Imagine a world where you can create a new app in minutes, design a drug in hours, and automate work that once required an entire department.

Now imagine that same world waiting three years for a transformer.

That’s the strange contradiction emerging from the artificial intelligence revolution. We’re making intelligence cheaper and more accessible at a breathtaking pace. But the physical world isn’t moving nearly as fast.

AI can write code, analyze medical scans, and design factories. It cannot conjure copper out of the ground, build a power plant overnight, or manufacture a specialized laser simply because demand doubled.

And that gap could create one of the biggest investment opportunities of the next decade.

I call it the Scarcity Economy.

The Abundance Paradox

For decades, technology has been a deflationary force. Computing power became cheaper. Data storage became cheaper. Communication became cheaper. Today, AI is pushing that trend into areas we once thought required uniquely human intelligence.

That’s exciting, and it’ll unlock extraordinary productivity gains and create entirely new industries.

But there’s a catch. Every digital breakthrough still depends on something physical. Servers need electricity. Chips need advanced materials. Robots need motors. Data centers need cooling systems, land, construction crews, and connections to the grid.

Software can scale almost instantly. Physical infrastructure can’t.

A utility transformer can take years to manufacture and deliver. A new mine can take up to a decade to permit and build. And a sophisticated semiconductor production line requires equipment that only a handful of companies know how to make.

The faster innovation moves, the more valuable those constraints may become.

Seven Things AI Still Needs

I’m watching seven areas where demand could outrun supply.

First is electricity. AI data centers consume tremendous amounts of power, and that electricity has to reach the buildings reliably. Eaton (ETN) and Powell Industries (POWL) supply equipment that helps make that possible.

Second is copper. Electrification means more wiring, transformers, and transmission infrastructure. Freeport-McMoRan (FCX) is one way to gain exposure to the metal.

Third is photonics. This is a particularly fascinating opportunity. As AI systems grow, moving data between chips and servers becomes a major engineering challenge. Coherent (COHR) and Fabrinet (FN) participate in the optical supply chain that helps solve it.

Fourth is cooling. Powerful chips generate extraordinary heat, and overheated equipment doesn’t produce much value. Vertiv (VRT) and Modine (MOD) provide thermal-management technologies.

Fifth is specialized manufacturing. Even if demand for advanced chips surges, manufacturers need the tools to test and produce them. FormFactor (FORM) and Teradyne (TER) help address that challenge.

Sixth is critical minerals. Rare earths and specialized materials are essential for motors, electronics, and advanced industrial systems. MP Materials (MP) offers exposure to part of that supply chain.

And seventh is skilled labor. AI can help design a data center, but electricians still have to wire it. Companies such as Quanta Services (PWR) and EMCOR (EME) provide the people and expertise to build critical infrastructure.

Notice what’s missing from this list?

Not one AI chatbot.

The Next Wave of Winners

I’ve argued repeatedly that this innovation bull market looks more like 1995 than 1999. We’re still building the foundation for technologies that could reshape the economy over the next several years.

That doesn’t mean every infrastructure stock is a buy. Scarcity can attract competition, customers can delay projects, and even a great business can be a terrible investment at the wrong valuation.

The real opportunity is finding companies with difficult-to-replicate capabilities, strong balance sheets, and the pricing power to benefit when customers have few alternatives.

That’s where I want to focus my research.

The First Move

Wall Street loves the companies making AI smarter. I’m just as interested in the companies making AI possible.

Because the next phase of this revolution won’t be decided by algorithms alone. It’ll also be decided by whether the world can deliver enough power, materials, bandwidth, equipment, and skilled workers to support them.

AI may make intelligence abundant.

But abundance could make scarcity the better investment story.