NASDAQ · Utilities · Independent Power Producers
CEG Constellation Energy Corporation
As of 2026-05-20 · Baltimore, MD
Business summary
Constellation Energy is the largest private-sector power producer in the United States and the country's largest operator of nuclear plants. Following the January 7, 2026 acquisition of Calpine, the company owns ~55 GW of generating capacity across nuclear, natural gas, geothermal, hydro, wind, and solar — roughly 10% of U.S. clean energy generation. Pre-Calpine, the owned fleet was 31,676 MW (FY2025 10-K, p.8), with another 4,798 MW under unit-specific contracts. Calpine added ~23 GW (72 generation + battery storage assets), of which the majority is natural-gas–fired generation concentrated in Texas (ERCOT), California, and the Northeast.
The business model has two legs. The generation side sells output into wholesale power markets (PJM, ERCOT, NYISO, MISO) and, increasingly, under long-term Power Purchase Agreements directly to corporate customers — most importantly hyperscalers building AI data centers. The retail side is one of the largest competitive retail electricity suppliers in the U.S., serving ~2.5 million customer accounts including three-quarters of the Fortune 100. Post-Calpine, the retail platform adds ~62 TWh of annual load. Roughly 55-60% of revenue is generation/wholesale; the balance is retail and ancillary services (pro-forma estimate from Q1 2026 segment data).
The moat is structural and time-bound. The 21 reactors CEG operates (plus Crane/TMI restarting) cannot be replicated; new U.S. nuclear capacity has been effectively unbuildable at scale since the 1990s, and even SMR projects are 5-10 years from material commercial output. As hyperscaler power demand grows faster than grid interconnections can be approved, the operators of existing dispatchable, low-carbon nameplate capacity in load-pocket regions (PJM, ERCOT) hold an option-like position: they can sign PPAs at premium prices that didn't exist 3 years ago. The Microsoft TMI restart and the Meta Clinton PPA are the public face of a procurement scramble that the rest of the market is still pricing in.
Connected theses
- T001 — Power, not silicon, is the binding constraint on the AI build-out · core
Largest single position in the AI-power thesis. Original sizing (8% / 30 shares @ $262 = $7,860) reflects highest conviction within T001. The Calpine close materially diversifies CEG away from pure-nuclear and into ERCOT gas, which strengthens — not weakens — the thesis: more dispatchable capacity in the hottest data-center growth market.
Key metrics
| Nuclear fleet size (operating) | 21 reactors at 14 stations Largest US nuclear fleet by operating capacity. | FY2025 |
|---|---|---|
| Total capacity post-Calpine | ~55 GW Up from 31.7 GW pre-Calpine close. | Q1 2026 |
| Q1 2026 operating revenues | $11,122M (vs $6,788M Q1 2025; +64% YoY GAAP, +22% pro forma) Calpine integration is the primary driver of YoY; pro forma figures isolate organic growth. | Q1 2026 10-Q |
| Q1 2026 net income (common) | $1,590M (vs $118M Q1 2025; $147M pro forma) Includes $582M of mark-to-market hedging gains in net income — bonus from favorable forward power curves, will not recur at this magnitude. | Q1 2026 10-Q |
| DOE loan guarantee for Crane | $1.0B unsecured Federal backing reduces effective cost of capital for the restart. | 2025 |
| Capacity factor (nuclear fleet) | 94-95% (industry-leading) See educational_notes — capacity factor explainer. | Multi-year average |
Valuation snapshot
| Price | $262.00 |
|---|---|
| Market cap | $82.0B |
| Forward P/E | 28.5× |
| EV / EBITDA | 14.8× |
| FCF yield | 3.1% |
CEG trades at a premium to its 5-year historical multiple (28× fwd vs 19× avg), reflecting the structural re-rating from hyperscaler PPAs + Calpine accretion. The bull case is that traditional multiples understate the optionality on 24/7 nuclear in a power-constrained market — the right comp may now be 'AI infrastructure' (which trades 30-50× fwd) rather than 'utility' (which trades 15-20×). The bear case is multiple compression if hyperscaler capex slows or PPA economics normalize. At ~3% FCF yield it's not cheap on cash flow, but the FCF inflection from Crane (commercial 2027-28) and Calpine synergies should expand the denominator over 2-3 years.
