markets · uranium · ai
Is AI Creating the Next Uranium Bull Market?
· 6 min read
When most people think about the AI boom, they think about the obvious names. Nvidia. Microsoft. OpenAI. Google. The chips, the models, the software. What the headline names obscure is that every AI model, every chatbot reply, and every data center resolves down to one input: electricity.
Key takeaways
- A GPU is a space heater that happens to do math. At scale, intelligence is a power bill.
- The IEA expects data center electricity use to more than double to about 945 TWh by 2030, near the entire annual consumption of Japan, with AI as the primary driver.
- Data centers run 24/7, so nuclear is back in focus as reliable baseload that does not stop when the wind drops or the sun sets.
- The IAEA’s high case sees global nuclear capacity more than double by 2050, from 377 GW to roughly 992 GW.
- Key risks: build timelines, politics, supply response, and a single AI efficiency breakthrough re-rating demand overnight.
That creates a problem the demand side never fully priced in. How do you generate enough reliable electricity to run the next generation of AI infrastructure? One of the answers being taken seriously is nuclear. And if nuclear takes on a larger share of that load, uranium becomes one of the more overlooked beneficiaries of the entire AI story.
Why AI needs more power
Training large models and running AI data centers consumes far more electricity than the internet services that came before them. The numbers are not subtle.
The International Energy Agency estimates data centers drew around 415 terawatt-hours in 2024, roughly 1.5 percent of global electricity. It projects that to more than double to about 945 terawatt-hours by 2030, close to the entire annual consumption of Japan, with AI as the primary driver. Microsoft, Google, Amazon, and Meta are already committing tens of billions to data center expansion. The demand is not a forecast. It is under construction.
Why nuclear energy is back in focus
Meeting that load takes a mix of sources, and renewables remain central. But data centers need power that does not stop when the wind drops or the sun sets. They run flat out, 24 hours a day, 7 days a week.
That requirement is what put nuclear back on the table. Unlike solar and wind, a reactor delivers consistent baseload regardless of weather, which is exactly the profile an always-on data center wants. The shift is in the framing. Nuclear used to be argued almost entirely on climate grounds. Now it is increasingly discussed as infrastructure, the boring and reliable kind that AI growth quietly depends on.
Why uranium matters
If nuclear generation rises, uranium demand tends to follow. And uranium has a structural quirk that makes it interesting: supply cannot turn on quickly. New mines take years to permit, finance, and build. Meanwhile, existing reactors are having their lives extended and new projects are being announced.
The scale of the projected build is larger than most casual framings suggest. In its most recent high case, the IAEA expects global nuclear capacity to more than double by 2050, with small modular reactors doing a meaningful share of the work.
| Metric | 2024 | 2050 (high case) | Change |
|---|---|---|---|
| Global nuclear capacity | 377 GW | ~992 GW | ~2.6x |
| Data center electricity use | ~415 TWh (2024) | ~945 TWh (2030) | +128% |
So the setup is this. Electricity demand climbing as AI scales. Nuclear gaining support as a reliable source. Uranium supply still slow to respond. That combination is why some investors treat uranium as a long-fuse beneficiary of both the energy transition and the AI buildout.
The risks investors should consider
The thesis is clean, which is its own warning. Clean theses attract crowds, and crowds move price before fundamentals do.
Time
Reactors take years to build, and delays are the norm rather than the exception. A multi-year construction cycle means the demand you are underwriting today shows up slowly, if it shows up on schedule at all.
Policy
Nuclear runs on politics as much as physics. A single permitting decision or an election can move the timeline in either direction, and public sentiment toward nuclear has historically been volatile.
Supply and demand
Higher prices invite more production, which can erode the scarcity the bull case rests on. And the whole structure assumes AI keeps getting hungrier. One credible efficiency breakthrough and the demand curve re-rates overnight, not in your favor. Like any commodity, uranium is volatile, and it belongs inside a portfolio context rather than as a single conviction bet.
What investors should watch
Three things will tell you where this goes over the next several years.
1. AI electricity demand forecasts
If projections keep climbing, the case for more generation strengthens. Watch how the major data center operators revise their capital plans and how grid operators respond.
2. Reactor development worldwide
New builds, life extensions, and policy support are the leading indicators of future uranium demand. Small modular reactors are the wildcard, and their progress is worth tracking closely.
3. Uranium supply
The race between supply growth and demand growth is what ultimately sets the price. People express the uranium view in different ways, from physical-uranium trusts to miner-focused funds to the producing equities themselves, each carrying a different blend of leverage, timing, and volatility. They are not interchangeable, and the right expression depends entirely on horizon. A multi-decade thesis held through a fast-moving instrument is a different animal than the same thesis held patiently. Match the instrument to the timeline, or the timeline will find you.
Final thoughts
For years the AI trade meant the companies building the software and the chips. Those names still matter. But the infrastructure underneath them, the power that makes any of it run, is where the less crowded opportunities tend to hide.
Every AI query, every model, every data center draws on the grid. As that draw grows, nuclear keeps entering the conversation, and uranium sits quietly upstream of it. None of that guarantees uranium prices surge. Plenty can break the chain. But it does mean the story is worth watching closely, and worth understanding before the crowd finishes arriving.
Sources
- Energy and AI: Energy supply for AI (International Energy Agency)
- Data Centres, Artificial Intelligence and Cryptocurrencies Eye Advanced Nuclear to Meet Growing Power Needs (International Atomic Energy Agency)
- Uranium Market Dynamics and Supply (U.S. Geological Survey)
- Is Nuclear Energy the Answer to AI Data Centers’ Power Consumption? (Goldman Sachs Research)
- IAEA Raises Nuclear Power Projections for Fifth Consecutive Year (International Atomic Energy Agency)
- World Nuclear Association
This article is for informational purposes only and is not investment advice. Trading involves risk, including the possible loss of capital. Do your own research before making any investment decision.