The existing US nuclear fleet is structurally undervalued relative to what AI-power demand will pay

Confidence: 8/10 Supporting evidence:

• Every major hyperscaler signed nuclear PPAs in 2024-2026 at estimated $80-120/MWh; operating cost of existing nuclear is $30-50/MWh — structural 30-70 yr margin expansion for 20+ years. Evidence: strong
• Constellation and Vistra (the two primary nuclear IPPs) are signing multi-decade PPAs that lock in earnings far above what prior merchant-power models priced in. Evidence: strong
• The existing fleet is permitted, built, connected, and low-carbon — none of these attributes can be replicated in under 10 years at comparable cost. Evidence: strong

Challenging evidence:

• Nuclear PPA prices are confidential; if they're below $60/MWh, the spread is narrower than assumed. Evidence: weak
• State/federal regulators could force cost sharing that limits operator margins. Evidence: low probability

Evolution:

• Jun 15, 2026 — Initial thesis at 8/10; driven by reported deal structures and Vistra Q1 EBITDA record

Depends on: conventional-nuclear, power-purchase-agreements, ipps-merchant-power Would change if: Published PPA prices reveal rates well below $60/MWh, or if FERC regulatory action materially caps merchant nuclear margins

---