In partnership with

The AI Boom Has a Bottleneck - Follow the Money

AI needs far more than chips. Here’s where the infrastructure money could flow - and a simple way to hunt for opportunities before they become obvious.

✍️ Editor's Note

When most investors hear “AI investing,” they immediately think of the biggest names in technology.

But I'm increasingly interested in what's happening one layer beneath them.

Every new AI model and AI-powered product ultimately depends on physical infrastructure: servers, data centers, networking, cooling systems, electrical equipment - and enormous amounts of power.

And the numbers are getting difficult to ignore.

The International Energy Agency says electricity consumption from AI-focused data centers surged 50% in 2025. Its latest outlook projects total data-center electricity consumption could roughly double from 485 TWh in 2025 to 950 TWh by 2030, with AI-focused facilities growing much faster than the overall sector.

That raises a potentially lucrative question:

What if some of the best opportunities aren't the companies creating AI - but the companies AI can't grow without?

Today, we're going into the Vault to look for them.

Launch your own fund, without the usual price tag.

Launching a fund has traditionally cost tens of thousands of dollars and taken months. That cost kept many capable investors out.

Hedgia rebuilt the process as software. You answer plain English questions for about twenty minutes. The platform prepares draft offering documents for your review, forms your fund and manager LLCs, files your Form D, and helps you open bank and brokerage accounts with partner institutions in the fund's name. Bookkeeping, fund accounting, and investor onboarding run in the background.

Your investors subscribe, sign, and wire through your own portal. They never pay Hedgia anything.

From $89 a month, plus state filing and formation fees. Currently live in nine states.

Start my fund

_{Hedgia is a fund administration platform, not a law firm or investment adviser. Fees apply.}

🧠 The Big Idea: Don't Just Buy AI. Follow AI's Money.

Think about what happens when another billion dollars gets poured into AI infrastructure.

It doesn't all go to an AI software company.

Some goes toward processors.

Some toward memory.

Some toward servers.

Some toward networking.

Some toward cooling.

Some toward constructing and operating data centers.

And ultimately, somebody has to provide the electricity.

The chain looks something like this:

AI Demand → Compute → Data Centers → Networking → Cooling → Electricity → Grid Infrastructure

And each link potentially creates investable businesses.

This isn't merely theoretical. The IEA says a modern data center's electricity consumption includes servers, storage, networking and cooling, with cooling alone ranging from roughly 7% of electricity use at efficient hyperscale facilities to more than 30% at less-efficient enterprise data centers.

So instead of asking:

“Which AI company will win?”

Try asking:

“What will almost every AI company need?”

That's where the hunt gets interesting.

⚡ 5 Places I'd Be Looking

🖥️ 1. The Compute Supply Chain

The obvious play has been advanced AI processors.

But don't stop at the chip.

Think about everything required to manufacture, connect and operate increasingly powerful computing systems:

→ Memory
→ Semiconductor manufacturing equipment
→ Advanced packaging
→ Servers
→ Storage
→ Power-management components

The better question isn't:

“Who makes AI chips?”

It's:

“Who gets paid when the world builds more AI computing capacity?”

That opens a much larger universe of companies.

🌐 2. Networking

Thousands of processors sitting in a data center aren't particularly useful if they can't move enormous amounts of information between each other quickly.

As AI clusters become larger, networking becomes increasingly important.

That puts businesses involved in areas such as:

→ High-speed networking
→ Optical components
→ Switching equipment
→ Fiber infrastructure
→ Data-center connectivity

on my research radar.

They're not necessarily household names.

And that's precisely why this part of the ecosystem interests me.

❄️ 3. Cooling

This might be one of the least glamorous AI investment themes.

It could also be one of the most logical.

AI servers generate heat.

A lot of heat.

And AI is pushing power density higher. The IEA says AI-server power density increased roughly 11-fold between 2020 and 2025, with another fourfold increase projected by 2027.

Which gives us a beautifully simple thesis:

More AI compute → More heat → More cooling

As rack densities increase, technologies such as liquid cooling and specialized thermal-management systems become worth investigating.

Nobody is going to make a Hollywood movie about data-center cooling equipment.

That's okay.

Boring businesses can still benefit from extraordinary trends.

⚡ 4. Power & Electrical Infrastructure

Here's the one I'd watch particularly closely.

AI needs electricity.

Lots of it.

The IEA's current base case has global data-center electricity consumption approaching 950 TWh by 2030, roughly double 2025 levels.

In the United States, data centers could account for nearly half of electricity-demand growth through 2030.

Suddenly, an AI investment watchlist can extend well beyond Silicon Valley.

