Barclays Outlines the Top Nuclear Energy Stocks Powering the AI Boom

Barclays says AI infrastructure spending could exceed $1 trillion annually before peaking in 2028, boosting nuclear and uranium names positioned to supply reliable baseload power. The article highlights constructive company-specific updates including earnings beats from Southern Company, Duke Energy, Public Service Enterprise Group and Talen Energy, along with favorable analyst actions on Constellation and Dominion Energy. Cameco also secured a conditional DOE commitment for up to $17.5 billion tied to Westinghouse reactors, reinforcing the sector’s AI-driven growth narrative.

Analysis

The market is beginning to re-rate nuclear not as a utility subsector, but as an AI-enabled capacity bottleneck play. The first-order beneficiaries are the owners of scarce, dispatchable baseload assets with existing interconnection and fuel security — that makes CEG, TLN, and the regulated nukes more attractive than pure-play reactor developers, which still depend on long-dated permitting and balance-sheet support. The second-order winner is the nuclear services/fuel chain: uranium miners, component suppliers, and EPC firms should see backlog convert before the market fully capitalizes the need for firm power.

The key mispricing is that hyperscaler capex optimism does not automatically accrue to every nuclear name equally. Names with merchant exposure or financing needs will see higher volatility if power prices lag infrastructure demand, while balance-sheet-sensitive projects remain hostage to rates and credit spreads. That creates a clear dispersion trade: ownership of existing assets with near-term cash flow should outperform speculative SMR names over the next 6-12 months unless there is a genuine policy breakthrough or major rate cut.

The contrarian risk is that the market is extrapolating AI load growth faster than grid and regulatory reality can absorb it. If data center demand slips from announced to funded capex, or if large load customers push behind-the-meter gas, battery, or demand-response solutions instead of utility nuclear, the multiple expansion could stall within one or two quarters. Also, nuclear supply-chain bottlenecks can backfire: if the market prices in too much future uranium tightness too quickly, the physical market may not validate the equity move until 2026-27.

From a relative-value standpoint, regulated utilities with nuclear exposure offer cleaner downside protection than the higher-beta nuclear complex, but the strongest convexity sits in CEG and TLN if the AI power narrative keeps gaining credibility. Upside is more likely to come from estimates revisions than from near-term capacity additions, so the trade should be built around fundamentals inflecting over several quarters rather than a one-week squeeze.

Key Decisions for Investors

• Long CEG vs short NEE for a 3-6 month pair: CEG has direct scarcity value in existing baseload nuclear, while NEE carries more rate/financing sensitivity and less immediate AI-load optionality.
• Long TLN into the next 1-2 earnings prints: the co-location model gives it the cleanest direct monetization path from AI load growth; target a 10-15% rerating if management confirms incremental data-center demand.
• Add CCJ on pullbacks for a 6-12 month trade: it is the cleaner way to express the fuel-cycle bottleneck without taking full reactor-construction execution risk; downside is better protected than in SMR developers.
• Avoid or short-basket NXE/OKLO/SMR against CEG/CCJ until financing visibility improves: these names are more dependent on capital markets and policy catalysts than on actual power delivery, making them vulnerable if rates stay sticky.
• Tactically buy SO or DUK as lower-volatility hedges against a broader nuclear basket: if the AI power thesis proves real but sentiment reverses, regulated cash flows should preserve capital better than high-beta pure plays.