Evidence
- 20-year PPA signed September 2024; Microsoft purchases full output of restarted TMI Unit 1 (~835 MW). $1.0B DOE loan guarantee backs the restart. Renamed Crane Clean Energy Center.
- Meta PPA covers Clinton nuclear station (~1,080 MW); supports Meta's 24/7 clean-energy goal and continues operation of a plant that had been at retirement risk.
- primary Calpine acquisition closesCash + stock acquisition of Calpine closed Jan 7, 2026. Adds ~23 GW (72 assets), largest U.S. gas + geothermal generator. Strong ERCOT/California/Northeast footprint. Adds 62 TWh of annual retail load and ~2,500 employees.
- primary CEG Q1 2026 10-Q filingQ1 2026 operating revenues $11.1B (+64% YoY GAAP, +22% pro forma). Net income to common $1,590M vs $147M pro forma a year ago. Notable: $582M of mark-to-market hedging gains in net income — non-recurring at this magnitude.
- primary NY ZEC program extensionNew York extended Zero-Emission Credit program in January 2026, reaffirming state-level support for existing nuclear in NY market (Ginna, FitzPatrick, Nine Mile Point).
- Nuclear Production Tax Credit (originally enacted in IRA 2022) maintained under the One Big Beautiful Bill Act (OBBBA). Provides federal price floor for existing nuclear generation.
- CEG intends to file applications to extend nuclear unit licenses to 80 years where policy support continues — extending revenue tail by decades beyond current 60-year terms.
Catalysts
- Q2 2026 earnings — first full quarter post-Calpine close high
What to watch: Pro-forma revenue/earnings growth ex-hedging gains; integration cost trajectory; PPA backlog updates; capacity-factor stats; any commentary on Crane restart timing. - PJM Base Residual Auction results (typically Q3) high
What to watch: Capacity clearing prices — 2025 saw record-high prints; sustained pricing would be a major positive. - Crane Clean Energy Center restart — commercial operation high
What to watch: NRC license restart approval, mechanical readiness milestones, first power delivery to Microsoft. - 80-year license extension applications medium
What to watch: First filings; NRC engagement; precedent from PSEG/Salem & Hope Creek and STPNOC who already have 20-year renewals.
Falsifiers
- Loss or material renegotiation of either the Microsoft (Crane) or Meta (Clinton) PPA at lower realized economics
armed · Monitor 8-K filings + hyperscaler announcements. A 'PPA termination' or 'amended at lower rate' disclosure would be the canary. - Extended unplanned outage (>60 days) at any single reactor station — operational risk specific to a 21-reactor fleet
armed · NRC event reports + quarterly capacity factor disclosure. CEG typically operates at 94-95% capacity factor; sustained dip below 90% would be a flag. - Calpine integration produces cost overruns or culture issues; pro-forma EPS guidance revised down
armed · Quarterly earnings — watch integration-cost line (Q1 2026 was $126M; rising trend is concerning) and any commentary on retention/operational mishaps. - PJM capacity auction clears at materially lower prices than 2025 levels for 2 consecutive years (would compress merchant power earnings)
armed · PJM Base Residual Auction results — typically posted annually. Combine with company commentary on capacity revenues. - Repeal of nuclear PTC or significant narrowing of eligibility
armed · Federal tax legislation. The PTC was preserved in OBBBA; the bear case requires bipartisan reversal.
