Disruptive Themes: AI Energy
Nuclear Energy
AI doesn’t run on chips.
It runs on power.
And power is becoming the real bottleneck.
Every major AI breakthrough eventually runs into the same constraint:
electricity.
Not someday. Right now.
The market is obsessed with GPUs and models.
But AI is becoming a physical infrastructure story.
More compute means:
More electricity
More cooling
More transmission
More reliable baseload power
That changes the investment landscape completely.
This is how I mentally organize the AI Energy Stack.
1. Uranium - The Foundation
No uranium.
No nuclear power.
No scalable AI.
As AI data centers expand, demand for reliable energy rises with them.
That puts uranium back into focus.
Key Players:
Cameco (CCJ)
NexGen Energy (NXE)
Denison Mines (DNN)
Energy Fuels (UUUU)
Uranium Energy (UEC)
This layer is cyclical and volatile.
But if nuclear power becomes critical to AI infrastructure, uranium supply could become strategically important again.
2. Fuel Processing - The Hidden Bottleneck
Mining uranium is only step one.
Fuel still needs enrichment before reactors can use it.
That capacity is limited.
Scarcity creates leverage.
One company attracting attention here:
Centrus Energy (LEU)
Every major infrastructure boom eventually reveals hidden choke points.
This may be one of them.
3. Reactors - The Future Buildout
AI data centers need constant, reliable power.
Not intermittent power.
Not theoretical power.
24/7 baseload electricity.
That’s why Small Modular Reactors (SMRs) are gaining attention.
Key names:
Oklo (OKLO)
NuScale Power (SMR)
BWX Technologies (BWXT)
Nano Nuclear Energy (NNE)
This layer carries the highest uncertainty.
Some companies will fail.
But if SMRs become commercially viable, they could reshape how future AI campuses are powered.
This is where speculation turns into infrastructure.
4. Utilities - The Cash Flow Layer
These companies already generate electricity at scale.
They are not selling a future vision.
They operate the grid today.
Key players:
Constellation Energy (CEG)
Vistra Energy (VST)
Duke Energy (DUK)
Exelon (EXC)
Southern Company (SO)
NextEra Energy (NEE)
This layer looks boring.
That’s usually where durable compounding hides.
AI infrastructure still needs actual electricity providers behind the scenes.
5. Grid Infrastructure - The Picks & Shovels
No transformers.
No switchgear.
No grid modernization.
No AI energy expansion.
As electricity demand rises, infrastructure spending rises with it.
Key companies:
GE (GE)
Eaton (ETN)
Honeywell (HON)
Emerson Electric (EMR)
This may be the safest layer in the stack.
These companies benefit regardless of which reactor technology wins.
The Bigger Picture
Most investors still think AI is mainly a software story.
It isn’t.
It’s becoming an energy and infrastructure buildout on a massive scale.
The stack looks something like this:
Uranium → Fuel Processing → Reactors → Utilities → Data Centers
Hardware connects every layer.
That’s how infrastructure compounds:
slowly at first, then all at once.
Final Thought
AI may change the world.
But scaling AI requires physical systems:
power plants, grids, cooling, fuel, transmission, and electrical infrastructure.
The market is still heavily focused on chips and software.
Eventually it may realize the real bottleneck is power.
And when markets discover bottlenecks, capital flows upstream.
That’s where some of the biggest opportunities may emerge over the next decade.
Disclosure: For informational purposes only. Not investment advice. It reflects my personal opinions for research and discussion purposes only. I may hold positions mentioned and may change positions at any time without notice. Do your own research.