Energy / Competitive power & retail electricity · Deep dive

NRG Energy

The twice-restructured Texas power-and-retail giant that Elliott has raided twice, that bought a door-to-door alarm company for $2.8B and got its CEO fired over it — and that just doubled its gas fleet with a $12B LS Power deal to chase an AI power-demand supercycle it has yet to sign a single definitive hyperscaler contract for.

well positioned

NRG enters a genuinely supply-short power market owning 25 GW of dispatchable gas bought below replacement cost, the largest competitive retail book in Texas to sell it through, and $5.5B of guided 2026 EBITDA — a position that compounds even if the headline data-center deals keep slipping, though the leverage leaves little room for a stumble.

My take ★★ Very interesting ★ Interesting − Maybe ✗ Not interesting

Snapshot

NRG Energy is one of the two big competitive power-and-retail platforms in America: roughly 25 GW of generation after the LS Power deal closed on January 30, 2026, about 8 million retail electricity, natural gas, and smart-home customers under the Reliant, Direct Energy, Green Mountain, and Vivint brands, $30.7B of 2025 revenue, and $4.09B of 2025 adjusted EBITDA (reported February 24, 2026). It matters now because it is the purest listed bet on the thesis that AI data centers will make dispatchable gas power scarce and expensive: NRG was the best-performing stock in the S&P 500 in the first half of 2025 (up ~78%), doubled its fleet with a $12B acquisition, and formed a 5.4 GW new-build venture with GE Vernova and Kiewit — while carrying a debt load, an activist-scarred governance history, and a data-center pipeline that is still mostly letters of intent.

Founding story

There is no founder; there is a survivor. NRG was created in 1989 as the unregulated generation arm of Northern States Power, a Minneapolis utility, and rode the 1990s deregulation wave into a global merchant-plant buying spree. A May 2000 carve-out IPO raised $423M; the post-Enron power crash then buried the company under roughly $9.2B of obligations, and NRG filed Chapter 11 in May 2003 — one of the era’s defining energy bankruptcies. It emerged that December, independent of Xcel, under David Crane, who spent twelve years building a coal-heavy cash machine while evangelizing a green consumer-energy future the board eventually fired him over (December 2015). Trader-turned-COO Mauricio Gutierrez took the seat, executed Elliott’s 2017 shrink-to-grow plan, then re-expanded into retail and the home with Direct Energy and Vivint — and was removed on November 20, 2023 after Elliott’s second campaign. The current CEO, Larry Coben, is the board’s institutional memory made executive: a director since December 2003, a Yale/Harvard Law-trained former IPP entrepreneur (Tremisis Energy) with a working second career as a University of Pennsylvania archaeologist. He was named interim in November 2023 and permanent on August 1, 2024, at 65.

How it works

NRG’s machine has two ends that hedge each other. The generation end burns fuel: post-LS Power, ~25 GW of mostly gas and dual-fuel plants across Texas, the Northeast, and seven other states, selling energy and capacity into ERCOT, PJM, and NYISO wholesale markets. The retail end sells that power to homes and businesses at fixed or indexed rates through Reliant (the dominant Houston brand), Direct Energy, and Green Mountain. The integration is the point: when wholesale prices spike, generation profits offset retail supply-cost pain, and vice versa — a natural hedge that turns two volatile businesses into one relatively stable EBITDA stream (Texas segment: $1,877M in 2025, up $295M year over year). The physical tail risk is real: Winter Storm Uri in February 2021 produced a $967M quarterly loss when plants failed while retail obligations stayed on, mitigated to a final ~$380M full-year hit. The newer layers monetize the customer beyond the meter. Vivint’s ~20,000-person sales machine installs cameras, doorbells, and smart thermostats on five-plus-year subscription contracts ($1,092M of 2025 EBITDA). Those thermostats feed a residential virtual power plant — 150 MW enrolled in Texas in 2025, targeting 650 MW by 2030 — that NRG can dispatch as negative demand when ERCOT tightens; CPower, acquired with LS Power, does the same for commercial and industrial load at ~6 GW scale. The data-center play is the third act: sell existing plant sites and new gas capacity to hyperscalers at premium, contracted rates.

