Source Report

Crane Clean Energy Center (TMI Unit 1) Restart and Microsoft PPA

Constellation Energy transformed a 2019 economic shutdown into a hyperscaler-backed revival by signing a 20-year virtual PPA with Microsoft for the full ~835 MW output of Three Mile Island Unit 1 (rebranded Crane Clean Energy Center after late CEO Chris Crane), where Microsoft buys energy, capacity, and carbon-free attributes to match its PJM-region data center consumption—mechanistically, this dedicates the plant's dispatchable baseload to Microsoft's 24/7 carbon-free matching goals, bypassing intermittent renewables' intermittency via virtual offsite delivery while Constellation handles grid injection.[1]
- Deal announced September 20, 2024; expected online 2028 (accelerated to H2 2027 target amid progress), producing ~7 million MWh/year at 95%+ capacity factor.[2][3]
- $1.6 billion capex (cash from operations; $1 billion DOE loan guarantee closed November 2025, maturing 2055 at Treasury +37.5 bps).[4][5]
- Regulatory: NRC name change approved; full safety/environmental review, license renewal to 2054 pending (reviews targeted end-2027); FERC/PJM interconnection via Eddystone capacity transfer requested (PJM warns potential 2031 delay, monitor opposes waivers as of April 2026).[6][7]
- Economics: ~$785 million annual revenue implied (~$100-102/MWh fixed-price est. per analysts); 3,400 jobs, $3+ billion taxes, $16 billion PA GDP add over life.[8]

Implication for competitors/entrants: This proves restarts yield 2-3x merchant nuclear margins via long-term fixed premiums, but demands NRC/FERC/PJM navigation—new builds face 10+ year timelines vs. restarts' 3-4 years, favoring incumbents like CEG with existing sites/fuel.

Meta's Clinton Clean Energy Center PPA

Meta's 20-year virtual PPA locks in Clinton's full 1,121 MW (post-30 MW uprate) starting June 2027, replacing expiring Illinois ZEC subsidies with market-based hyperscaler revenue—Constellation sells output (energy/attributes) to the MISO grid while Meta claims carbon-free credits for regional data centers, enabling relicensing to ~2046+ without ratepayer aid and opening SMR site potential.[9]
- Announced June 3, 2025; preserves 1,100 jobs, $13.5 million annual taxes, $1 million charity over 5 years; avoids 34 million metric tons CO2 if closed.[10]
- No explicit pricing (undisclosed); supports uprates/NRC early site permit extension for advanced reactors.[9]

Implication for competitors/entrants: ZEC-like subsidies' end creates urgency; hyperscalers' willingness to back 20-year deals reprices at-risk nuclear at 50-100% merchant premiums (~$70-100/MWh est.), but requires 90%+ capacity factors—gas peakers or solar can't replicate 24/7 dispatch.

Broader Hyperscaler Nuclear PPAs Involving CEG

CEG lacks direct PPAs with AWS/Google (AWS/Talen Susquehanna co-location ~960 MW phased; Google/Kairos SMRs), but post-Calpine acquisition (closed Jan 2026, +23 GW gas/nuclear), added CyrusOne 380 MW (expandable) at Texas Freestone for data centers—packaging generation/grid access as "behind-the-meter" alternatives to pure nuclear.[11]
- U.S. gov't contract won for nuclear supply; fleet uprates target 1,000+ MW.[12]
- No CEG-AWS/Google nuclear specifics; focus remains Microsoft/Meta as templates.

Implication for competitors/entrants: CEG's 21-reactor fleet (largest U.S.) + Calpine scales to 50+ GW, capturing 10-20% data center load growth—but FERC co-location risks (e.g., Talen/AWS rejection) push virtual PPAs, favoring CEG's grid-integrated model over pure behind-meter.

AI Data Centers Repricing Nuclear via 24/7 Carbon-Free Procurement

AI hyperscalers shifted from annual RECs to 24/7 hourly carbon-free matching (Google/Microsoft 2030 targets), repricing nuclear's ~95% capacity factor at $100+/MWh premiums vs. $40-55/MWh merchant/PTC floors—mechanism: PPAs bundle energy/capacity/attributes, insulating from intermittency (solar ~25% CF) while guaranteeing dispatch for 100+ MW data loads, turning stranded nuclear into $700 million+/year cash cows per plant.[13][14]
- Demand: Data centers +35 GW by 2030; nuclear <10% today but rising (e.g., CEG PPAs ~2 GW).[15]
- Repricing: Microsoft Crane est. $100-102/MWh; Talen/AWS ~$70/MWh; IRA 45U PTC ($40-43.75/MWh inflation-adjusted) as floor enables 2x EBITDA vs. gas.[8]

Implication for competitors/entrants: 24/7 mandates create nuclear moat—new SMRs (Google/Kairos) 2030+; restarts/uprates win near-term, but supply scarcity bids prices to $100-150/MWh, compressing gas/solar valuations 30-50%.

