CEG: A $5 Quarter and a $3 Billion Cash Test

The 60-second read

Five things to know

1. The strong quarter raised the annual bar. Adjusted EPS reached $5.06 and the full-year midpoint rose to $12.65; a quarterly earnings burst is not a $20.24 annual run rate.
2. The contract book has become more substantial. The November deck shows 91 million annual MWh of long-term agreements in 2032, supporting about 45% of expected nuclear and geothermal generation.
3. Cash improved sharply, but allocation remains demanding. Computed Q3 OCF less total capex was $2.486B. Nine-month free cash flow was $1.518B against $3.001B of repurchases and $459M of dividends.
4. The uprate funding plan is conditional. The up-to-$5B DOE loan commitment is not cash received. The deck identifies roughly $3B of incremental growth investment during 2028–2032.
5. Our prior HOLD and $285 value remain. New contracting supports the old operating thesis, while unresolved project timing, merchant earnings and the cash cost keep a large premium out of our base case.

Three scenarios, one probability-weighted number

Valuation scenarios at the November 9 reference date; probabilities are analytical judgments

Signposts: what would change our mind

Only the annual EPS range and identified project/contract figures are issuer statements. All green/red thresholds, deadlines for our review and safety-price rules are our analytical choices; they are not company promises or confirmed earnings dates.

The price and the prior call

The saved November 9 quote is $273.95, up 9.60% from the preceding $249.95 close. It is an intraday observation, not a completed-session return. The saved market capitalization is approximately $97.1B. A price recorded during the release day can change before publication; this packet anchors its model to the dated observation instead of quietly changing the denominator between the video and the research page.

Our Q2 episode used HOLD, conviction three out of five, a $285 base value and a $268.03 reference price. The base value remains $285. The current quote is about 2.2% above that prior reference before dividends. That observation is not a measured investment recommendation return: it ignores trading timing, transaction costs and the fact that our earlier call was not an instruction to buy. The relevant track record is whether the operating tests and valuation logic still hold.

The latest quarter strengthens the argument for valuable existing generation and long-duration customer relationships. It also makes the timing of reinvestment more concrete. A higher annual EPS guide does not automatically warrant a higher present value if a larger share of future cash is committed to growth, buybacks or integration. Our unchanged rounded value is the result of a refreshed model, not an assumption that nothing changed.

Historical release-day close-to-close movements below are descriptive. The provider calendar dates and daily closes are saved, but we do not claim that earnings alone caused every move or that each release occurred at an identical time. Today's incomplete session is excluded from those historical bars. The operating analysis stands independently of whether a short-term share-price jump continues.

Source: Saved FMP market data, retrieved November 9. Ratios and valuation arithmetic are computed; company forecasts and our assumptions are labeled separately.

A strong quarter, a narrower guide

Revenue was $9.550B, compared with $6.570B a year earlier. GAAP net income attributable to common shareholders reached $1.835B, or $5.19 diluted EPS, compared with $930M and $2.97. Adjusted operating earnings were $1.792B, or $5.06 per share, versus $952M and $3.04. The release explicitly attributes adjusted improvement to favorable market and portfolio conditions and the addition of Calpine, partly offset by interest and nuclear production-tax-credit portfolio effects.

The share denominator matters. Average diluted shares rose to 354M from 313M, so growth in dollar earnings is not the same as growth per share. The acquisition changed both the asset base and ownership base. Calling the entire 45.4% revenue increase organic would be wrong; annualizing the $5.06 quarter would be another error because summer demand, capacity revenue, commercial performance and maintenance schedules are seasonal.

Management raised and narrowed annual adjusted EPS guidance to $12.50–$12.80 from $11.50–$12.50. The midpoint increased from $12.00 to $12.65, a computed 5.4%. The revised guide uses expected annual diluted shares of 356M. Subtracting the three reported quarterly adjusted EPS values from the annual range gives an approximate remaining-quarter indication of $2.15–$2.45. That is our arithmetic, not a separately issued Q4 guide, and annual versus quarterly share weighting limits exact additivity.

FMP's saved results row carries $3.90 of EPS expectations and $9.233B of revenue expectations. Against those figures, adjusted EPS exceeded the snapshot by 29.7% and revenue by 3.4%. The provider row was retrieved after the release; it is not an independently archived pre-announcement survey. The comparison is useful context, while the issuer's actual statements determine the numbers we analyze.

Source: Current earnings release, P&L and reconciliation. Ratios and valuation arithmetic are computed; company forecasts and our assumptions are labeled separately.