Agent notes
High conviction core hold in T001. The thesis I wrote on May 18 framed CEG as a nuclear pure-play; the 10-K reading clarifies that post-Calpine CEG is materially more diversified (gas, geothermal, renewables, retail) — which actually strengthens the thesis. Three reasons: (1) ERCOT/Calpine gas + storage exposure puts CEG in the hottest data-center market with dispatchable capacity that's even harder to replace than nuclear in that geography; (2) the retail platform is a hedge against pure wholesale price volatility; (3) the policy environment (NY ZEC extension, OBBBA preserving nuclear PTC, state-level legislation favoring dispatchable generation) is a multi-year tailwind that I underweighted in v0. The main risk I'd watch is the Q1 net income figure being flattered by $582M of MTM hedging gains — strip those out and the run-rate is more modest. Q2 print will clarify the true earnings power.
User bought a few shares May 20. My recommendation: this is a position to add into weakness (any pullback below $245), not chase here. The structural re-rating is real but the next leg up needs either a confirming Q2 print, a new hyperscaler PPA announcement, or a strong PJM auction. Patient holding ≥18 months is the right frame.
Educational notes
📚 capacity factor
Capacity factor is the share of the year a power plant actually produces electricity at its full nameplate output. Nuclear plants like CEG's run at 94-95% capacity factor — basically always on. Wind runs ~35%, solar ~25%, because they depend on weather. Combined-cycle gas can run anywhere from 30% to 85% depending on whether it's a baseload or peaker plant. This matters enormously for AI data centers, which need 24/7 power: a 1 GW solar farm only delivers ~250 MW of round-the-clock equivalent power, while a 1 GW reactor delivers ~940 MW. That's why hyperscalers will pay a premium for nuclear PPAs even when wind/solar PPAs are nominally cheaper per MWh.
📚 Power Purchase Agreement (PPA)
A PPA is a long-term contract where an electricity buyer (corporate or utility) commits to buy a power plant's output at a fixed price (or a price formula) for 10-25 years. PPAs let the seller finance and operate the plant without exposure to spot power markets, and let the buyer lock in cost and carbon attributes. For CEG specifically, the Microsoft Crane and Meta Clinton PPAs are both 'physical' PPAs — Microsoft and Meta actually take title to the electrons (or attributes) — at prices believed to be in the $80-110/MWh range, vs. wholesale prices of $40-60/MWh in those markets. That spread is why CEG is being re-rated.
📚 PJM capacity market (RPM)
PJM is the regional grid operator covering the Mid-Atlantic and Midwest US. Beyond paying for actual electricity generated, it also runs an annual 'capacity auction' (Base Residual Auction) where generators get paid just for being available to produce power three years out. This is the Reliability Pricing Model (RPM). For owners of dispatchable capacity like CEG, capacity payments are a meaningful chunk of revenue separate from energy sales. The 2025 auction cleared at record-high prices because PJM is short on dispatchable resources relative to projected data-center load growth — a structural tailwind for CEG.
📚 Nuclear Production Tax Credit
The IRA (2022) created a per-MWh tax credit for existing nuclear plants, ramping the credit up when wholesale power prices fall below a threshold and ramping down when prices are high. It's effectively a floor on cash flow for nuclear generators — if power prices collapse, the PTC catches some of the downside. The OBBBA (One Big Beautiful Bill Act, 2025) preserved this PTC despite the broader rollback of IRA energy provisions, reflecting bipartisan support for keeping the existing nuclear fleet operational.
Open questions
- What is the Crane restart cost basis vs. the implied PPA economics? The $1B DOE loan guarantee helps but the all-in NPV of the Microsoft deal would clarify pricing power.
- How much of Q1's $1,590M net income is structurally repeatable vs. one-time MTM gains? Run-rate normalized EPS matters more than headline.
- What is the Texas Energy Fund (TEF) opportunity set for Calpine assets beyond the $278M Pin Oak Creek loan?
- Are there additional hyperscaler PPAs in the pipeline? CEG hasn't explicitly hinted at deal #3 yet.