The Nuclear Renaissance Explained: 3 Ways to Invest

Goldman Sachs expects data-center power demand to rise 175% by 2030 from 2023 levels, supporting a potential nuclear-energy renaissance driven by AI infrastructure needs. The article highlights three investment routes: SMR developers NuScale and Oklo, nuclear-focused utility Constellation Energy, and uranium ETFs including Roundhill Uranium ETF (0.77% expense ratio) and Global X Uranium ETF (0.69% expense ratio; more than $5 billion in assets). Constellation's existing Big Tech power agreements and restart of nuclear facilities underscore utilities' direct exposure to rising demand, while SMR companies offer higher-risk, higher-upside exposure.

Analysis

The investable bottleneck is not simply reactor capacity; it is deliverable, firm power inside constrained data-center load pockets. CEG has near-term earnings optionality because existing nuclear output can be repriced through new power contracts well before new-build capacity arrives, while MSFT and peers face higher all-in power costs and potentially slower campus commissioning where transmission queues—not generation—are binding. Grid equipment and transmission beneficiaries (ETN, PWR, GEV) may capture a larger and earlier share of AI-power capex than SMR developers.

OKLO and SMR are effectively long-duration regulatory, construction, and financing options rather than near-term AI revenue exposures. Their valuations require customer commitments to convert into bankable PPAs, site permits, fuel availability, and project-finance terms; any one slippage can materially dilute equity holders before commercial output. The 1-3 month catalyst path is contract announcements, NRC milestones, and data-center power procurement disclosures, but the meaningful cash-flow test is 6-18 months away and should be judged by funded backlog rather than memoranda or nonbinding demand claims.

Consensus may be overpaying for the direct AI-to-SMR narrative while underweighting the value of operating assets and fuel-cycle security. Uranium equities add substantial country, operating, and equity-beta exposure, so URA is not a clean uranium-price hedge; CCJ or a physical-uranium vehicle offers more targeted exposure if contracting tightens. Conversely, if AI capex moderates or hyperscalers favor gas-backed generation as the fastest bridge solution, nuclear developers de-rate first while established generators retain broader load-growth support.

The key falsifiers are: CEG failing to secure incremental contracted prices above its replacement-power economics; SMR/OKLO failing to report fully funded customer projects and permitting progress; or evidence that data-center energization schedules are being delayed by transmission rather than generation. A sustained decline in power forwards in key PJM/ERCOT nodes would also weaken the near-term scarcity thesis.

Key Decisions for Investors

• Prefer a 6-12 month long CEG / short equal-dollar basket of OKLO and SMR as a quality-duration pair: CEG monetizes existing assets and contracted-load scarcity, whereas the short leg is exposed to permitting, funding, and dilution risk. Reassess if either developer announces a binding, fully financed project with a credible commercial-operation date.
• Build a 3-9 month watch position in ETN and PWR rather than chase SMR beta; prioritize entries after broad AI-power enthusiasm pulls back. The risk/reward improves if utility interconnection backlogs and transmission awards continue to rise, while a material slowdown in hyperscaler capex is the primary stop condition.
• For uranium exposure, favor a measured CCJ or URNM allocation over URA when uranium term-contracting data strengthen; use a 12-18 month horizon and cap sizing because miners remain leveraged to operational execution and commodity volatility. Avoid treating uranium ETFs as substitutes for physical uranium.
• Do not add MSFT solely on this theme: higher power procurement costs are a modest margin headwind relative to its AI revenue opportunity. Monitor whether power availability delays data-center capacity additions; that is a more relevant earnings risk than the absolute electricity price.