CEG Earnings Visibility and Valuation Synthesis

Microsoft/Meta PPAs + gov't deals + Calpine lock ~13-20% base EPS CAGR to 2030 ($11-12/share 2026 guide), shifting 70% merchant exposure to contracted (PJM auctions/clears provide bridge)—long-dated revenue de-risks weather/price volatility, compounding via 45U PTC/fleet uprates while $3.9 billion 2026 capex yields 94%+ CF.[16][12]
- Visibility: Crane/Meta ~$1.5 billion annual by 2028; Calpine +$2 billion FCF; dividend +10% 2026.
- Valuation: Trades ~22-26x 2026 EPS, 18-20x EV/EBITDA (premium to utilities' 12-15x); analysts ~$370-400 targets (20-30% upside), but 40x trailing reflects AI froth—deals justify 25-30x if 20% growth holds.[17][18]

Implication for competitors/entrants: PPAs elevate CEG to "AI toll-road" (tech-like 25x+ multiple vs. utility 15x), but execution risks (PJM delays, capex overruns) cap rerating—rivals need 5+ GW nuclear to compete, else trade as gas/solar proxies at 10-12x.

Sources:
- https://www.constellationenergy.com/news/2024/Constellation-to-Launch-Crane-Clean-Energy-Center-Restoring-Jobs-and-Carbon-Free-Power-to-The-Grid.html [web:64,169]
- https://investors.constellationenergy.com/news-releases/news-release-details/constellation-meta-sign-20-year-deal-clean-reliable-nuclear [web:109,170]
- https://www.constellationenergy.com/news/2025/constellation-meta-sign-20-year-deal-for-clean-reliable-nuclear-energy-in-illinois.html [web:110]
- https://www.nrc.gov/info-finder/reactors/ccec [web:19]
- https://www.utilitydive.com/news/pjm-market-monitor-constellations-nuclear-crane-waiver/818216 [web:99]
- https://finance.yahoo.com/news/constellation-energy-stock-target-lifted-on-microsoft-ppa-deal-93CH-3629727 [web:154]
- https://seekingalpha.com/article/4894827-constellation-energy-riding-nuclear-demand-and-ai-power-boom [web:84]
- https://finance.yahoo.com/quote/CEG/key-statistics [web:139]
- Additional: [web:50,66,67,83,131,152] for estimates/context. Data as of April 2026; no direct pricing in official releases (analyst inferences marked). Confidence: High on facts/timelines (official/NRC); medium on pricing/revenue (estimates). Further SEC 10-Qs strengthen.

---

Recent Findings Supplement (April 2026)

Crane Clean Energy Center (Former Three Mile Island Unit 1) Progress

Constellation secured a $1 billion DOE loan on November 18, 2025, closing simultaneously with commitment—the first for LPO—slashing restart financing costs to $1.6 billion total ($1,916/kW) via interest-bearing funds under the Energy Dominance program, enabling 2027 online despite prior 2028 target; this leverages Microsoft's 20-year PPA for full 835 MW output (energy, capacity, RECs) to match PJM data centers hourly, proving hyperscalers' willingness to pay premiums for nuclear's 94%+ capacity factor over intermittent renewables.[1][2][3]
- DOE loan powers turbine/generator restoration, controls, cooling; >80% staffed (500+ workers); NRC inspections on schedule per Feb 2026 reports.[4][5]
- NRC reviewing 3+ license amendments (e.g., tech specs, security, emergency plan) submitted Jul-Oct 2025; Federal Register notice Feb 24, 2026 opened hearings (deadline Apr 27); SAFSTOR status, restart panel active, no major issues; name changed to Christopher M. Crane Clean Energy Center May 2025.[6][5]
- FERC filing Mar 31, 2026 (ER26-2028) seeks waivers to transfer 760 MW Capacity Interconnection Rights from Eddystone Units 3/4 (under DOE emergency order) to Crane, bypassing PJM's 2031 queue; PJM neutral, monitor opposes but FERC approval likely by Jun 1 for 2027 auction bid.[7][8]
Competition Implication: New entrants face 5-10 year queues/barriers; CEG's data moat (real-time merchant sales visibility) + federal backstop locks hyperscaler deals, deterring pure-plays without existing fleet.

