CEG — Research Report (2026-07-23)

TL;DR

• What it is: America's largest nuclear power generator (Constellation Energy), spun off from Exelon in 2022, now selling carbon-free electricity directly to Big Tech data centers.
• Overall take: A genuinely scarce asset (nuclear power AI companies desperately want) trading at a valuation that already assumes years of flawless execution.
• Decision summary: Watch, Medium conviction — great business, priced for perfection; wait for a better entry or clearer regulatory signal (educational framework, not advice — see §Decision below).
• Revenue $25.5B in 2025 (steady growth, not explosive — nuclear plants can't scale output fast)
• Net income $2.3B in 2025, down from $3.7B in 2024 (one-off gains faded, see §2)
• Free cash flow flipped positive in 2025: +$1.3B (first positive FCF year since spin-off — funding the AI buildout used to eat all the cash)
• Stock down ~35% from its 2026 high (market's re-pricing regulatory/timeline risk, not a business breakdown)
• Strongest bull point: 20-year fixed-price power deals with Microsoft and Meta lock in demand nobody else can supply (nuclear is the only large-scale 24/7 carbon-free source).
• Strongest bear point: Goldman Sachs flags CEG trades ~3 turns above peer Vistra on EV/EBITDA with the lowest free-cash-flow yield in its coverage — the premium requires everything to go right.
• Watch next: FERC's final order on the Crane (Three Mile Island) restart timeline — a slip pushes out the Microsoft deal's earnings contribution.

1. Business overview

So what: CEG doesn't make anything you'll ever hold — it makes electrons, and it owns the country's biggest fleet of the one power source that runs nonstop without emitting carbon: nuclear reactors.

Constellation Energy Corporation was spun off from Exelon in February 2022 as the standalone generation ("competitive") business, while Exelon kept the regulated wires-and-poles utilities. CEG owns and operates roughly 21 nuclear reactors (~21 GW) plus natural gas, hydro, wind, and solar assets, concentrated in PJM (the mid-Atlantic/Midwest grid) and ERCOT (Texas).

Revenue model: CEG generates electricity and sells it three ways: (1) wholesale into competitive power markets, (2) retail supply contracts to businesses and municipalities, (3) increasingly, long-term bilateral power purchase agreements (PPA) — a contract where a buyer locks in a fixed price for power over many years — directly with data-center operators. It is not a regulated utility with a guaranteed rate of return; its earnings ride real-time and contracted power prices.

Customer concentration: Historically diversified (grid operators, municipal utilities, commercial/industrial). The strategic shift of 2024–2026 is toward a small number of hyperscale tech customers — Microsoft (Three Mile Island/"Crane" restart, $16B 20-year PPA, 835 MW) (Yahoo Finance), Meta (20-year nuclear PPA) (Every Nuclear-Powered Data Center Deal, smrintel.com), and reportedly Amazon (1,920 MW PPA) (enkiai.com). This concentrates counterparty quality (investment-grade tech giants) but also concentrates revenue mix risk into a handful of contracts.

Major structural event: CEG acquired Calpine (gas/geothermal generator) in 2025–2026; to win regulatory approval, it agreed to divest 4.4 GW of PJM gas assets to LS Power for $5B (Constellation press release). Net effect: CEG is shedding secondary gas assets and concentrating into a "nuclear + geothermal premium" portfolio — a deliberate pure-play tilt, not a retreat.

2. Financials (5-year, $ millions unless noted)

So what: Revenue is steady, not explosive (physics — you can't spin up a new reactor in a year); the real story is cash flow finally turning positive in 2025 after years of funding growth from the balance sheet.

*Net debt = long-term debt minus cash. Source: EDGAR companyfacts (SEC), CEG CIK 0001868275.

Revenue CAGR 2020→2025: ~7.7%/yr — respectable, not a growth-stock number. Nuclear output is largely fixed by reactor capacity; growth comes from price, not volume, until new capacity (uprates, restarts) comes online.

Why 2024 net income ($3.7B) beat 2025 ($2.3B) despite AI-deal headlines: operating income fell from $4.35B to $3.09B — 2024 included favorable mark-to-market and hedge accounting gains that partly reversed in 2025 (typical for a merchant generator; power-price hedges swing both ways). This is not a deterioration in the underlying nuclear fleet's operations — it's the normal noise of a commodity-exposed income statement.

FCF turned positive in 2025 for the first time since the spin-off. From 2021–2024 CEG burned cash (peak burn −$7.7B in 2023) funding growth capex and working capital around the PPA buildout. 2025's +$1.3B is the first year cash generation outpaced investment — worth watching whether it holds as Calpine integration capex ramps.

