NRG Energy reported third-quarter results on Monday, November 9, and the numbers alone would have been a strong quarter: adjusted earnings of $4.68 a share against a consensus estimate of about $3.44, revenue of $9.47 billion against $7.64 billion a year earlier, and adjusted EBITDA of $2.01 billion against $1.21 billion. Adjusted net income rose to $983 million from $537 million, GAAP net income more than tripled to $561 million, and free cash flow before growth investments came in at $1.24 billion. Shares closed the day up about 10%.
But the quarter’s real story arrived in the same announcement. NRG agreed to sell six gas-fired generation assets totaling just under 4.9 gigawatts across New York, Texas, and New England to LS Power for $4 billion in cash, at an implied multiple of 8.0 times the assets’ expected 2027 adjusted EBITDA. The deal is expected to close in the second quarter of 2027, subject to antitrust clearance, Federal Energy Regulatory Commission approval, and approval from the New York State Public Service Commission. Chief executive Robert Gaudette said LS Power approached NRG with an unsolicited offer the company decided was too attractive to refuse.
Start, as this publication’s earnings-report checklist prescribes, with the two EPS numbers and what sits inside them. The $4.68 adjusted figure compares with $2.78 a year earlier, a gain of about 68%, and the beat against consensus was roughly 36% — wide by any standard, and wider than the revenue beat, which suggests the quarter was won on margins rather than volume. The $409 million year-over-year gain in GAAP net income came from three sources: contributions from the LS Power portfolio, higher realized capacity prices in the East, and stronger margins in Texas, helped by favorable weather.
The segment split shows where the money was made. Texas adjusted EBITDA rose to $1.02 billion from $807 million a year earlier. The East did the heavier lifting: $614 million against $107 million a year ago, a nearly sixfold increase driven by the capacity prices that have transformed Eastern power markets. The LS Power portfolio sits inside both numbers — it is the engine of the quarter and, in part, the asset being sold.
The sale, in numbers
The six plants are Ravenswood in New York, Ocean State Power and Wallingford in New England, and Jack County, Johnson County, and R.W. Miller in Texas. At 8.0 times expected 2027 adjusted EBITDA, the $4 billion price implies roughly $500 million of annual EBITDA leaving with the plants — the number against which the buybacks, the debt reduction, and the retained cash all have to be measured.
The proceeds are split three ways: $2 billion for debt reduction upon closing, $1 billion for incremental share repurchases, and $1 billion retained as cash for growth or further deleveraging. The incremental buyback lifts total planned repurchases across 2026 and 2027 to $3 billion, with the additional $1 billion to begin this year — ahead of closing, funded from the balance sheet and operating cash flow before the sale cash arrives. The base 2027 plan already called for $1 billion of debt reduction and $1 billion of buybacks, so with the sale the company is targeting $3 billion of debt reduction in 2027 in total. The board also plans an 8% dividend increase for 2027, bringing the annualized payout to about $2.05 a share.
The debt math is the reason the split looks the way it does. NRG’s January acquisition of the LS Power portfolio — about 13 gigawatts across 18 gas and dual-fuel facilities plus CPower, for 24.25 million shares and $6.4 billion in cash — was financed in part with $4.4 billion of notes and a $2.5 billion revolving credit facility. Total debt and finance leases stood at nearly $22.9 billion at the end of September, and current maturities of long-term debt had risen to nearly $1.08 billion from just $31 million at the end of 2025.
What NRG is keeping
The more revealing part of the announcement is what is not for sale. NRG retains its entire PJM generation fleet and the CPower demand-response business acquired in January. That retention is the strategy in one line: keep the assets in the market where capacity prices have just repriced higher, and sell the assets in markets where the price on offer exceeds what NRG believes it can earn by holding them.
The logic is straightforward once the capacity-price map is in view. PJM’s capacity auctions have delivered prices that make existing gas plants extraordinarily profitable, and NRG’s PJM fleet is the direct beneficiary. The plants being sold sit in NYISO, ISO New England, and ERCOT, where the earnings stream is thinner, more volatile, or both. LS Power knows these assets intimately — it sold most of them to NRG ten months ago — and values the cash flows differently; NRG values the cash more. In January NRG bought roughly 13 gigawatts from LS Power; in November it is selling 4.9 gigawatts back. That is capital recycling, not a reversal: NRG keeps about 8 gigawatts of the acquired fleet, plus CPower, and converts the rest into $4 billion at a premium multiple.
The guidance math
The company reaffirmed its 2026 guidance — adjusted EPS of $7.90 to $9.90 and adjusted EBITDA of $5.33 billion to $5.83 billion — while noting it expects to finish 2026 slightly below the midpoint of the EBITDA range. The 2027 guidance, initiated alongside the results, is explicitly stated excluding the asset sale: adjusted EBITDA of $5.5 billion to $6.0 billion, a midpoint of $5.75 billion; adjusted EPS of $8.05 to $10.05, a midpoint of $9.05; and free cash flow before growth of $3.15 billion at the midpoint. The exclusion matters. Post-sale guidance has not yet been provided, and the $9.05 midpoint belongs to a company that still owns the six plants. When the sale closes, roughly $500 million of EBITDA on the deal’s own multiple leaves with them, and the buybacks and debt reduction will have to offset part of that loss. One illustration put that offset at roughly $177 million to $207 million of annual after-tax earnings — partial, not full.
The honest way to read the 2027 guidance, then, is as a baseline for the retained business rather than a forecast of reported results. The $5.75 billion EBITDA midpoint says the retained fleet — PJM capacity, Texas optionality, retail, and Smart Home, where adjusted EBITDA rose 14% in the quarter — can support earnings at that level before the sale’s effects. Whether the sale is accretive, dilutive, or neutral per share will depend on the repurchase price, the interest saved, and the earnings that depart, none of which is fully knowable until post-sale guidance arrives.