Operations and contracted generation

The operational evidence is strong. Nuclear generation including owned output from Salem and South Texas Project was 46,677 GWh, compared with 46,477 GWh. The capacity factor for the operated nuclear fleet, excluding those two plants, was 97.2%, compared with 96.8%. Planned refueling outages consumed 20 days versus 23; non-refueling outages were six days versus five. Those metrics have different scopes and should not be combined into one fabricated fleet-efficiency ratio.

The deck's 99.2% summer capacity factor covers June through August. It is not the September-quarter capacity factor. Both can be true at the same time. Reliable output is valuable when electricity prices and system demand are high, but output rose only slightly while earnings rose sharply. Portfolio scope, pricing and commercial execution explain why profit grew much faster than nuclear megawatt-hours.

Long-term agreements support 91 million annual MWh in 2032, roughly 45% of expected clean baseload generation. The deck describes 15-to-20-year contracts and a customer-credit mix concentrated in strong ratings. These arrangements can replace uncertain future merchant revenue with more predictable economics. They do not mean 45% of all company revenue is guaranteed, and the clean baseload denominator includes nuclear and geothermal generation rather than every gas plant and retail account.

We show operating families rather than inventing current segment net-income figures that the earnings attachment does not supply. The gas, oil and pumped-storage fleet's equivalent forced outage factor was 5.9%; renewable energy capture was 96.4%. A lower forced-outage factor is generally preferable, whereas a higher capture factor is generally preferable. Their units, direction and operating meaning differ. Comparing the two as if they were margins would mislead.

Source: Operating and contracted-generation presentation. Ratios and valuation arithmetic are computed; company forecasts and our assumptions are labeled separately.

Read the reconciliation in both directions

The adjusted number is slightly below GAAP this quarter. Starting from $1.835B of GAAP common earnings, the reconciliation removes $127M of unrealized fair-value gains and $108M of tax-related benefits. It adds decommissioning adjustments of $56M, acquired commodity-contract amortization of $39M, Calpine integration costs of $73M, pension and other postemployment costs of $25M, and legal/environmental adjustments of $9M. Plant retirement/divestiture and noncontrolling items reduce the bridge by another $10M.

The net adjustment is negative $43M, producing $1.792B of adjusted operating earnings. Rounded per-share figures can differ by a cent when summed; the dollar bridge is the arithmetic check. Some exclusions remove noncash volatility, while integration costs can require real cash. The label adjusted does not establish that every excluded expense is economically irrelevant, and GAAP is not automatically more conservative in a quarter with favorable marks or tax effects.

Operating income was $2.492B. Net interest expense of $217M and other expense of $44M reduced pretax income to $2.231B; tax expense of $405M and equity income of $9M resulted in the $1.835B bottom line. The effective tax rate computed from those current-quarter amounts was 18.2%. It differs from the marginal rates used in specific non-GAAP adjustments and from modeling tax rates in the deck. A single quarter's favorable tax outcome should not become a permanent valuation assumption.

The previous episode focused on a quarter in which GAAP and adjusted earnings moved in opposite directions. That old headline does not describe this print: both measures grew. The more useful question now is how much improved earnings can finance the next stage of growth after maintenance, financing and shareholder distributions. The accounting bridge supports that question rather than replacing it.

Source: GAAP / adjusted operating earnings reconciliation. Ratios and valuation arithmetic are computed; company forecasts and our assumptions are labeled separately.

Nine checks on earnings quality

Quarterly operating cash flow of $3.426B exceeded GAAP common earnings by a factor of 1.87. That is favorable evidence for this quarter, but it is not a sustainable conversion promise. Collections, collateral, tax payments and refueling timing can move cash between quarters. Year-to-date cash therefore matters alongside the headline summer result. The nine-month statements show $4.979B of OCF and $3.461B of total capex, leaving $1.518B by the simple subtraction used here.

Historical cash-flow classification is especially important for this company. The audited parent cash-flow statement shows negative OCF in 2023 and 2024 while deferred-purchase-price collections of $7.340B and $10.217B appear in investing activities. Mechanically presenting those negative OCF values as operating collapse would confuse classification with business economics. We show both the GAAP OCF-minus-capex series and a separately labeled OCF-plus-DPP-minus-capex diagnostic; neither is the company's FCF-before-growth measure.

