery slots, all-in $/kW build cost, expected returns, and how much is de-risked by signed offtake vs. merchant?
8. What is the Vivint segment's standalone cash contribution and goodwill-impairment sensitivity to churn/discount-rate, and is divestiture ever on the table?
9. How exposed is the gas fleet to carbon policy / EPA rules, and do hyperscaler clean-power commitments threaten your addressable share of data-center demand vs. nuclear/renewables?
10. The $650M 10.25% Series A Preferred is expensive — what is the refinancing/redemption plan, and why was that the right instrument for the LSP stack?
11. How do you frame ERCOT market-design / PUCT regulatory risk (capacity mechanisms, large-load tariffs, performance credits) to the durability of your Texas economics?
12. Replacement-cost framing: you bought LSP at ~50% of new-build cost — does that imply organic new-build (GEV) is value-destructive at current returns, or are the use-cases different?
13. What fuel/transport and reliability risks does the doubled gas+dual-fuel fleet carry (another Winter Storm Uri/Fern), and how is that hedged after the East cost spike this quarter?
14. How should investors weight the ≥14% per-share growth guide given it excludes large-load upside — what's the downside case to that number?
15. What is the long-term capital-allocation hierarchy once <3.0x is reached (80/20 reinvest/return) — organic new-build vs. M&A vs. buyback vs. dividend?

---