Think:

→ Utilities
→ Transformers
→ Grid equipment
→ Power management
→ Natural gas
→ Nuclear power
→ Renewable generation
→ Energy storage

The question becomes:

Who gets paid if America needs dramatically more reliable electricity?

That's an investing rabbit hole worth exploring.

🏢 5. Data Centers

Finally, somebody has to house all this equipment.

That puts data-center operators, infrastructure providers, developers and certain real-estate businesses directly in the AI ecosystem.

But there's a major trap here.

Finding an industry that's growing doesn't automatically mean you've found a good investment.

Which brings us to the rule I don't want you to forget.

🚨 Great Trend ≠ Great Investment

Imagine you discover the perfect company.

Revenue is growing.

AI demand is exploding.

Management is bullish.

Everyone loves the story.

There's just one problem:

The stock price already assumes everything goes perfectly for the next five years.

That's why identifying an AI beneficiary is only step one.

Before getting excited, investigate four things:

Growth: Is revenue actually accelerating?

Profitability: Is increased demand turning into earnings and free cash flow?

Valuation: How much future growth are investors already paying for?

Real AI exposure: Is AI materially changing the business—or did management simply discover that saying “AI” makes investors excited?

That last one matters.

We're looking for economic exposure to AI, not marketing exposure to AI.

🤖 Put AI to Work: Build Your Own AI Infrastructure Watchlist

Here's where AI itself becomes useful.

Don't ask ChatGPT or another AI:

“What are the best AI stocks?”

That's far too easy.

Make it behave more like a skeptical research analyst.

Copy this:

🔎 IMPORTANT: Verify What AI Tells You

Never buy an investment because an AI recommended it.

The output above should give you research candidates, not buy orders.

For every company AI identifies:

1️⃣ Go to the source.
Check the company's latest 10-K and 10-Q through the SEC's EDGAR database.

2️⃣ Check investor relations.
Read the latest earnings release and investor presentation directly from the company.

3️⃣ Verify the numbers.
Revenue, earnings, margins, debt, free cash flow and valuation metrics should be independently confirmed.

4️⃣ Verify management claims.
If AI says management reported booming AI demand, find the actual earnings call or filing and read the surrounding context.

5️⃣ Recalculate valuation using current data.
A perfectly accurate P/E ratio from months ago can be completely wrong today.

6️⃣ Try to destroy the thesis.
Don't ask AI only why the company could win. Search specifically for evidence that your investment thesis is wrong.

7️⃣ Check every source date.
Stale information can be accurate and still lead you to a terrible conclusion.

If an important claim can't be independently confirmed:

Treat it as unverified.

AI can make investment research dramatically faster.

It cannot outsource your responsibility to think.

🔓 From the Vault

If you remember only one thing from today's issue, make it this:

Don't just chase the winner. Find the bottleneck.

If AI adoption continues expanding, somebody has to provide more computing.

Somebody has to connect it.

Somebody has to cool it.

Somebody has to house it.

And somebody has to power it.

The investing opportunity isn't necessarily finding the company with the flashiest AI demonstration.

It may be finding a business quietly collecting a toll every time the AI ecosystem expands.

What we're ultimately hunting for is this:

Real AI demand + defensible business + attractive economics + reasonable valuation

Finding all four isn't easy.

But that's precisely why it's worth doing the research.

📌 Your Vault Checklist

Before putting any AI-related company on your investment shortlist, ask:

☑️ What does this company actually sell?

☑️ Why should AI increase demand for it?

☑️ Can I see that demand appearing in its financial results?

☑️ Does it have an advantage competitors can't easily replicate?

☑️ What's the strongest argument against my thesis?

☑️ How much future growth is already priced into the stock?

If you can't answer those questions yet, you probably don't understand the investment well enough yet.

And that's okay.

Keep digging.

🔐 The Bottom Line

The biggest winners from AI won't necessarily have AI in their names.

Some may manufacture equipment.

Some may cool servers.

Some may move data.

Some may generate electricity.

Some may own the buildings where all that computation happens.

That's what makes this investment cycle fascinating.

AI isn't creating one investment opportunity.

It's creating an entire economic ecosystem.

And sometimes the most interesting opportunity isn't where everyone is staring.

It's one layer deeper.

Vaulting Your Wealth Forward,
– T. D. Thompson

AI Investing Vault

The content above is for educational and informational purposes only and does not constitute financial advice or a solicitation to buy or sell any financial instruments. Trading and investing involve significant risk of loss, and past performance is not indicative of future results. Always consult with a licensed financial advisor or conduct your own research before making any investment decisions. Use of AI tools and strategies mentioned above is at your own discretion and risk. AI Investing Vault may receive compensation if you purchase tools or services mentioned in this email, at no additional cost to you.