Receivables of $4.537B equate to a simple end-balance proxy of 43.7 revenue days using the 92-day quarter. Inventory of $3.386B is about 32.6 revenue days on the same denominator. These are screening ratios, not average-balance working-capital cycles; inventory includes fuel and other operating assets. Calpine changes the balance-sheet scope, so a year-over-year movement cannot be attributed entirely to deteriorating collections.

The latest annual report records $385M of stock-based compensation in 2025, versus $332M in 2024 and $178M in 2023. We retain those primary annual figures and do not imply that an incomplete current-quarter cash-flow add-back is the total economic compensation expense. Current-quarter SBC is marked not separately verified in this summary. Missing precision is a reason to label a limit, not to manufacture a reassuring zero.

Source: FY2025 audited parent statements and Note 20. Ratios and valuation arithmetic are computed; company forecasts and our assumptions are labeled separately.

The capital-allocation test

The summer quarter repaired the first-half cash deficit. Subtracting the six-month amounts from the nine-month statements yields $3.426B of quarterly OCF and $940M of capex, or $2.486B of simple free cash flow. Nine-month simple free cash flow is lower because the first half consumed cash after capex. This is an important improvement over the prior episode's negative first-half cash observation, and it should be acknowledged plainly.

The allocation test remains demanding. Repurchases used $3.001B over nine months, nearly twice the $1.518B simple FCF. Dividends consumed another $459M. This comparison does not trace each dollar of financing to an individual buyback, and it does not prove that repurchases are irrational. It shows that current cash generation, asset transactions, debt and retained liquidity must be considered together. Saying all buybacks are comfortably funded by operating surplus would go beyond the statement.

The balance sheet has $603M of unrestricted cash, $406M of restricted cash, $3.989B of short-term borrowings, $359M of current long-term debt and $19.039B of longer-dated debt. Our net-debt convention subtracts only unrestricted cash, producing $22.784B. It is a deliberately clear convention; it is not the company's rating-agency or covenant leverage measure. Restricted cash, collateral facilities and derivative assets do not all behave like spendable cash for common shareholders.

Nuclear decommissioning trust assets of $20.382B support dedicated obligations and are not a spare cash pile to distribute. Asset retirement obligations were $12.887B, while regulatory-agreement liabilities and tax effects complicate a simple trust-minus-obligation calculation. The company highlights investment-grade ratings, credit facilities and long asset lives. Those are relevant strengths, but the funding plan still needs to accommodate uprates, integration, the pending RISEC purchase and operating volatility.

Source: Balance sheet and nine-month cash flows. Ratios and valuation arithmetic are computed; company forecasts and our assumptions are labeled separately.

Three routes and the market-implied case

Our base is $285, with high uncertainty and conviction three out of five. Route one assumes $18 of adjusted earnings power in 2029, a 22-times exit multiple and a 9% cost of equity over 3.14 years from the research date. It yields $302.12 today. The $18 input is our assumption, informed by management's base/enhanced framework and contracting progress; it is not a newly issued company 2029 total-EPS guarantee. In particular, a statement about growth in base earnings must not be applied indiscriminately to total adjusted earnings.

Route two uses $13.30 of assumed 2027 normalized EPS, a 23-times multiple and the same 9% discount rate for 1.14 years, yielding $277.28. The assumption is close to the saved $13.32 provider consensus, but that snapshot may not reflect today's report. A premium multiple recognizes scarce reliable generation and contracting prospects. It also exposes the stock to interest-rate and sentiment changes if expected growth becomes less certain.

Route three discounts equity cash per share of $9, $12, $15, $17 and $18 for 2027 through 2031 at 9%, followed by 3% perpetual growth. It yields $251.46. These are analyst assumptions after financing, taxes, economic compensation and reinvestment; they are not issuer FCF-before-growth guidance. Because the modeled cash is already an equity cash stream, subtracting net debt again would double-count financing. The terminal value dominates this route, which is why we show it as a sensitivity check rather than a precise measurement.

Weights of 50%, 30% and 20% on those routes produce approximately $284.54, rounded to $285. The $273.95 market quote is only about 4.0% below that base. At a 22-times 2029 multiple and 9% discount rate, it already implies roughly $16.32 of 2029 EPS. The market does not require the bull case to support today's price, but it also does not provide a large cushion against execution errors. A 25% margin of safety would be $213.75 while the operating thesis holds; that is an analytical entry reference, not a forecast that the price must trade there.

The bear case uses $14 of 2029 EPS, 18 times and a 10% discount rate, approximately $187 before rounding to $185. The bull case uses $22, 24 times and 8.5%, approximately $409 before rounding to $410. Base is triangulated rather than merely rounded from route one. The probability-weighted value of $291.25 is distinct from the base. Expected value is not a promise, and a wide range is an explicit admission that power-market and project outcomes can differ materially.

