Hyperscalers building out AI infrastructure have run into a hard constraint: electricity. Training clusters and inference farms need round-the-clock baseload power, and grids in Virginia, Texas, and the Pacific Northwest are already strained. That has pushed Microsoft, Amazon, and Google toward direct purchase agreements for nuclear power with utilities and small modular reactor developers, with all three having signed such agreements over the past two years. The three exchange-traded funds most directly exposed to that buildout are the Sprott Uranium Miners ETF (NYSEARCA:URNM), the Range Nuclear Renaissance Index ETF (NASDAQ:NUKZ), and the Themes Uranium & Nuclear ETF (NASDAQ:URAN).
Uranium exposure comes in three distinct flavors across these funds. URNM focuses almost entirely on pure uranium miners and physical uranium, giving it the most concentrated link to the commodity. NUKZ takes a wider view of the sector by spreading its holdings across the full nuclear value chain, including reactor operators and SMR developers. URAN offers a quieter blended approach that mixes miners with broader nuclear‑related names.
Recent performance has moved in the same direction but at different speeds. URNM has climbed roughly 89% over the past year, NUKZ has gained about 73%, and URAN has advanced around 65%, all of which have benefited from the surge in AI‑driven power demand.
Why nuclear became the AI power story
Data center load growth is the force that broke the grid‑planning assumptions of the past decade. Solar and wind can add capacity, but they cannot provide the firm, round‑the‑clock supply that AI training demands without storage costs that wipe out the economics. Natural gas can meet the load, yet it brings carbon obligations that clash with the net‑zero commitments hyperscalers have already made to shareholders. Nuclear ends up as the only scalable zero‑carbon baseload option, which explains why operators have begun restarting retired reactors and signing long‑dated offtake agreements with cloud providers. The investment case for the three funds below flows directly from that supply‑demand setup.
URNM: the concentrated uranium miner play
URNM is the most direct way to express a view that uranium prices keep rising as utilities and hyperscalers compete for fuel. The fund tracks the North Shore Sprott Uranium Miners Index and concentrates capital in companies whose earnings move with the spot and term uranium prices rather than with broader nuclear services revenue.
The portfolio leans heavily on a small group of names. Cameco sits at the top at about 21% of net assets, followed by the Sprott Physical Uranium Trust at roughly 14% and NexGen Energy near 13%. The top three holdings make up about 47% of the fund. The physical uranium trust position gives URNM something most peers do not: a slug of direct commodity exposure that responds to spot price moves without the operational risk of a single mine.
Geographic spread spans North American producers such as Uranium Energy, Denison, and Energy Fuels, Australia’s Paladin, UK-listed Yellow Cake, and Kazakhstan’s Kazatomprom. The expense ratio is about 0.8%, on the higher end for thematic ETFs but standard for a specialized commodity miner fund.
The trade-off with URNM is the risk of concentration in both names and themes. A weak quarter from Cameco or a uranium price pullback hits the fund harder than it would a diversified nuclear vehicle. Year-to-date, URNM has gained about 18%, and the five-year return is near 130%, capturing both the upside and the volatility associated with uranium-linked equities.
NUKZ: the broader nuclear renaissance vehicle
NUKZ is the fund that most closely aligns with the AI data center thesis. Where URNM bets on uranium fuel pricing, NUKZ holds companies that actually sell power to hyperscalers, build next-generation reactors, and supply engineering services for new construction. The portfolio combines reactor-operating utilities, advanced-reactor and small-modular-reactor developers, uranium miners, and nuclear construction and services firms.
That mix matters because the AI nuclear story spans fuel demand, utilities like Constellation negotiating long-term offtake contracts, about SMR developers like Oklo and NuScale pursuing first commercial deployments, and about engineering firms positioned to capture reactor restart and new-build work. NUKZ holds across those layers in a single product, which means a single hyperscaler PPA announcement tends to lift multiple holdings at once rather than concentrate gains in the miner side of the trade.
Performance over the past year reflects that breadth. NUKZ has returned about 15% year to date and 73% over the past year, with a smoother trajectory than URNM because utility holdings dampen the swings tied to uranium spot moves. The tradeoff is the inverse of URNM’s: investors give up direct fuel-price leverage in exchange for diversified exposure to the buildout. If uranium spikes sharply, URNM captures more of it. If the story plays out as a multi-year capex cycle across utilities and reactor builders, NUKZ captures more of that.
URAN: the lower-cost blended option
URAN is the overlooked option in this category. Themes ETFs, the issuer behind the fund, has built its product line around competing on cost against established thematic peers, and URAN follows that pattern. The fund holds a global mix of uranium miners and nuclear-energy companies, which puts its exposure profile somewhere between URNM’s pure-miner concentration and NUKZ’s broader value-chain approach.
Smaller AUM and a more recent launch mean URAN trades with wider bid-ask spreads than its larger peers, which matters more for active traders than for buy-and-hold investors. The fund has logged about an 11% year-to-date gain and a 65% one-year return, tracking the category but at a lower cost basis if the issuer’s typical expense ratio positioning holds. For investors seeking blended uranium-plus-nuclear exposure without paying the premium associated with category-leading AUM, URAN serves as a reasonable substitute.
The tradeoff is liquidity and track record. The fund has not been through a full uranium cycle, and lower trading volume can produce execution friction during sharp moves in the underlying equities.
Choosing between the three
Choosing between the three funds starts with deciding what an investor wants the trade to express. URNM speaks to a view that uranium spot and term prices keep climbing as utility and hyperscaler demand outstrips mine supply, and its concentration in Cameco, NexGen, and physical uranium magnifies that bet. NUKZ lines up most directly with the AI data‑center story because it bundles utilities signing long‑dated PPAs, SMR developers drawing hyperscaler capital, and the engineering firms building new capacity. URAN appeals to a cost‑conscious investor who wants blended exposure and is comfortable trading lower liquidity and a shorter history for a lighter expense profile.
The three funds are not substitutes for one another. The right choice depends entirely on which slice of the nuclear‑AI buildout an investor wants to own.
Top Gaining Stocks
Top Losing Stocks