This Unstoppable Trend Could Mint the Industrial Sector's Next Big Winners

EPRI projects U.S. data centers’ electricity use could rise from ~5% today to ~20% by 2035, reinforcing the energy bottleneck for AI compute. The article highlights a nuclear build-out target of tripling global capacity by 2050 (U.S. aiming to quadruple), with specific policy and funding tailwinds including a $400M DOE grant for the BWRX-300 via TVA and $900M to Centrus to expand HALEU enrichment by 2029. It frames nuclear fuel, reactor equipment (e.g., Cameco, GE Vernova), and HALEU supply (Centrus Energy) as beneficiaries, while emphasizing higher-risk, pre-revenue SMR developers (Oklo, NuScale).

Analysis

The cleanest expression of this theme is not the reactor developers; it is the scarce-input bottleneck. LEU has the strongest near-term pricing power because enrichment/HALEU capacity is the gating item for next-gen reactors, so the market can re-rate it on contract visibility well before any meaningful reactor FCF appears. CCJ is the lower-volatility way to own the same macro, but its upside is more dependent on sustained term-uranium discipline than on headline enthusiasm.

GEV is a second-order winner, but mostly because power customers will pay up for anything that can shorten the path to reliable electrons. The nuclear optionality is a long-dated call embedded in an already-strong power franchise, so the stock likely captures more of the secular capex budget than the eventual SMR economics. META is interesting mainly as a buyer of power certainty: if nuclear becomes the acceptable decarbonized baseload for AI campuses, hyperscalers may accept higher upfront power costs to secure capacity, which is supportive for the largest balance sheets and negative for smaller peers that cannot pre-fund their own supply.

The contrarian risk is that investors are extrapolating 2030s capacity additions into 2026 equity returns. For OKLO and SMR, the main falsifier is still execution: permits, financing, and signed utility/offtake contracts over the next 6-18 months, not the policy narrative. If uranium spot softens or DOE-backed projects slip, the trade likely migrates from "nuclear winners" back to a narrower fuel-scarcity trade, and the pre-revenue names can de-rate fast.

Key Decisions for Investors

• Long LEU on pullbacks for a 3-6 month trade: best risk/reward on the theme because scarcity of HALEU capacity can reprice on contract wins before the market sees revenue inflection. Falsify if enrichment ramp milestones slip or no commercial offtake emerges by the next two quarterly updates.
• Long CCJ as a higher-quality uranium beta versus pre-revenue reactor names; use it as the core 6-12 month expression. Add only if uranium term pricing stays firm; if spot rolls over materially, trim because the valuation case weakens quickly.
• Pair trade: long LEU / short SMR for 1-3 months. This captures the supply-constrained part of the cycle while fading duration risk in pre-revenue SMR developers; the risk is a surprise utility contract or financing package for SMR that re-prices the short leg.
• Buy GEV selectively on weakness as a lower-beta beneficiary of AI power capex, but size smaller than LEU/CCJ. Treat it as a power-infrastructure hedge, not a pure nuclear trade; thesis weakens if order growth is driven entirely by gas turbines without any incremental nuclear licensing progress.
• Avoid chasing OKLO on momentum unless there is a new signed offtake or financing milestone. Without that, the stock is mostly a sentiment vehicle; the overhang is that any delay in permitting or commercialization can compress multiple sharply.