CEG — Constellation Energy Corporation
Investment Read as of 2026-07-20
The Read
CEG is a SOLID investment. Its core value—providing irreplaceable, firm nuclear power for AI data centers—is reinforced by recent regulatory approvals and sustained high demand signals, supporting the running conviction of P=0.86.
Bull case
- Irreplaceable Baseload Power: CEG owns some of the nation's most reliable, carbon-free electricity sources (Pillar 1). This firm power is exactly what AI infrastructure requires when other grids struggle with variable supply.
- Operational Stability: The successful subsequent license renewal application for the R.E. Ginna plant confirms regulatory confidence in CEG’s existing asset base [Federal Register].
- Hyperscaler Demand: Major tech companies are actively signing Power Purchase Agreements (PPAs) directly for nuclear power, proving the market values this specific source of clean, reliable energy (Pillar 2).
Bear case / what breaks it
- Restart Delays: If the Crane/TMI-1 restart slips past 2028, CEG loses a major near-term growth pillar [Thesis].
- Capacity Price Collapse: If PJM capacity auctions clear below $200/MW-day for two straight periods, it signals that market demand is weakening and the premium pricing model fails.
- Demand Flattening: A sudden, durable efficiency breakthrough in AI inference (the process of running LLMs) could significantly flatten future power demand curves, undermining the core thesis.
What the latest signal says
The recent regulatory filing for the Ginna plant confirms CEG's ability to maintain and expand its existing asset base [Federal Register]. Separately, academic papers detailing optimized tensor scheduling for LLM inference confirm that AI compute demands are complex and continue to drive specialized power needs [Reddit/LocalLLaMA], reinforcing the structural demand side of the thesis.
Posterior history
| Date | P | Δ | Call | Driver |
|---|---|---|---|---|
| 2026-07-15 | 0.86 | +0.05 | UP/MED | PJM capacity auction cleared at its price cap again — the exact opposite of CEG's break-tr |
| 2026-07-07 | 0.81 | +0.05 | UP/MED | Walmart signs its first-ever nuclear PPA — with Constellation, for Illinois ops — billed a |
| 2026-06-12 | 0.76 | -0.03 | REVIEW/- | Stable, tilting soft. The structural case (scarce firm nuclear into AI demand) is intact, |
| 2026-06-03 | 0.79 | +0.05 | UP/MED | Constellation's Three Mile Island (Crane) restart got a FERC waiver letting it transfer ca |
| 2026-05-29 | 0.74 | -0.04 | REVIEW/- | Stable, but the last move up looks soft — the 2026-05-27 bump to 0.78 rode tangential narr |
| 2026-05-27 | 0.78 | +0.10 | UP/HIGH | Multiple signals (Jefferies ESG, SK Submarines, MI Grid Plan) boost the narrative around n |
Thesis detail
Core thesis
Largest US nuclear operator (~55 GW post-Calpine, ~95% capacity factor) sitting
on irreplaceable firm baseload exactly when AI is starved for it — converting
uncontracted nuclear into 20-year hyperscaler PPAs at scarcity prices.
Pillars (with priors)
1. Existing nuclear fleet = the scarcest firm, carbon-free power for AI · P = 0.85
2. Hyperscaler PPAs (MSFT 835 MW; ~147M MWh uncontracted) re-rate the fleet · P = 0.70
3. Crane / TMI-1 restart adds ~835 MW by Q3–Q4 2027 · P = 0.60
4. Calpine synergies land as underwritten · P = 0.55
Expected news (the prior)
- Periodic new PPA announcements (MW, $, hyperscaler counterparty)
- Crane restart progress + FERC interconnection updates
- PJM capacity-auction prints; EPS roughly in line with ~$11.75 guide
Surface the residual: anything *not* on this list (a cancellation, a delay,
a competing supply source) is the signal.
Thesis-breaking triggers (→ set P near 0)
- ☐ Crane/TMI-1 restart slips past 2028
- ☐ PJM capacity clears <$200/MW-day for two consecutive auctions
- ☐ Calpine synergies miss by 30%+
- ☐ Durable AI inference-efficiency step-change flattens power demand
Leading vs lagging indicators
- Leading: PPA signings, FERC interconnection rulings, restart milestones, capacity-auction guidance
- Lagging: EPS, share price
Key metrics
- Crane restart status (~$219M per $5/MWh) · PJM capacity price (Dec 2026) · uncontracted MWh signed + price
Valuation anchor
Premium (~27x fwd P/E); priced for execution. Consensus ~$368 (range $272–441).
The premium is the bear's lever.
Cross-arena sensors
B2 (PPAs/capacity/restart), B1 (AI demand + the efficiency risk), B7 (FERC/policy).
Posterior log
- 2026-07-15 · P 0.81→0.86 ↑ · UP · PJM capacity auction cleared at its price cap again — the exact opposite of CEG's break-trigger (two auctions <$200/MW-day). Fresh print confirms firm-power sca · https://heatmap.news/am/pjm-capacity-auction
- 2026-07-07 · P 0.76→0.81 ↑ · UP · Walmart signs its first-ever nuclear PPA — with Constellation, for Illinois ops — billed as among the first retailer/nuclear deals of its kind. A genuinely new · https://www.utilitydive.com/news/walmart-signs-nuclear-ppa-with-constellation-to-support-illinois-operations/824604/
- 2026-06-12 · P 0.79→0.76 ↓ · MAINTENANCE · Stable, tilting soft. The structural case (scarce firm nuclear into AI demand) is intact, but the last two upticks (0.78 on 5/27, 0.79 on 6/
- 2026-06-03 · P 0.74→0.79 ↑ · UP · Constellation's Three Mile Island (Crane) restart got a FERC waiver letting it transfer capacity interconnection rights, so the unit can deliver its full output · https://www.utilitydive.com/news/constellation-three-mile-island-crane-nuclear-ferc-waiver/821836/
- 2026-05-29 · P 0.78→0.74 ↓ · MAINTENANCE · Stable, but the last move up looks soft — the 2026-05-27 bump to 0.78 rode tangential narrative signals (Jefferies ESG, SK Submarines, Michi
- {{date}} · created · — · AI-Energy-Thesis-Scaffold