Balance sheet: Total assets grew from $48.1B (2021) to $57.2B (2025); stockholders' equity grew from $11.2B to $14.5B. Net debt fell from $7.1B (2023) to $3.6B (2025) even while capex rose — cash generation, not deleveraging via asset sales, drove this (the LS Power divestiture closes later in 2026 and isn't yet reflected).

Shares outstanding fell from 327M (2022) to 312M (2025) — modest net buybacks, not dilution.

3. Unit economics

So what: the one number that matters for a nuclear generator is realized price per megawatt-hour (MWh) versus fleet-wide production cost — nuclear's fuel cost is tiny, so almost every dollar of power price above the "all-in cost to run the plant" (roughly $30–40/MWh across the industry) drops to margin.

CEG does not break out a clean "$/MWh" unit economic line in the GAAP tables provided, but the structural logic is:

• Fixed output, variable price: the ~21 GW nuclear fleet produces roughly the same number of MWh every year (reactors run at 90%+ capacity factor almost regardless of demand). Revenue growth is almost entirely a function of the price CEG can charge, whether wholesale market price or a locked-in PPA rate.
• PPA lock-in economics: the Microsoft Crane deal reportedly locks in $16B over 20 years for 835 MW — that math implies roughly $115/MWh average realized price for that single contract (a premium to typical wholesale power prices, reflecting the tech buyer's willingness to pay for guaranteed carbon-free baseload).
• Operating margin is the best available proxy for unit economics here: it swung from 18.5% (2024) to 12.1% (2025) — power-price and hedge volatility, not a change in the physical cost structure.

(Estimated — CEG's public disclosures don't provide a single clean "$/MWh" cheat-sheet number; the $115/MWh figure above is derived from press-reported deal terms, not company guidance.)

4. Metrics table with the WHY

So what: every ratio below is a lens on the same question — is CEG's scarce nuclear fleet worth the price the market is charging for it?

4b. Sector valuation benchmarks

So what: CEG is expensive even for an expensive sector — the premium is the whole investment debate.

• Metric used: EV/EBITDA, because CEG is capital-heavy (nuclear plants, big capex) and carries meaningful debt — EV/EBITDA is debt-neutral and the standard yardstick for power generators, unlike P/E which ignores leverage.
• CEG has traded at an average 35% valuation premium to its independent power producer (IPP) peer group (Vistra, NRG, Talen), attributed to its superior credit profile, scale, and carbon-free portfolio (source: Lambda Finance, "Vistra vs Constellation vs Talen").
• Peer comps: Vistra ~10.0× EV/EBITDA, Talen ~10.7×; recent M&A in the sector (NRG/LS Power gas deal, Vistra/Lotus deal) has priced at 7–8× 2026 EV/EBITDA for un-differentiated gas assets (Nasdaq).
• Goldman Sachs (Neutral, $305 PT) states CEG trades "~3 turns above VST on FY2 EV/EBITDA and at the lowest FCF yield in coverage" (Yahoo Finance) — i.e., roughly 13× vs. peers' ~10×.
• Read: CEG's premium is real and sizeable, and multiple sell-side desks (Goldman, and price-target cuts from Citi/Mizuho/JPMorgan/BofA/UBS/KeyBanc/BMO/Raymond James/Barclays/Scotiabank since February 2026) treat it as a risk, not a bargain. The premium is earned by scarcity (nuclear can't be built quickly) but leaves little room for execution slips.

5. Ownership & management

So what: management is executing the AI-power pivot fast — the open question is discipline on price paid, not strategic direction.

• CEO: Joseph Dominguez, President & CEO since the 2022 spin-off; a former Exelon regulatory/government-affairs executive, publicly framing CEG as the "largest carbon-free U.S. generator" in investor and media appearances (CERAWeek, CNBC 2026) (Constellation on X, CNBC).
• Insider activity: A Form 4 filing exists for early 2026, but specific transaction details (shares, price) weren't retrievable from search snippets — flagged as open; verify directly at SEC EDGAR Form 4 filings for CIK 0001868275.
• Near-term supply-side risk: lockup expiration on 25 million Calpine-linked shares around June 30, 2026 creates a technical overhang independent of business fundamentals.
• No activist involvement or contested control situation found in this pass.

6. M&A track record

So what: CEG's two biggest deals (Calpine acquisition, LS Power divestiture) both point the same direction — trading breadth for a cleaner, higher-quality nuclear/geothermal story.