Calpine Acquisition Closes, Unlocking Data Center Scale

CEG completed $22 billion Calpine buy (50M shares + $4.5B cash) on Jan 7, 2026—originally $26.6B announced 2025—merging 23 GW nuclear with 26-32 GW gas/geothermal for 55-60 GW total, creating U.S.'s top private producer; mechanism: Calpine's "Powered Land" co-location delivers on-site power/grid/site infra to data centers, pairing nuclear baseload with gas flexibility for 24/7 dispatchable CF energy, sidestepping transmission delays while monetizing underused sites at premium pricing ($80-100+/MWh inferred from peers).[9]
- Enables >10 GW contracted (Microsoft/Meta/CyrusOne); post-close, Calpine inked 380 MW Freestone (TX) deal + 380 MW Phase 2 exclusive with CyrusOne (total >1.1 GW TX data centers), atop 400 MW Thad Hill.[10][11]
- Synergies + capex ($3.9B 2026) target 20% EPS growth; $5B buyback boosted Feb 2026.[11]
Competition Implication: Rivals like Vistra/Talen lack CEG's nuclear scale (22 GW); entrants need $2,600+/kW new-build gas vs. Calpine's $1,142/kW acquired, but integration risks (debt up) cap aggressive bidding.

Hyperscaler PPAs Expand Beyond Microsoft, But No New CEG Nuclear Ties

Meta's Clinton (IL) nuclear deal referenced in analyst notes (1.1 GW, 20-yr starting 2027 + uprate/SMR option) confirms prior structure, but no post-Oct 2025 CEG-specific updates; CyrusOne (post-Calpine) adds non-nuclear data center power (760+ MW TX), signaling hyperscalers' hybrid tolerance (gas bridge to nuclear) amid 24/7 CFE mandates—nuclear's always-on (no hourly mismatches like solar/wind) commands 20-50% premiums, repricing merchant assets 2-3x vs. 2024.[12][10]
- No new AWS/Google/CEG nuclear PPAs confirmed; Amazon eyes Calvert Cliffs site (2k acres), but exploratory.[13]
- AI demand quadruples data center TWh to 1,600 by 2034; nuclear PPAs now operational necessity (e.g., MSFT TMI template).[14]
Competition Implication: Non-nuclear (gas) bridges buy time, but CFE goals favor CEG's 94.7% nuclear CF; others must partner or face 10-yr SMR delays.

Earnings Visibility Boosted by PPAs, Guidance Affirms AI Tailwinds

Q4/2025 adjusted EPS $2.30 (beat $2.20), full-year $9.39/$7.40 GAAP; 2026 guidance $11-12/share (Mar 31 call, ~25% growth, below $11.75 est.), assumes 2% inflation + PTC; Calpine adds >$2/share, >10 GW deals ensure revenue certainty into 2030s, derisking 40% volatile earnings.[11][15]
- 10% dividend hikes 2026-27; $2.32B 2025 net income; analysts trim PT $370-400 (34x P/E), but 13%+ CAGR to 2030.[16][17]
Competition Implication: PPAs lift multiple 25-27x fwd (vs. utilities 15-20x); new entrants trade pre-revenue discounts, needing CEG-scale contracts for re-rating.

AI Demand Structurally Reprices Nuclear at 24/7 CFE Premiums

Hyperscalers' 24/7 CFE (hourly matching) exposes renewables' 20-30% CF limits, forcing nuclear/gas premiums ($100+/MWh bull case); CEG's fleet (94.7% CF) reprices via PPAs tying output to data loads, auto-deducting payments like Shopify lending—defaults near-zero, revenues inflation-linked (PTC to $56/MWh at 3.5%).[18]
- Demand: 10 GW+ CEG deals; hyperscalers $635-665B 2026 capex, but grid queues force restarts/co-location.[19]
Competition Implication: Data moat = product; pure developers (Oklo/NuScale) -50%+ YTD, pre-commercial; CEG's 5% U.S. nuclear share compounds FCF $2B+/yr.