2 Green Flags and 2 Red Flags for Nuclear Stocks After This Year's Sell-Off

U.S. electricity demand is projected to rise 60% between 2025 and 2045, versus just 10% over the prior 20 years, supporting a long-term case for always-on nuclear generation as AI and EV demand expands. NextEra's planned acquisition of Dominion would add nuclear scale and improve capital-market access, while Constellation has signed AI-related reactor agreements. However, SMR developers Oklo and NuScale remain loss-making and largely unproven, while nuclear-linked valuations remain elevated: Constellation trades at 25x earnings versus roughly 19x for utilities and Cameco at 157x.

Analysis

The investable bottleneck in the power-demand theme is not reactor design; it is deliverable, interconnectable generation in constrained load pockets. CEG has the cleanest near-term exposure because existing nuclear capacity can monetize higher capacity and bilateral power prices without construction risk, but its premium embeds continued data-center contracting. NEE/D creates a better risk-adjusted platform if the transaction closes: regulated rate-base capacity and a lower cost of capital can fund grid and generation additions that speculative SMR developers cannot.

OKLO and SMR remain duration-heavy financing trades, not earnings trades. Their equity value is highly sensitive to rates, dilution and commercial milestones; a favorable federal funding headline does not solve first-of-a-kind construction, fuel qualification, customer-credit or fixed-price EPC risk. Second-order beneficiaries are grid equipment suppliers such as ETN, PWR and GEV, where transmission upgrades and data-center interconnection spending can arrive years before incremental nuclear output.

Near term, sector flows may continue to punish high-beta nuclear names if long-end yields rise or AI-capex expectations soften. Over 1-3 months, signed power purchase agreements, utility commission approvals and the NEE/D financing/closing process matter more than broad nuclear sentiment. Over 6-18 months, the key falsifier for the incumbent-nuclear thesis is evidence that gas generation, storage and demand-response clear new data-center load at lower all-in cost, limiting nuclear's scarcity premium.

Contrarian view: the market may be assigning too much value to nuclear as the sole solution to firm power demand. Data-center operators optimize for time-to-power, and will often accept gas-backed or hybrid solutions if nuclear capacity is unavailable on their construction schedule. That favors diversified utilities and grid contractors over pure-play reactor developers.

Key Decisions for Investors

• Prefer a 6-18 month long NEE versus short OKLO pair: NEE offers regulated/contracted cash-flow support and financing scale, while OKLO remains exposed to milestone slippage and equity issuance. Reassess if NEE/D closing economics deteriorate materially or OKLO secures binding, financeable customer contracts with a credible construction timetable.
• Maintain CEG as a tactical long only on pullbacks or after independently verified long-duration power-contract announcements; avoid chasing multiple expansion. Upside comes from scarcity pricing, but trim if forward EBITDA/FCF guidance fails to rise with reported contracting activity.
• Build a basket long ETN and PWR over 6-18 months as a lower-technology-risk expression of load growth. Use project backlog, margin guidance and utility capex plans as validation; reduce if interconnection queues or hyperscaler capex plans show broad deferrals.
• Avoid fresh long exposure to SMR and OKLO absent visibility on cash runway, expected dilution and customer-backed project financing. Treat permitting, fuel-supply and EPC announcements without committed capital as trading catalysts rather than fundamental de-risking events.
• Monitor long-end Treasury yields and regional power-price forwards over the next 1-3 months: rising yields disproportionately pressure pre-revenue nuclear developers, while sustained tightening in power markets supports CEG and utility/grid-capex beneficiaries.