Product and business overview

Four reported segments. Texas — the crown jewel: integrated ERCOT generation plus Reliant/Green Mountain retail; $1,877M of 2025 adjusted EBITDA. East — retail and generation across PJM, New York, and New England; $981M in 2025, now the home of most acquired LS Power capacity. West/Other — $137M in 2025, shrunk by the Airtron sale (September 2024) and the Cottonwood lease expiry (May 2025). Vivint Smart Home — 1.9M+ subscribers at acquisition (2023), $1,092M of 2025 EBITDA on record adds and rising margin per customer. Layered across them: the GE Vernova/Kiewit venture (announced February 26, 2025) with turbine slots and EPC capacity reserved for 5.4 GW of new combined-cycle plants — 1.2 GW targeted in service by 2029, 1.2 GW by 2030, ~3 GW by 2032; letters of intent with data-center developers Menlo Equities and PowLan (400 MW initial, up to 6.5 GW potential); a LandBridge agreement (September 23, 2025) for a possible 1.1 GW plant serving a Permian Basin hyperscale campus; and a 415 MW Texas Energy Fund-backed peaker nearing completion in Houston (Utility Dive, 2026).

Business model and pricing

Revenue books three ways: retail energy sales at contracted or variable rates (the bulk of $30.7B in 2025), wholesale energy and capacity sales from the fleet, and subscription revenue from Vivint (roughly $2.1B in 2025, recognized monthly over long contracts). Real price points: Texas retail plans are commodity-plus — Reliant’s fixed-rate plans price competitively against dozens of ERCOT rivals, and billing disputes, not rates, drive most complaints; Vivint monitoring runs roughly $30-70 per month plus financed equipment. The 2026 guidance issued February 2, 2026 — adjusted EBITDA of $5,325-5,825M, adjusted net income of $1,685-2,115M — implies ~36% EBITDA growth at midpoint, almost entirely from eleven months of LS Power. Capital allocation is buyback-heavy: $1.3B repurchased in 2025 plus $344M of dividends, with $1.0B of buybacks and ~$400M of dividends planned for 2026 and the quarterly dividend raised 8% in January 2026 to $1.90 annualized. The balance sheet is the constraint: $4.9B of new notes in October 2025, total liquidity of $9.6B at year-end 2025, and a stated deleveraging path analysts consider the gating item on further returns.

Traction over time

Q1 2026 (reported May 6, 2026): adjusted EBITDA $1,080M, adjusted EPS $1.49, GAAP net income $125M — guidance reaffirmed, but the stock fell ~7.5% on the soft quarter. Customers total about 8 million across North America (company, May 2026). The share price tells the sentiment arc: +78% in the first half of 2025, a $189.96 high on February 25, 2026, then a ~30% slide by mid-2026 as data-center announcements underwhelmed. Q2 2026 reports August 4, 2026.

Market analysis

US power demand is growing for the first time in two decades, and NRG sits in the two hottest markets. The EIA projects retail electricity sales in ERCOT growing ~11% annually across 2025-26 and PJM ~4%, driven overwhelmingly by data centers; ERCOT’s own planning scenarios run from a base-case peak of ~98 GW in 2026 to interconnection-request-inflated scenarios approaching 280 GW by 2029 (ERCOT, April 2026) — numbers everyone agrees are inflated by duplicate requests, and nobody agrees by how much. The EIA warned data-center load could drive a 79% ERCOT wholesale price increase in 2027 (Utility Dive, 2026). Structural forces favoring NRG: gas turbines are sold out into the 2030s (hence the pre-reserved GE Vernova slots), new-build costs above $2,400/kW make NRG’s ~$920/kW LS Power purchase look cheap, and retail books grow with load. Forces against: if even a fraction of announced data centers slip, merchant gas EBITDA mean-reverts; Texas politics caps retail pricing freedom after Uri; and hyperscalers increasingly prefer nuclear (Constellation, Talen, Vistra) for carbon accounting, leaving gas-heavy NRG the less fashionable counterparty.

Competitive intel

The sidebar carries the set; the analytical read: NRG is the cheapest of the big four IPPs per unit of story. Constellation ($104B market cap) closed Calpine on January 7, 2026 and owns the nuclear fleet plus the only signed marquee hyperscaler deals; Calpine’s Texas gas fleet and retail arm now compete inside ERCOT with Constellation’s balance sheet behind them. Vistra ($56B, January 2026) is the direct doppelganger — integrated Texas generation-plus-retail — valued at roughly twice NRG partly for its nuclear and storage. Talen ($17B) is smaller but demonstrated the monetization NRG promises, contracting Susquehanna nuclear capacity to AWS. NRG’s edge is the retail machine (no rival matches Reliant’s Texas residential density) and demand response at both residential (Vivint VPP) and C&I (CPower) scale — an aggregated flexibility story competitors lack. Its weakness is that its data-center pipeline remains LOIs and site agreements while rivals announce definitive contracts; the August 2025 stock drop after underwhelming deal disclosures showed how impatient the market is.