The Texas data-center project
The quarter’s other strategic thread is the 1.2-gigawatt Texas data-center project under NRG’s “Bring Your Own Power” initiative, announced in August. Gaudette was emphatic on the call: the project is going to happen, terms have been agreed with the unnamed hyperscaler customer, and approved agreements and an announcement are expected by year-end. The political environment since the election, he said, has reinforced the strategy of pairing new generation directly with new load rather than drawing on the existing grid.
The project is a combined-cycle gas plant expected to cost about $3.2 billion and begin operating in late 2029, backed by a contract of at least 15 years with an investment-grade guarantee, and expected to generate roughly $500 million in annual adjusted EBITDA. It received conditional inclusion in ERCOT’s Batch Zero large-load interconnection process in September. NRG is also in talks with the same customer about expanding the arrangement to as much as 2.4 gigawatts at the same location, and demand for its turbines now exceeds the allocations under its development agreement with GE Vernova and Kiewit. Partnership financing is under discussion; on funding, management was unequivocal: not equity issuance. That connects directly to the asset sale: the $1 billion of retained proceeds is the growth capital most likely to fund this pipeline. A merchant generator selling 4.9 gigawatts of existing gas plants while building 1.2 gigawatts of new gas generation for a single customer is not shrinking for its own sake — it is swapping merchant exposure, earnings that rise and fall with capacity auctions and Texas weather, for contracted exposure under a 15-year agreement with an investment-grade guarantee. The sale finances the swap.
The context is less settled than the CEO’s confidence suggests. Texas paused new data-center grid hookups in August while ERCOT reviews more than 474 gigawatts of connection requests, about 90% of them from data centers, with a report expected in early December. NRG’s project is designed to sit outside that queue — the generation is built for the customer, not drawn from the grid — but conditional Batch Zero inclusion is not final approval, and late 2029 is three years of construction, permitting, and financing away.
Discipline at the top of the cycle
Step back, and the quarter reads as a statement about where merchant power value sits now. Capacity prices in the East are the windfall, and NRG’s response is not to double down on scale but to harvest it: sell assets at 8.0 times forward EBITDA, retire debt taken on to buy them, return cash to shareholders, and keep the balance sheet flexible enough to fund contracted growth. That is discipline at the top of the cycle — the opposite of what merchant generators have done at previous tops, when windfalls were spent acquiring more of the same assets at higher prices.
The discipline has a cost, and the honest reading names it. Roughly $500 million of annual EBITDA is leaving with the six plants, and the buybacks and debt reduction offset only part of that on an earnings basis. The retained PJM fleet has to keep earning at capacity prices that reset at auction — prices that rose sharply can fall, as they have in every market that has ever repriced. The Texas business has to keep producing the margins this quarter delivered, in a market where weather is a genuine variable and the data-center pipeline is still a set of agreements expected by year-end rather than signed contracts in hand. And the $1 billion of retained cash is optionality only until deployed; held too long, it is drag.
What this quarter does not resolve
Three things this quarter does not settle, and an honest reading should name them. First, the post-sale earnings picture. Until NRG issues guidance that includes the divestiture, the market is pricing a company whose 2027 midpoint of $9.05 excludes the sale’s effects, and the size of the gap depends on repurchase prices, interest rates, and the plants’ actual 2027 earnings — none yet fixed.
Second, the capacity-price question. The East segment’s jump from $107 million to $614 million of quarterly EBITDA is the quarter’s most dramatic number, and the least durable-looking if capacity prices normalize. NRG is selling into that strength, which is the right instinct, but it is also keeping its entire PJM fleet exposed to it. The retained fleet is a bet that the repricing holds; the sale is a hedge against the possibility that it does not.
Third, the data-center execution question. The 1.2-gigawatt project has agreed terms, a conditional interconnection slot, and a customer in expansion talks — but no signed agreements yet, no financing partner yet, and a late-2029 start date. The $500 million of expected annual EBITDA is a management estimate for a plant that does not yet exist.
What to watch
Four markers, in order of consequence. First, the year-end announcement on the Texas data-center project: signed agreements, a named financing partner, and any expansion beyond 1.2 gigawatts would convert the year’s most confident management statements into contracts. Second, the regulatory path for the sale — antitrust clearance, FERC, and the New York commission — which determines whether the second-quarter 2027 closing holds. Third, post-sale 2027 guidance, which will show how much of the roughly $500 million of departing EBITDA the buybacks and debt reduction actually replace. Fourth, the next PJM capacity auction results and the East segment’s fourth-quarter print, which will say whether the capacity windfall is a new plateau or a peak.
NRG’s quarter is best read as a progress report on a thesis, not a verdict on it. The earnings arrived, the beat was wide, the capacity windfall is real, and the company used it to shrink the fleet, cut the debt taken on to build it, and return $3 billion to shareholders across two years — while keeping the PJM assets where the repricing pays best and the Texas optionality where the data-center demand is forming. What has not yet arrived is proof that the smaller, more focused company earns as much per share as the larger one did, and that proof comes in signed data-center agreements, a closed sale, and post-sale guidance that closes the arithmetic. The sale is disciplined. Whether it is also accretive is a question for 2027.
None of this is investment advice, and nothing in this analysis recommends buying or selling any security. It is a reading of one quarter’s evidence — a wide beat, a $4 billion sale at 8.0 times forward EBITDA, $3 billion of planned buybacks, and a data-center project still awaiting its signed agreements — with the windfall separated from the durable earnings before anyone calls the quarter a trend.