You are human visitor number — on this page

Language · ภาษา

Utilities

CEG Constellation Energy

Nuclear utility; thesis orthogonal.

Neutral Rank 35 · Nasdaq-100 constituent

Last price

$296.21

Market cap

$107.3B

As of

18 April 2026

Live quote sourced from Yahoo Finance. Prices cited in narrative below reflect the original memo date and may be stale.

Scores · adapted framework

Enabler

2 / 10

Autopilot adoption

1 / 10

Disruption risk

0 / 10

Efficiency upside

2 / 10

The Sequoia matrix

Intelligence / Judgment

Intelligence-leaningGrid optimization is pattern-recognition heavy; safety judgment remains critical.

Copilot posture

EmergingPredictive maintenance and grid monitoring are emerging; not primary product.

Autopilot posture

MinimalSome autonomous grid functions; not outcome-based services.

Data moat

StrongPlant operations and grid data; proprietary to utility operations.

Execution layer

LimitedExecution is plant and grid operations; no external services layer.

The memo

State of play · CEG

Trading ~$296 in mid-April 2026. Market cap ~$70B. Q4 FY25 revenue $19.8B (+8% YoY); ~$6B net income. On March 31, 2026: Announced accelerated restart of Three Mile Island Unit 1 for Microsoft, with long-term power-purchase agreement. Amazon-Talen Energy deal also live. Next earnings: mid-May 2026.

Thesis angle

Constellation Energy operates nuclear and renewable power plants. The business is utility operations and power generation—not software or services. AI may optimize grid scheduling or plant maintenance, but these are marginal. Regulatory economics dominate; thesis does not apply.

The framing

CEG is the single utility in the Nasdaq-100 that is materially a Sequoia-thesis play—and it is already monetizing it. Nuclear power for AI hyperscalers is a direct, contracted outcome: the company sells electricity to cloud platforms at premium rates, bundled with carbon-free operations. This is not regulatory-capture; it is outcome-priced generation.

Two forces, opposite directions

Tailwind · AI hyperscaler electricity demand at outcome-scale

Microsoft-Three Mile Island long-term PPA (March 2026): 835 MW baseload nuclear
Amazon-Talen Energy: 960 MW data-center dedicated capacity deal
Hyperscalers explicitly bundling carbon-free compute into AI product offers
Premium pricing: ~2-3x spot rates for reliability and decarbonization
No commodity pricing: all megawatts are outcome-contracted to compute outcomes

CEG is not selling kWh to the grid; it is selling 24/7 compute-enabling power at outcome-linked rates.

Headwind · Regulatory approval risk and execution complexity

TMI-1 restart requires NRC approval, permitting, and construction (timeline: 2028)
Hyperscaler load is lumpy: one customer cancellation is material risk
Political risk: future administrations may restrict datacenter-nuclear partnerships
Legacy utility operations remain regulated; not all of CEG is outcome-priced

Scale-up from ~15% outcome-priced generation to material mix is multi-year and execution-dependent.

CEG portfolio: regulated vs. outcome-priced generation

Asset Class	MW Capacity	Economics	Thesis Fit
Regulated utility gen	~28,000	Rate-base returns (~7%)	Orthogonal
Three Mile Island-1 (Microsoft)	835	Long-term outcome PPA	Strong thesis play
Talen-Amazon dedicated	960	Compute-outcome pricing	Strong thesis play
Other hyperscaler contracts	~2,000	Premium PPAs	Strong thesis play

CEG is evolving from pure regulated utility to hybrid: legacy rate-base (~75% of mix), plus emerging outcome-priced generation (~25% mix target by 2028).

Bull case

Three Mile Island-1 restart is the cleanest Sequoia autopilot-play in the entire utilities cluster.

CEG builds and operates the plant; Microsoft pays outcome-linked rates for AI compute enablement. This is outcome-pricing of infrastructure services.

Hyperscaler capex dedications are multi-decade contracts; not cyclical.

AI development cycles and training require continuous power; Microsoft and Amazon are locking in 10-20 year terms.

Nuclear baseload is the only 24/7 carbon-free generation at scale.

Hyperscalers have committed to carbon-neutral AI compute; CEG is the sole qualified supplier of large baseload capacity. Competitive moat.

Margin expansion from outcome-pricing is structural.

Rate-base generation earns 7%; outcome-priced nuclear earns ~12-15%. Portfolio mix shift drives ROIC expansion.

Bear case

TMI-1 restart is not assured; regulatory and execution risk is real.

NRC approval, environmental review, and construction cost overruns are all live risks. 2028 start-up is optimistic.

Customer concentration is an existential risk.

If Microsoft cancels or delays datacenter buildout, CEG's outcome-revenue drops materially. This is not portfolio diversification.

Political opposition to nuclear may resurface.

Future administrations or state legislators could challenge nuclear-subsidies narrative or carbon-credit valuations.

Stranded asset risk in legacy regulated generation.

Coal plants are being retired; baseload fossil generation is structurally declining.

Sequoia-framework fit

CEG is the single utility uniquely positioned as a Sequoia-thesis beneficiary. The company is pivoting from regulated-utility economics toward outcome-priced generation for hyperscaler AI compute. This is not services-as-software in the traditional sense—it is outcome-based infrastructure pricing. CEG earns premium margins by selling AI-enabling power at compute-outcome rates rather than commodity-grid rates. Execution risk is real (TMI-1 permitting, customer concentration), but the strategic thesis is sound and already monetized.

Investor takeaway

Upgraded framing: CEG is a thesis-aligned outcome-pricing play, not an orthogonal utility.

· · ·