Disruptive Stocks - Futurist

Disruptive Stocks - Futurist

AI Energy

AI’s Next Bottleneck Is Power: 2 Stocks Behind the Data Center Buildout

Futurist's avatar
Futurist
Sep 08, 2026
∙ Paid

Everyone is watching the chip race.

Nvidia this. Blackwell that.

But a chip needs electricity to run and a way to shed heat once it does.

Skip either step and the fastest chip in the world is a paperweight.

On August 29, Elon Musk put a number on exactly how big that skip is.

Roughly 15 gigawatts of AI compute expected in 2027 may not actually come online that year.

X avatar for @elonmusk
Elon Musk@elonmusk
Consensus estimate is that ~15GW of AI compute produced in 2027 cannot be turned on in 2027. This is harder than just finding power, as you also need to build out all the transformers, wiring, liquid-cooling, (massive) chillers & complex networking. grok.com/share/bGVnYWN5…
9:17 PM · Aug 29, 2026 · 6.64M Views

2.04K Replies · 2.65K Reposts · 19.6K Likes

The bottleneck is moving beyond chips.

Musk specifically pointed to:

Power · Transformers · Wiring · Liquid Cooling · Chillers · Networking

For the last few years, the AI trade looked like this:

More AI → More GPUs

The next phase looks more like:

More AI → More Power → More Heat → More Infrastructure

The AI race is moving into the physical infrastructure required to make compute usable.


The Power Chain Behind AI

Before a GPU can run, electricity has to pass through an entire electrical chain.

The GPU gets the headlines.

But every layer before it determines whether the GPU can turn on.

No power infrastructure. No compute.


The Cooling Bottleneck

Once the GPU turns on, the problem reverses.

Almost all the electricity it consumes eventually becomes heat.

That heat has to leave.

AI racks are packing more compute into the same physical space.

That means more electricity and more heat per rack.

That is why liquid cooling is becoming increasingly important.


AI Economics - Megawatt to Tokens

Here is the simplest way to understand the AI power trade:

Tokens are the output. Megawatts are the input.

Every AI token starts with electricity.

More tokens require more compute.

More compute requires more power.

More power requires more infrastructure.

So I am looking for two things:

The companies that already control the power AI needs.

The companies that make that power usable inside the data center.

Two companies stand out.


Where I See the Opportunity

The AI power bottleneck creates two clear opportunities.

1. Own the Megawatts

AI data centers need reliable electricity at scale.

That favors companies that already control large amounts of generation.

2. Own the Infrastructure Around the GPU

Once the electricity arrives, it still has to reach the rack.

Then the heat has to be removed.

That means:

Power Distribution · UPS · Switchgear · Liquid Cooling

So I want exposure to both sides:

Company 1 supplies the power.

Company 2 makes that power usable inside the data center.

Those are the two stocks I am focusing on.:

My 2 AI Power Infrastructure Stocks:

This post is for paid subscribers

Already a paid subscriber? Sign in
© 2026 10x Disruptive Stocks · Publisher Privacy ∙ Publisher Terms
Substack · Privacy ∙ Terms ∙ Collection notice
Start your SubstackGet the app
Substack is the home for great culture