• Calpine acquisition (2024–2026): added gas and geothermal generation, expanded scale — but triggered the antitrust review that forced the LS Power divestiture.
• LS Power divestiture ($5B, 4.4 GW of PJM gas assets, ~$1,142/kW) — required by FERC/DOJ as a condition of the Calpine deal; framed as sharpening the portfolio into a "premium" nuclear/geothermal mix (PowerMag).
• Blue Energy investment (2026) — a smaller stake tied to nuclear-adjacent demand (Fool.com) — a minority stake, not a control transaction.

7. Market & value-chain position

So what: CEG doesn't compete on price like a normal power company — it holds a bottleneck asset (existing licensed nuclear reactors) that can't be replicated on a relevant timescale.

• Bottleneck or commodity? Bottleneck. New nuclear plants take a decade-plus to permit and build; CEG's ~21 GW of already-licensed, running reactors is fixed, non-reproducible supply. Data centers can't run on intermittent solar/wind alone — hence 20-year PPAs at premium prices.
• Who has pricing power: CEG has real pricing power over data-center buyers for existing nuclear output. It does not have pricing power over regulators — FERC/DOJ can slow or reshape deals, and PJM rulings can swing near-term revenue.
• Named competitors: Vistra (VST), NRG Energy (NRG), Talen Energy (TLN) — none matches CEG's nuclear scale.
• Regulatory dependency: the Crane restart underlying the Microsoft deal is contingent on FERC's final order — a bottleneck-on-the-bottleneck.

8. PE/quality lens

So what: a private-equity buyer would love the scarcity and contracted cash flows; would fear paying today's public-market premium and inheriting regulatory risk they can't control.

• Moat: Strong and structural — existing nuclear licenses are close to un-replicable.
• Pricing power: Demonstrated — Microsoft/Meta PPAs signed at premium rates for guaranteed carbon-free baseload.
• What a buyer would love: contracted 20-year revenue visibility; FCF just turned positive; falling net debt despite rising capex.
• What a buyer would fear: paying a ~35% sector premium for assets exposed to regulatory timeline risk, plus commodity power-price volatility that swung operating margin from 18.5% to 12.1% in one year.

9. Bull / Base / Bear (attributed)

So what: Wall Street is split between "own the irreplaceable asset" and "you're paying too much for it" — both camps agree on the business, disagree on the price.

Bull — Morgan Stanley (Overweight, PT raised to $364 from $359): confidence in CEG's premium nuclear franchise and PPA pipeline despite the 2026 sector pullback. BofA also upgraded to Buy, projecting double-digit growth through 2026 (Investing.com).

Base — Consensus (21 analysts, mean target $367–368, street high $441): 19 of 21 rate Buy/Outperform, but targets have been trimmed across the board since February 2026 as banks factor in updated contracting timelines and more conservative long-term multiples (TIKR, Simply Wall St).

Bear — Goldman Sachs (Neutral, $305 PT): the valuation itself is the risk — CEG trades ~3 turns above Vistra on forward EV/EBITDA with the lowest FCF yield in Goldman's coverage, meaning the price already assumes scarcity premium, PPA optionality, and Calpine synergies all land cleanly (Yahoo Finance). Citigroup (Neutral, PT cut to $297 from $348) flagged PJM reliability-market risk in its July 2026 downgrade (TheStreet). Barclays' 2029 earnings outlook sits below consensus, citing execution risk.

Catalysts: FERC's final order on Crane; more hyperscaler PPA announcements; PJM rule clarity; Q2 2026 earnings (~Aug 6, 2026).

What would prove the thesis wrong: a FERC delay/denial on Crane; a PJM ruling capping merchant power prices; a hyperscaler PPA renegotiation/cancellation; sustained margin compression below 2025's 12%.

10. Practitioner decision layer

10a. Reverse expectations (Mauboussin-style approximation — not a DCF)

At ~$271.62/share and 2025 GAAP EPS of ~$7.43, CEG trades at ~36.6× trailing P/E. Consensus 2026 EPS is ~$11.74, implying a forward P/E of ~23× — a premium to typical regulated-utility multiples (14–18×).

For that multiple to be "earned," the market needs sustained double-digit EPS growth for several years. Judgment: stretched, not implausible — plausible given signed contracts, but leaves no margin for a single miss. This is a rough consensus-and-multiple sanity check, not a DCF.

10b. Variant perception

1. Consensus believes: CEG's nuclear-to-AI pivot justifies a structural re-rating above historical multiples; 19 of 21 analysts rate it Buy.
2. This report's differentiated view: None found beyond what Goldman Sachs and Citigroup have already published.
3. Evidence: Multiple banks have already cut targets since February 2026 citing exactly the valuation/execution risks a variant view would raise.
4. Consensus agreement — no supported variant perception found. The bear case here is already Wall Street's own stated Neutral view, not a contrarian call.