History and evolution

1989: formed inside Northern States Power. May 2000: $423M carve-out IPO. May 2003: Chapter 11 (~$9.2B restructured); December 2003: emerges under Crane. 2009: buys Reliant’s Texas retail business; 2010: Green Mountain Energy. December 2012: $1.7B GenOn merger. December 2015: Crane forced out; Gutierrez in. February 2017: Elliott and Bluescape arrive; July 2017: Transformation Plan — renewables sold to GIP for $1.375B, GenOn’s 15.4 GW handed to creditors; NRG is the S&P 500’s best performer that year. January 5, 2021: Direct Energy closes ($3.625B). February 2021: Uri — $967M Q1 hit, ~$380M final. March 10, 2023: Vivint closes ($2.8B equity). May 2023: Elliott returns with a 13%+ position and a scathing letter. November 20, 2023: Gutierrez out; Coben interim. August 1, 2024: Coben permanent. February 26, 2025: GE Vernova/Kiewit 5.4 GW venture plus Menlo/PowLan LOIs. May 2025: $12B LS Power deal announced. September 23, 2025: LandBridge Permian data-center agreement. October 2025: $4.9B notes issued. January 30, 2026: LS Power closes — fleet doubles to ~25 GW. February 24, 2026: FY2025 results; stock peaks the next day. May 6, 2026: soft Q1; guidance reaffirmed; shares slide.

What people say

The case for. Sell-side sentiment is strongly bullish: over 80% of covering analysts positive, a Strong Buy consensus, and a median target near $200 against a mid-2026 price around $130 (Simply Wall St/TipRanks, mid-2026), implying ~50%+ upside. Bulls cite the LS Power fleet bought far below replacement cost just as ERCOT and PJM tighten, the natural generation-retail hedge that produced fourteen years of steady free cash flow, Vivint’s record 2025 customer adds and $1.09B EBITDA quietly vindicating the hated deal, and Coben’s capital discipline — he has publicly said NRG has “zero interest” in speculative, uncontracted new build (Utility Dive, 2025). Fortune’s 2025 profile framed NRG as the S&P’s biggest riser with the demand supercycle still early. Glassdoor employees rate the company 3.8/5 across ~963 reviews with 76% recommending it (2026), praising power-markets exposure and pay.

The complaints. Elliott’s 2023 letter remains the sharpest critique: Vivint was “the single worst deal in the power and utilities sector in the past decade,” and the governance lesson — that NRG’s board needed two activist interventions in six years — still colors the multiple. The consumer record is genuinely rough: Reliant logged 449 Texas PUC complaints in twelve months (electricrates.org, 2026), nearly four times the industry average in absolute terms, with billing disputes the dominant theme across Trustpilot and ConsumerAffairs reviews; Vivint paid the FTC $20M in April 2021 for misusing consumer credit reports (“white paging”), has faced enforcement actions from 16 state attorneys general over sales misconduct, and lost a ~$190M deceptive-practices verdict to CPI Security, affirmed on appeal in July 2025. Financial skeptics focus on leverage — the October 2025 notes plus assumed obligations leave, in Simply Wall St’s May 2026 framing, “very high leverage” that limits flexibility — on the LS Power share overhang, on a Q1 2026 miss that dropped the stock ~7.5%, and on the gap between data-center rhetoric and signed contracts: an August 2025 disclosure of underwhelming deal progress knocked the shares hard. Employees’ recurring gripes are layoff survivor workload and an abrupt shift to four-day office attendance (Glassdoor, 2026). And the Uri scar tissue is permanent: the model that hedges a normal year lost nearly $1B in a single quarter when the physical grid failed.

Outlook: well positioned or at risk?

