2 Nuclear Stocks Worth Buying as AI and Geopolitics Drive Demand for Power

The article argues that AI data centers could drive major new demand for reliable power, with global electricity use from data centers projected to more than double to 945 TWh by 2030. It highlights BWX Technologies’ $8.6 billion backlog and ongoing SMR work with GE Vernova and Hitachi, while noting Oklo’s longer-dated upside from its Aurora microreactor but also its NRC approval risk and expected $80 million to $100 million of 2026 cash burn. Overall, the piece is constructive on nuclear-related stocks, but it is largely opinion-driven and not immediate operating news.

Analysis

The second-order winner is not simply the reactor OEMs, but the regulated, high-trust suppliers that sit inside the qualification bottleneck. BWXT is better positioned than the headline AI power names because its revenue stream is anchored by mission-critical defense work while the commercial nuclear optionality gives it a free call on any SMR acceleration; that combination should compress downside while preserving upside if AI power demand translates into procurement. The market may still be underestimating how much of the value accrues to components, fuel handling, controls, and maintenance rather than to first-of-a-kind reactor developers.

OKLO is a different trade: it is effectively a venture-style financing instrument tied to a policy/regulatory milestone stack. The near-term catalyst path is binary and slow, which means the stock can outperform on narrative but still underperform on mark-to-model until certification clarity improves; any meaningful re-rate likely requires either a credible DOE demo signal or an external customer anchor in the data-center space. The bigger competitive risk is that non-nuclear baseload substitutes like fuel cells, gas-plus-carbon capture, and behind-the-meter grid upgrades can fill the same pain point with materially less regulatory risk and faster deployment.

Contrarian view: the consensus is likely overpricing the inevitability of nuclear as the AI default. The grid constraint problem is real, but the fastest solution over the next 12-24 months is usually interconnection workarounds, demand management, and hybrid on-site generation, not new reactors. That means the most attractive exposure is probably not the pure-play developers, but the picks-and-shovels names with backlog, pricing power, and multi-year order visibility.

From a portfolio perspective, the cleanest expression is a barbell: own the stable supplier and fund it with a short in the speculative developer basket if the market gets too far ahead of approvals. If the AI power trade broadens, expect the first beneficiaries to be industrials with nuclear-qualified engineering and fabrication capacity, not the software beneficiaries people instinctively reach for.

Key Decisions for Investors

• Long BWXT on a 6-12 month horizon; favor on pullbacks because the defense anchor limits earnings drawdown while SMR optionality adds free upside if AI-driven utility demand turns into orders.
• Avoid chasing OKLO ahead of regulatory milestones; if you want exposure, use small size only and treat it as a binary catalyst trade with a willingness to cut on any delay in NRC or DOE timing.
• Pair trade: long BWXT / short OKLO for the next 3-6 months to express a view that the market is overpaying for development-stage optionality relative to cash-flowing nuclear infrastructure exposure.
• Watch GEV as a cleaner proxy for commercialization of modular nuclear supply chains; a dip into weakness is preferable to chasing, since it benefits from the same theme with less single-project risk.
• If BE rallies on the same AI power narrative, consider fading into strength versus BWXT: the market may be temporarily grouping all distributed power solutions together, but fuel-cell economics are easier to displace if grid access improves.