Source: Issuer earnings framework; valuation assumptions and calculations are Charged Alpha. Ratios and valuation arithmetic are computed; company forecasts and our assumptions are labeled separately.

Wall Street context, with dates

The saved target news includes CIBC at $343 on October 28, Barclays at $345 on October 27, HSBC at $335 on October 22 and Morgan Stanley at $378 on October 21. All four predate the November 9 earnings release. They are not post-print endorsements, and the table is a selected recent set rather than a complete analyst consensus. We distinguish each firm's dated target from our present-value model instead of presenting them as interchangeable estimates.

Our $285 reference is below each of those targets. The disagreement can be explained without assuming that the market or another analyst is irrational. Higher expected earnings, a lower discount rate, a longer premium-multiple period or more credit for prospective contracts can produce a higher value. Our main reservation is that durable contracted generation still requires capital and that a strong merchant quarter should not be capitalized as a perpetual run rate.

FMP's annual estimate snapshot shows EPS of about $12.17 for 2026, $13.32 for 2027, $16.96 for 2028 and $19.82 for 2029. The 2026 average is below the new issuer guidance range, a warning that provider aggregation can lag a fresh release or combine definitions. We do not describe that gap as an investor opportunity by itself. The 2029 EPS estimate has only four contributing EPS analysts in the snapshot, so its apparent precision overstates the depth of agreement.

The earnings record compares adjusted EPS expectations with adjusted actual EPS. GAAP and adjusted definitions differ materially for utilities with hedges, tax credits and decommissioning assets. A model that mixes the provider's adjusted expectations with GAAP actual EPS can manufacture a misleading beat or miss. The research retains both series and describes the matched definition explicitly.

Source: Saved price-target news and analyst-estimate responses. Ratios and valuation arithmetic are computed; company forecasts and our assumptions are labeled separately.

Contract growth and capital discipline

Management's clearest operating achievement is converting customer demand into longer-duration agreements while keeping the existing fleet available. The Amazon arrangement at Calvert Cliffs supports investment and a capacity expansion; the Google collaboration supports incremental nuclear capacity across existing units. The release distinguishes the 20-year power agreement from a separate 15-year financial load-following agreement. Combining every stated megawatt into one equivalent new-generation total would exaggerate what is being built.

The deck identifies 1,165 MW of uprates with approximately $4.3B of project cost. About $1.3B of investment is already within the 2026–2027 growth-capital guidance, while approximately $3B of incremental growth capital is identified for 2028–2032. Those amounts concern the specified uprate portfolio. They should not be casually added to every other press-release investment headline, because scopes and infrastructure programs may overlap.

The DOE commitment is conditional and for up to $5B. The company describes financing through a Federal Financing Bank loan with a 30-year term and rates fixed on advances at a spread above comparable Treasuries. An attractive indicated spread improves potential financing economics, but an undrawn conditional commitment is neither cash in the bank nor debt already outstanding. Execution, eligible spending, project approvals and definitive documentation remain relevant.

The prior episode's positive test for additional long-term nuclear contracting has gained evidence. The prior concern about debt-funded repurchases while cash remains weak is partly improved by the strong Q3 cash generation, but nine-month repurchases still exceed simple FCF. The old 2027 EPS warning cannot yet be graded against a new formal total-EPS outlook. We label it pending instead of treating a provider estimate or a base-earnings framework as a final company guide.

1. How much of the incremental contract margin is available after uprate capital and recurring maintenance?
2. Which project approvals and customer milestones govern the first incremental megawatts?
3. How will repurchases respond if merchant conditions soften before contracted cash arrives?
4. What portion of the DOE commitment is expected to be drawn, and when do definitive terms become binding?
5. How much of the 2029 opportunity is already included in base earnings versus still optional?

Source: Capital program, contracting and modeling slides. Ratios and valuation arithmetic are computed; company forecasts and our assumptions are labeled separately.

The bull case and the failure modes

The bull case starts with physical scarcity. Existing nuclear generation can supply reliable power on a timescale that is difficult to replicate with a new plant. Long-lived assets and strong customers can support attractive financing and a premium valuation when contracts convert demand into durable cash. Reliable summer operations and the current quarter's cash rebound demonstrate operating capability. The bullish argument deserves more than a dismissive claim that everything is hype.

