Hey there, bargain hunter.
Here’s something most people keep skipping over when they talk about AI’s energy problem: every data center running large language models needs power that runs 24 hours a day, 7 days a week, with no interruptions. Solar doesn’t do that. Wind doesn’t do that. Nuclear does.
And right now, uranium is one of the most structurally interesting commodities in the entire market.
Let’s set the scene quickly. Spot uranium briefly touched roughly $101 per pound in late January 2026 before a geopolitical shock knocked it back. In early Q1 2026, spot was around the mid-$80s per pound. Bank of America has published bullish uranium price outlooks in recent years, but I could not verify a specific $135 per pound target “by 2027” from Bank of America in credible primary sources. The long-term contract price has already reached $90 per pound, its highest level since 2008.
What’s driving this? Three things converging at the same time.
First, AI infrastructure is eating electricity alive. The numbers aren’t subtle. AI models require more electricity than a standard web search, but I could not verify the specific “roughly 10x” figure as a settled consensus statistic. The hyperscalers have started signing long-duration power purchase agreements for nuclear energy because it’s the only carbon-free source that can run continuously. Constellation Energy’s long-term PPA with Microsoft is the template. Others will follow.
Second, supply is structurally broken. Years of underinvestment after the Fukushima disaster left the industry with fewer mines, fewer enrichment facilities, and fewer skilled workers. The U.S. alone consumes on the order of tens of millions of pounds of uranium per year, while domestic uranium mine output was just 677,000 pounds of U3O8 in 2024. The world’s largest producer, Kazatomprom in Kazakhstan, has issued updated 2026 guidance ranges and has made downward revisions to its sales guidance in prior updates—but I could not verify the specific claim that it “deliberately cut its 2026 production guidance by 10%.”
Third, policy is now openly supportive. On May 23, 2025, President Trump announced four executive orders aimed at reinvigorating America’s nuclear energy industry, with an objective of expanding U.S. nuclear capacity from about 100 gigawatts to 400 gigawatts by 2050. More than 30 countries have endorsed major nuclear expansion by the same year. Over 75 reactors are currently under construction globally.
The Names Worth Watching
Cameco (CCJ) is the blue-chip anchor. It’s one of the world’s two largest uranium producers, with a 49% stake in Westinghouse Electric, the company that builds and services nuclear reactors. That Westinghouse piece is important: it transforms Cameco from a pure miner into a broader nuclear fuel-and-services platform. In Q1 2026, Cameco reported uranium revenue of $712 million, with uranium sales volumes up 13% year over year. Delivery targets for 2026 sit at 19.5 to 21.5 million pounds.
The risk worth knowing: CCJ is not cheap on a forward P/E basis. The valuation reflects a lot of good news already. If uranium prices stall or Kazakhstan supply returns faster than expected, the stock will feel it. The Key Lake mill and McArthur River mine have faced periodic disruptions in recent years, and operational disruptions are part of this sector’s DNA. Cameco reports Q2 numbers on July 31 and investors want to see production recovery, contract pricing holding, and Westinghouse pulling its weight.
Centrus Energy (LEU) is the less-talked-about play. Centrus operates the only NRC-licensed HALEU production facility in the United States (at Piketon, Ohio), and began HALEU enrichment in 2023 under a DOE-backed effort. Centrus has announced HALEU-related agreements with Oklo (including a June 2026 letter of intent). However, I could not verify the claim that Centrus already has HALEU supply agreements with TerraPower and X-Energy. Its backlog has been reported at about $3.9 billion through 2040.
Slight tangent, but it matters: small modular reactors are still largely pre-commercial. NuScale did receive U.S. NRC design approval/certification earlier in the decade, but I could not verify the specific claim that its stock surged 22% in late April 2026 on “new commercial deployment contracts.” Oklo has discussed a model that places compact reactors near customer sites, but these are higher-risk bets, and timelines remain uncertain.
NexGen Energy (NXE) sits at the speculative end of the quality spectrum. Its Rook I project in Saskatchewan is widely considered one of the best undeveloped uranium deposits in the world. The Canadian Nuclear Safety Commission issued NexGen a Licence to Prepare Site and Construct on March 5, 2026. NexGen is pre-revenue, so it’s a longer-duration bet. But the asset quality is hard to argue with.
What the Market Is Actually Saying
The Global X Uranium ETF (URA) and Sprott Uranium Miners ETF (URNM) give you diversified exposure without single-name risk. Both can be volatile; however, I could not verify the specific claim that both “pulled back roughly 4%” in recent sector unwinds.
The global uranium market was valued at about $9.7 billion in 2025, with some industry market-research forecasts projecting roughly $13.6 billion by 2033. (Treat these as estimates, not audited facts.)
The part people often skip: the enrichment and services side of the nuclear chain may have better margin profiles and less commodity price risk than pure miners. Cameco understood this when it bought into Westinghouse. The real picks-and-shovels angle in nuclear isn’t just the dirt movers. It’s the fuel cycle companies that sit between the mine and the reactor.
If uranium prices hold above $80 and reactor construction continues at the current pace, the miners look underpriced relative to the demand story. If policy support softens or Kazakhstan normalizes supply faster than expected, the thesis has real air pockets. Size positions accordingly.