🚨24-Year-Old Trader Turns $225M into $5.5B: He's Betting Not on AI Models, but on AI's "Electricity"

There's a crazy trading story in the market recently.

A 24-year-old fund manager turned $225 million into $5.5 billion in less than 12 months.

Many assumed such returns must have come from leading AI stocks.

But a closer look at his portfolio reveals a completely different logic.

He's not betting on AI software.

But on AI infrastructure bottlenecks.

His largest single holding is:

$Bloom Energy(BE.US)

A position of roughly $885 million, accounting for about 20% of the entire fund.

What is Bloom Energy's core business?

Providing distributed energy systems for data centers.

In other words—

AI needs computing power, and computing power needs electricity.

This is the most overlooked link in the entire AI infrastructure chain.

His second major bet is:

The CoreWeave ecosystem.

He recently added another $300 million, bringing his investment in CoreWeave to $700 million.

He also holds about a 10% stake in Core Scientific.

One point many overlook is:

A large number of Bitcoin mining farms are being converted into AI data centers.

The reason is very simple:

Mining farms already possess the three things AI needs most:

Electricity

Cooling

Infrastructure

So he's also positioning in mining companies, such as:

$Cipher Digital(CIFR.US)

$Bitfarms Canada(BITF.US)

These companies are gradually transitioning from Bitcoin computing power → AI computing power.

Even more interesting is his short position.

He is massively shorting Infosys.

The logic is equally straightforward:

With the development of AI programming tools like Claude Code and Codex, the traditional IT outsourcing model may be impacted.

In short, his entire investment thesis boils down to one sentence:

The biggest bottleneck for AI is not models.

It's electricity, data centers, and infrastructure.

This is also why his capital is rapidly shifting towards:

Electricity

Energy

Computing infrastructure

He has even reduced positions in some traditional AI chip companies, such as:

$NVIDIA(NVDA.US)

$Intel(INTC.US)

If this logic holds, the true value distribution within the AI industry chain may change.

Model companies attract attention.

But what truly limits the speed of AI expansion might be those industries that seem less "sexy":

Electricity

Energy

Data centers

When the entire industry enters a GW-scale computing power construction cycle, these infrastructure companies could become the new critical nodes.