The bear case starts with the price paid for distant benefits. Contracted demand still needs approved projects, grid access, maintenance, workers and capital. A long contract does not eliminate counterparty, regulatory or construction risk. Merchant and commercial earnings can decline before the new contracted earnings arrive. If a premium valuation assumes smooth progress, a delay can hurt both the numerator of earnings and the multiple investors apply.

Tax-credit economics require precise language. The deck describes a 2026 PTC maximum of $15/MWh, a gross-receipts threshold of $26/MWh and phase-out after $44.75/MWh, subject to stated conditions. That does not create a universal unconditional cash floor for every asset and every hour. In the issuer's own $35/MWh example, the stated credit brings value to $42.80/MWh before its separate tax-adjusted illustration. We do not turn a phase-out threshold into a guaranteed realized selling price.

Our position between the cases is HOLD near modeled value. The company has valuable assets and improved contracted visibility, but the valuation cushion is modest and financing remains part of the analysis. A lower quote alone would not make the stock attractive if reliable earnings power or project economics deteriorated simultaneously. Conversely, better cost and cash evidence could justify raising value without waiting for a large price decline.

Source: Issuer risk factors, operations and forward-looking qualifications. Ratios and valuation arithmetic are computed; company forecasts and our assumptions are labeled separately.

What would change the view

The next annual earnings report should make the cash-allocation test easier to judge. The dated checks here are review deadlines, not confirmed company reporting dates. We want a same-definition comparison of the $12.50–$12.80 annual guide with actual results, a clear separation between total capex and growth spending, and a bridge between repurchases and cash available after investment. Meeting the earnings guide while using materially more debt would be a different outcome from meeting it with stronger free cash generation.

Project milestones matter before all the revenue arrives. Definitive DOE financing terms, the RISEC acquisition decision and incremental uprate approvals can change risk even if quarterly EPS does not immediately move. We will focus on binding terms and completed work. A press release about an opportunity, a conditional commitment and an executed financed project are separate stages of evidence.

New contracts can increase value when they improve risk-adjusted cash returns. They can also require more capital, carry unusual cost-allocation provisions or crowd out attractive merchant exposure. The next review should examine those economics before adding a simple per-gigawatt premium to valuation. The issuer expressly warns that some opportunity sensitivities may not be additive; our model preserves that warning.

The practical decision checkpoints are therefore earnings delivery, cash after total investment, project cost, financing execution and contract retention. None requires predicting tomorrow's quote. If those checks improve together, our normalized earnings and cash assumptions can rise. If they deteriorate, holding the old $285 number merely for consistency would be a mistake.

Source: Project timing and forward-looking disclosures. Ratios and valuation arithmetic are computed; company forecasts and our assumptions are labeled separately.

Data, methods and limits

All latest-quarter reported amounts come from the November 9 SEC-filed earnings release and presentation. The retained dataset also includes nine quarters of provider income statements, balance sheets and cash flows, three years of primary annual cash-flow and compensation data, the saved market quote, daily price history, earnings expectations and dated analyst targets. FMP historical standardization is not a substitute for issuer definitions; the current quarter has been checked directly against the release.

The parent company is Constellation Energy Corporation, listed on Nasdaq as CEG. The covered security is common stock, with no ADR conversion. The 2025 annual report identifies PricewaterhouseCoopers as auditor and reports effective year-end controls. That audit conclusion is dated December 31, 2025; it is not a newly inferred Q3 certification. Consolidated energy-project variable-interest entities are disclosed, including restricted asset recourse; this is distinct from an offshore contractual-control equity structure.

The annual report lists 15,339 employees at year-end 2025. Earlier employee counts and a complete current-quarter plant-level segment-profit history were not independently reconstructed for this packet, and no invented figures fill those gaps. The financial history is presented at the consolidated level. Operating metrics are shown only with their reported scope. A current Q3 Form 10-Q was not in the initial SEC submission snapshot; the complete earnings attachments supply the current statements and guidance without treating that normal same-day filing sequence as a delay.

Free cash flow in our simple comparisons means GAAP operating cash flow minus total capital expenditures. The company's FCF before growth excludes or adjusts additional items and is a different measure. Net debt excludes restricted cash; enterprise value is market capitalization plus that net debt convention. Valuation cash flows are equity cash flows after financing and reinvestment. Probabilities, multiples, discount rates and monitoring thresholds are our judgments. Precision in the arithmetic does not remove uncertainty in those assumptions.

Source: Primary current-period evidence. Ratios and valuation arithmetic are computed; company forecasts and our assumptions are labeled separately.