Listen to the audio version of this article (generated by AI).

Hello, Reader.

In a crossword puzzle, every answer connects. Solve one clue, and another piece suddenly falls into place.

Here’s one: Three words investors hate to see: “supply is ____”

Seven letters across. One answer.

Limited.

That word is becoming one of the most important clues in the AI investment puzzle.

The world wants more AI – more chips, more servers, more electricity, more data centers. But the supply of these critical resources is failing to keep up with demand.

And when supply runs short, the companies supplying the resources could emerge as the biggest winners. That’s why it’s essential for investors to consider the bottlenecks forming within the AI industry.

So, in today’s Smart Money, I’ll examine the two growing constraints of AI, how they may influence which companies will thrive, and the proper ways to invest in them.

Where AI Is Hitting Its Limits

Let’s start with what makes the AI Revolution go ’round: Energy. 

Data centers are filled with expensive chips from companies like Nvidia Corp. (NVDA) and Advanced Micro Devices Inc. (AMD). But those chips must be powered to do any work… otherwise they are just pricey doorstops.

In other words, power isn’t just important to AI growth. It is AI growth. And it has become one of AI’s biggest bottlenecks. 

Demand for power near data centers is already straining local grids. In some areas, electricity now costs up to 267% more than it did five years ago. That means the next AI winners may not just be the companies building smarter machines, but the companies supplying the energy needed to run them.

Meeting this demand will require an all-hands-on-deck approach. That means wind, solar, nuclear, and natural gas. Hyperscalers like Microsoft Corp. (MSFT), Alphabet Inc. (GOOGL), and Amazon.com Inc. (AMZN) are already investing in nuclear, natural gas, and other dedicated power sources to guarantee electricity for future AI infrastructure. For example…

• Microsoft signed a 20-year deal to buy electricity from the planned restart of Three Mile Island nuclear plant in Pennsylvania.
• Alphabet partnered with Kairos Power to develop electricity from small modular reactors (SMRs).
• Amazon is investing $20 billion in Pennsylvania AI data centers, including a campus near the Susquehanna nuclear plant to secure the power needed for AI.

Electricity is clearly becoming a competitive advantage. But it’s only one chokepoint. The next bottleneck is something every AI system needs to function…

What Every AI System Needs

It needs memory, also known as DRAM.

Without enough DRAM, AI systems simply run out of room to process information. And the shortage may persist for years. Nearly 100 gigawatts of new data centers are scheduled to come online over the next four years. But there’s only enough DRAM to support roughly 15 gigawatts over the next two years. 

Without memory, artificial intelligence literally can’t think. 

Nvidia CEO Jensen Huang put it plainly: “The memory bottleneck is severe.” 

And Elon Musk just announced in Space Exploration Technologies Corp.’s (SPCX) first earnings report this past week: “The limiting factor currently is memory.”

So, don’t just take it from me. Take it from the titans of the AI industry.

These bottlenecks are very real, and they will affect how the AI investing unfolds. But it is still missing a key piece; energy and memory are only two constraints.

The Bottleneck Blueprint

This isn’t the first time technology has created a shortage of essential resources. The same pattern appeared during the dot-com boom – when the internet’s rapid expansion created unexpected winners beyond the companies building the digital world.

The explosion of internet infrastructure, personal computers, and networking hardware meant the world suddenly needed far more metals than usual. I’m talking about copper… tantalum… germanium… and other essential ingredients to build the physical internet.

But mining and refining capacity couldn’t expand overnight. The result was a classic supply bottleneck. But investors who anticipated which resources would become scarce had the chance to profit in extraordinary ways.

From 1998 to 2001, I recommended four mining stocks to my readers that went on to generate remarkable gains. These companies became the quiet winners of the late-1990s tech boom.

One of them was Antofagasta plc (ANTO.L), which had become a copper-focused mining company.

I recommended Antofagasta to my readers on December 18, 1998 – about a year before the mine began production.

• Over the next three years, the stock soared 205%, while the S&P 500 was essentially flat.
• Over six years, Antofagasta delivered an astonishing 778% gain, while the S&P continued to nurse its losses, down 27%!

Antofagasta solved the puzzle before most investors even saw the clue. It built capacity during the investment phase of the 1990s – then benefited enormously once the metals bottleneck tightened.

That’s the power of identifying bottlenecks early. Now, we have the opportunity to apply this strategy again.

The Hidden Clues Behind AI’s Next Winners

The word limited is only the first clue in the AI investment puzzle. To find the biggest opportunities, investors need to solve four more:

1. Where is demand overwhelming supply?
2. Which companies control the bottleneck?
3. Will increasing supply be easy or difficult?
4. Has the market recognized the opportunity yet?

If you want to know the answers to these questions, check out my free Market Shock presentation, where I dive even deeper into AI’s physical limitations: energy, memory, and the third bottleneck that could shape the next wave of AI winners.

I also reveal the types of companies that could benefit most from these constraints, including 15 free stocks – ticker symbols and all – that I believe are positioned to profit from the AI shortage problem.

Understanding AI’s power is essential when choosing stocks for your portfolio. But every great puzzle has hidden clues. By identifying the bottlenecks holding AI back, investors can uncover the companies positioned to benefit most from solving them.

Click here to learn how.

Regards,

Eric Fry