10c. Probability-weighted expected value

EV = 0.30×400 + 0.45×320 + 0.25×250 ≈ $326.50 — modestly above the current price, but the wide scenario spread argues against treating this as a clean margin of safety.

10d. Pre-mortem (assume price halved to ~$136 by mid-2028)

1. FERC/regulatory delay or denial on Crane, or PJM rules cap merchant prices. Early signal: docket delays; weak PJM capacity auction results.
2. A hyperscaler PPA is renegotiated, delayed, or cancelled. Early signal: an 8-K disclosing a PPA amendment; a customer's capex guidance cut.
3. Calpine integration costs or divestiture proceeds disappoint, FCF reverts negative. Early signal: FCF flipping negative again; net debt rising.

10e. KPI driver tree and sensitivity

Core drivers: realized price/MWh × nuclear capacity factor (~90%+, limited upside) × operating margin (12–18% band).

• Price/MWh +10% → operating income likely moves more than 10% given nuclear's low variable cost (operating leverage) — rough estimate, not a model.
• Capacity factor −10% (e.g., unplanned outage) → revenue drops roughly proportionally — the single biggest single-plant risk given concentration in ~21 reactors.
• Margin reverting to 2025's 12.1% from consensus assumptions near 15–18% would pressure the ~23× forward multiple.

10f. Scuttlebutt proxies

• Hiring-posting trend: unavailable in this search pass — check LinkedIn/Indeed directly.
• Employee-review tone: unavailable in this pass — would need a direct Glassdoor lookup.
• Customer/supplier earnings-call mentions: confirmed — Microsoft and Meta have both referenced nuclear PPAs with Constellation in their own investor communications per press reports, corroborating the deals beyond CEG's own press releases.

Decision (Educational — not investment advice)

• Decision: Watch
• Conviction: Medium
• Core assumption: CEG's nuclear scarcity value justifies a premium to IPP peers, but the current ~23× forward P/E already prices in flawless execution on FERC approvals and PPA ramp-up, leaving asymmetric downside if either slips.
• Review date: 2026-08-06 (Q2 2026 earnings, estimated)
• Controlling question: Has FERC issued its final order on the Crane restart, and did management reaffirm the PPA revenue ramp timeline on the Q2 call?
• Review trigger: A FERC decision (either direction) on Crane, a pullback toward the $250 bear case, or a break above $320 on confirmed catalyst news.

📚 What this company teaches

1. Scarcity beats growth rate. CEG's revenue grows slowly (~7.7%/yr, capacity-capped) yet commands a premium multiple because what it has cannot be built quickly by anyone else.
2. Cash flow and net income can tell different stories in the same year. 2025 net income fell from 2024, but FCF turned positive for the first time — GAAP profit includes hedge-accounting noise cash flow doesn't.
3. A "premium" valuation is a bet on things staying perfect. When multiple banks independently flag the same valuation gap, the market has already priced in the good news.

Self-test question: If CEG's forward P/E is ~23× versus a typical regulated utility's 14–18×, what would need to be true about future earnings growth for that premium to be "worth it" — and what single calendar event could most quickly prove that right or wrong?

📖 Glossary

• P/E: dollars paid today per $1 of annual profit.
• EV/EBITDA: enterprise value ÷ EBITDA; debt-neutral comparison across companies with different leverage.
• EV/Sales: enterprise value ÷ revenue; used when profit is near zero/negative.
• FCF: cash left after paying for the investment needed to keep running.
• OCF: cash generated from day-to-day operations, before investment spending.
• Capex: spending on long-lived physical assets like power plants.
• Net debt: total debt minus cash on hand.
• PPA: a long-term contract locking in electricity price for many years.
• Capacity factor: the share of a plant's max possible output it actually produces; nuclear runs 90%+.
• FERC: the US federal agency regulating interstate power transmission and wholesale markets.
• PJM: the regional grid operator covering CEG's biggest nuclear fleet.
• IPP: a company generating/selling power competitively, without a regulated utility's guaranteed returns.
• Merchant power: electricity sold at market prices rather than under regulated guaranteed-return contracts.
• Lockup expiration: end of a period barring shareholders (often post-M&A) from selling — its end can pressure the stock with no fundamental change.
• Uprate: an engineering upgrade raising an existing reactor's output without building a new plant.

Sources cited inline. EDGAR/GAAP figures per SEC companyfacts (CIK 0001868275). Market price, analyst targets, and deal details are as reported in web sources as of 2026-07-23 — re-verify against a live quote before any decision.

Note: I couldn't save this to investing/ceg-2026-07.md — the Write tool needs your approval first. Want me to retry the write, or is the inline output sufficient for now?