Well-positioned. Strip out the AI theater and the position is unusually good: NRG owns 25 GW of dispatchable capacity in the two fastest-growing power markets in America, bought half of it at roughly $920/kW when new gas costs $2,400+/kW and turbines are sold out past 2030, and it owns the retail and demand-response channels — 8 million customers, Reliant’s Texas density, CPower’s 6 GW of C&I flexibility — that let it monetize scarcity twice. The 2026 guidance (~$5.6B EBITDA at midpoint, +36%) does not require a single hyperscaler contract to be hit; every data-center deal that does land — Menlo, PowLan, LandBridge, or the promised 2026 agreement on the GE Vernova build — is upside on top. The bear case is real but is a leverage-and-execution case, not a structural-decay case: a demand disappointment or another Uri-class grid failure would hit a levered balance sheet hard, LS Power’s stock will overhang the shares, and NRG has now twice needed Elliott to correct capital allocation, which argues for humility about the next big idea. But the asset NRG defends — flexible generation plus the customer relationship in a supply-short market — is appreciating, not eroding. Vistra trades at double the market cap for a similar machine; the gap is NRG’s opportunity, and closing even part of it only requires the company to keep doing what the guidance already assumes.

How a challenger would attack it

Attack the retail book, where the complaints already live. NRG’s natural hedge depends on Reliant’s Texas residential density, and that book is defended by brand inertia, not satisfaction: 449 Texas PUC complaints in twelve months — nearly four times the industry average — with billing disputes the dominant theme across Trustpilot and ConsumerAffairs. In a deregulated ERCOT market where switching is one form away, a digital-first retailer with transparent billing and an honest smart-thermostat VPP (paying customers for flexibility rather than dispatching them into NRG’s wholesale position) peels off exactly the customers whose margin funds the hedge. The Vivint channel is even softer: an FTC settlement, sixteen state AG actions and a $190M deceptive-practices verdict mean a challenger selling smart-home security without door-to-door sales inherits the demand and none of the litigation tail. On the wholesale side, the attack is carbon accounting: hyperscalers increasingly prefer nuclear counterparties, and Constellation, Talen and Vistra all have signed deals while NRG’s pipeline is LOIs — a competitor doesn’t need to outbuild 25 GW of gas, it needs to keep winning the definitive contracts so NRG’s data-center premium never materializes and the leverage ($4.9B of fresh notes, an LS Power share overhang) has to be serviced from merchant economics alone. The stock’s ~30% slide from its February 2026 peak shows how little deal slippage that thesis tolerates.

Same playbook, new buyer

Integrated generation-plus-retail-plus-flexibility is a template ERCOT proved and other deregulated markets haven’t absorbed. The nearest port is PJM and the Northeast retail markets, where NRG’s East segment ($981M EBITDA) is generation-heavy but nothing like Reliant’s Texas retail density exists under one roof — a regional player assembling gas capacity, a retail brand and C&I demand response there runs the Texas playbook before NRG, busy integrating LS Power and deleveraging, can. The more interesting shift is the buyer: NRG aims its aggregated flexibility (Vivint VPP at 150 MW heading to 650, CPower’s ~6 GW) at wholesale markets; sold instead to utilities and co-ops in regulated states as a resource-adequacy product, the same machinery monetizes without owning a single plant — and NRG won’t chase that channel because its economics require the retail relationship it refuses to share. Third, the mid-size data center: NRG, Constellation and Talen all court gigawatt-scale hyperscalers, leaving 20-100 MW enterprise and colocation loads — too small for a LandBridge-style campus deal — with no packaged power-plus-flexibility offer. A developer standardizing behind-the-meter gas-plus-battery for that tier serves demand the giants’ deal teams structurally ignore, on timelines the sold-out turbine market makes precious.

Sources and further reading

• NRG Energy Reports Full Year 2025 Financial Results — NRG Investor Relations, February 24, 2026
• NRG Energy Reports First Quarter 2026 Results — NRG Investor Relations, May 6, 2026
• NRG Energy CEO Gutierrez Leaves After Pact With Elliott — Bloomberg, November 20, 2023
• Activist investor Elliott seeks new NRG Energy CEO, revamped strategy — Utility Dive, May 2023
• NRG Buying LS Power Gas-Fired Plants in $12-Billion Deal — POWER Magazine, May 2025
• NRG aims for 2026 next step in 5.4-GW gas deal with GE Vernova, Kiewit — Utility Dive, 2026
• Vivint Will Pay $20 Million to Settle FTC Charges — Federal Trade Commission, April 2021
• Vivint Loses Appeal in $189M CPI Dispute — Security Info Watch, July 2025
• EIA: Retail electricity sales in PJM and ERCOT set to soar, driven by data centers — Data Center Dynamics, 2025
• NRG rides the power wave as the S&P 500’s biggest riser — Fortune, 2025