Bet on These Nuclear ETFs to Ride the AI Data Center Boom Now

AI data-center electricity consumption is projected to rise from 448 TWh in 2025 to 980 TWh by 2030, supporting a projected 44% increase in global nuclear capacity to as much as 535 GW by 2036. Corporate demand is becoming tangible: Meta has secured 6.6 GW of nuclear power, while Constellation plans to restore 835 MW at Three Mile Island Unit 1 by 2027 under a 20-year Microsoft agreement. The article highlights diversified nuclear ETFs—including NLR ($3.74B in assets), NUKZ ($715.3M), RCTR ($29.8M), and URAN—as vehicles to gain exposure to uranium, utilities, reactor technology, and nuclear services while reducing single-company risk.

Analysis

The investable implication is less about aggregate electricity demand than scarcity of deliverable, firm power inside constrained transmission regions. Existing nuclear operators CEG and VST can monetize this scarcity well before new generation is built through higher contracted power prices, capacity payments and data-center interconnection economics; regulated PEG has less upside because much of the benefit is ultimately shared with ratepayers. GEV is a quieter second-order beneficiary: grid equipment, gas turbines and services may capture data-center capex regardless of whether the marginal long-run electron comes from nuclear.

The market is likely over-attributing near-term AI power demand to SMR developers. OKLO's valuation remains principally a duration/security-selection bet on licensing, financing, fuel availability and construction execution, whereas CEG/VST own operating assets and can convert contracted pricing into EBITDA on a 1-3 year horizon. New nuclear capacity is a 6-18 year story; the nearer bottleneck is transmission and dispatchable capacity, which can favor gas-plus-grid spending and dilute the pure-nuclear narrative.

No broad ETF trade is warranted on this promotional, low-novelty item: NLR/NUKZ blend commodity-sensitive miners, regulated utilities and pre-revenue reactor developers with materially different catalysts, while smaller products can carry execution and liquidity costs. Over the next 1-3 months, the relevant confirmations are incremental hyperscaler PPAs, CEG/VST forward-power and capacity-price disclosures, and GEV order/backlog conversion. Falsify the incumbent-operator thesis if 2027-28 realized power-price guidance or contracted load growth disappoints, or if data-center project deferrals meaningfully reduce utility load forecasts; a sustained decline in regional capacity prices would be the earliest market signal.

Key Decisions for Investors

• Maintain a 6-12 month relative-value bias: long CEG versus short OKLO in matched beta/volatility terms. The trade owns contracted, operating-generation cash flow against licensing and financing duration risk; reassess if CEG's forward EBITDA/FCF outlook fails to rise with new load contracts or OKLO secures fully financed, binding offtake and a material licensing milestone.
• Accumulate GEV only on post-results pullbacks or after order/backlog evidence confirms data-center-driven grid spend; target a 12-18 month horizon. Use quarterly electrification backlog, service-margin progression and free-cash-flow conversion as gates rather than treating nuclear announcements as the catalyst.
• Avoid RCTR and URAN for institutional-sized deployment given limited reported trading liquidity; if diversified exposure is required, use NLR as the more liquid implementation but size it as a thematic satellite, not a direct AI-power proxy.
• Set a monitoring trigger around CEG and VST earnings: add exposure only if incremental contracted load is accompanied by higher long-dated realized-price or capacity-revenue guidance. Do not chase price moves on nonbinding MOUs, political statements, or announced reactor concepts without disclosed economics and completion dates.