Nuclear power has become the default answer to AI's insatiable appetite for electricity, and Wall Street is betting on it through three competing exchange-traded funds that each capture a different slice of the nuclear supply chain. The Range Nuclear Renaissance Index ETF (NUKZ), the Global X Uranium ETF (URA), and the VanEck Uranium and Nuclear ETF (NLR) all ride the same thesis: AI needs round-the-clock power, reactors deliver it, so buy nuclear. Yet their one-year returns have already spread from 37% to 62%, revealing that the market is still figuring out which part of the nuclear chain will tighten first.

Why Is Nuclear Power Suddenly Critical for AI Data Centers?

The demand numbers tell the story. According to the Department of Energy's report on data center power use, data centers consumed about 4.4% of total US electricity in 2023 and could reach somewhere between 6.7% and 12% by 2028. A single hyperscale facility can pull more than a gigawatt, roughly the draw of 750,000 homes, and it runs that load around the clock. That steady, always-on profile is exactly what nuclear is built to serve. S&P Global's 451 Research forecasts that data center grid demand will rise by nearly three times by 2030.

The deals are already on paper. Microsoft contracted to take the full output of the restarting Three Mile Island Unit 1 over a 20-year term, with Constellation Energy securing a federal loan to fund the restart. Amazon expanded its arrangement with Talen Energy at the Susquehanna plant in Pennsylvania, locking in up to 1,920 megawatts of carbon-free nuclear power through 2042. These are not theoretical bets; they are binding commitments from the world's largest technology companies.

The pressure is also hitting household budgets. The Energy Information Administration's outlook for residential power prices sees the average residential rate near 18.2 cents per kilowatt-hour in 2026, a rise of close to 5%, and regulators in load-heavy regions are leaning toward firm, dispatchable generation. State utilities are now under scrutiny over record customer complaints about soaring electricity bills.

How Do These Three Nuclear ETFs Differ in Their Approach?

Each fund targets a different link in the nuclear supply chain, and understanding those differences is key to understanding why their returns have diverged so sharply.

• NUKZ (Reactor and SMR Builders): The youngest of the three, NUKZ screens for companies whose revenue ties to constructing, operating, and servicing reactors, including small modular reactor (SMR) developers, fuel fabricators, and engineering firms. Top weights cluster around Cameco, GE Vernova, Talen Energy, Constellation Energy, and BWX Technologies. The fund trades near $72, with a roughly 14% gain year to date and about 42% over the past 12 months. This is the most concentrated way to bet that hyperscaler contracts and modular reactors drive the next leg, but concentration risk is the price of that focus.
• URA (Uranium Miners and Fuel): URA owns the upstream end of the chain, tracking the Solactive Global Uranium and Nuclear Components index, with weight concentrated in pure-play miners and physical-uranium holdings. Leading positions run heavily toward Cameco, with NexGen Energy, Oklo, Uranium Energy Corp, and Sprott Physical Uranium Trust filling out the book. URA is up about 19% year to date and 62% over 12 months, with a 184% return over five years and 416% over ten years, the strongest long record of the three. This leverage cuts both ways; when uranium spot prices correct, miners typically fall further than utilities.
• NLR (Utilities and Grid): NLR blends regulated nuclear utilities with miners and a smaller slice of SMR developers across roughly 28 holdings, giving it a cash-flow profile closer to a utility fund than a commodity vehicle. Top names center on Constellation Energy, Cameco, BWX Technologies, Public Service Enterprise Group, and Oklo. It trades near $133, up about 7% year to date and 37% over 12 months, with a ten-year return near 270%. The returns are lower than its peers because regulated utilities cannot rerate as aggressively as miners or pure-play developers, but they carry rate-base earnings and steadier dividends.

The appeal of NLR is durability. Utilities that sign power purchase agreements directly with hyperscalers capture margin every time a data center plugs in, whether or not new reactors come online on schedule. If the trade compresses back into established operators, NLR holds up better than either alternative.

What Are the Key Risks Facing the Nuclear-AI Power Bet?

The demand thesis is durable, but the path to monetizing it through any one fund is not. Several pressure points sit under the trade right now. NUKZ faces concentration risk because a short holdings list means one weak quarter from a top name can drag the whole fund. URA shows volatile, high-beta exposure tied tightly to uranium spot pricing and prone to sharper swings in both directions. All three funds depend on regulatory approval and construction timelines that have historically slipped.

Oklo Inc., a pre-revenue nuclear power company with a market cap of around $12 billion, illustrates the execution risk. The company is developing small-scale nuclear reactors called Aurora to power data centers, with plans to deliver 15 to 75 megawatts of carbon-free, baseload power per unit. In late 2024, Oklo inked an important deal with data center operator Switch, which could reach 12 gigawatts over the next 20 years. In January, Oklo reached an agreement with Meta to power a campus in Ohio, aiming to deliver around 1.2 gigawatts by 2030. These are not yet binding contracts, but they show promise if the company can secure licensing for its systems.

For any nuclear company, the major hurdle is regulatory approval. The U.S. Nuclear Regulatory Commission (NRC) process is long, expensive, and unpredictable. Oklo's first application was denied in 2022. Today, the company is pursuing a faster Department of Energy (DOE) authorization pathway for its first plant at the Idaho National Laboratory, which allows construction to begin on a pilot version while the longer commercial NRC license proceeds in parallel. Even with some regulatory progress, execution risk remains high as novel nuclear power projects are rarely completed on time or on budget. Oklo has never generated revenue and does not expect to until at least 2028.

How Should Investors Think About These Three Competing Bets?

The choice comes down to which part of the thesis an investor finds most credible, and the three views do not have to be mutually exclusive. Pairing them adds breadth without much overlap. NUKZ and URA together cover infrastructure and fuel while underweighting regulated utilities; NLR and URA capture utilities and miners but skip most of the SMR and engineering names NUKZ holds. The funds were built to complement, not duplicate.

None of this is a clean ride, and the same features that make each fund attractive also mark where it can crack. The demand thesis is durable; the path to monetizing it through any one fund is not. Investors betting on nuclear power to solve AI's energy crisis are essentially betting on three different timelines: whether reactor builders will deliver on schedule, whether uranium miners can scale supply fast enough, or whether established utilities can capture enough margin from hyperscaler contracts to justify their valuations. The market's diverging returns suggest that no single answer has yet emerged.