Here’s the plot twist nobody saw coming: the technology that was supposed to supercharge the digital economy just broke it.
For four decades, America bet everything on software. We shipped factories overseas, told people to learn to code, and built an entire economy on the idea that intelligence—the kind you could package into an app—was scarce and therefore valuable. It worked. Tech companies became the foundation of American wealth. By 2025, they were basically running the show.
Then AI showed up and said: “Actually, intelligence isn’t scarce anymore.”
When you can generate intelligence on demand—faster, cheaper, and at basically zero marginal cost—the whole economics of software changes. Not overnight, but persistently. And when the economics change, the money moves.
The Great Flip
Welcome to what we’re calling the “Great Re-Materialization.” It’s a reversal of where value actually lives in the economy. For decades, it lived in software. Now it’s moving back to physical stuff: data centers, power grids, copper wire, cooling systems, and the raw materials that make it all work.
Think about it logically. AI doesn’t run on clever ideas. It runs on GPUs. GPUs live in massive data centers made of steel and concrete. Those data centers need insane amounts of electricity—so much that the U.S. power grid is basically freaking out. That power needs to be generated, transmitted, and managed by physical infrastructure that takes years to build. And the GPUs themselves? They generate enough heat to melt a small country, so you need sophisticated cooling systems made of copper tubing and specialized fluids.
The entire ecosystem needs fiber optic cable, networking equipment, and connectivity infrastructure. And underneath all of it: copper, silver, aluminum, rare earth elements, natural gas, water.
This is where value actually concentrates in the AI economy. Not in the software sitting on top. In the physical cascade from GPU to raw material.
The Market Already Knows
Year-to-date in 2026, the strongest performers read like an industrialist’s shopping list: Vertiv (data center power and cooling, up 67%), Modine Manufacturing (thermal management, up 50%), Corning (fiber optic cable, up 53%), Bloom Energy (distributed power generation, up 83%). All physical companies. All “boring.”
Meanwhile, the software darlings are getting crushed: Atlassian, MongoDB, Workday, HubSpot, Intuit. Every one is a software play facing the same problem—AI is making their core value proposition cheaper.
Why This Matters
Three forces are accelerating this shift simultaneously. First, the hyperscalers are spending over $600 billion on AI infrastructure in 2026, and that number keeps rising. Second, America is trying to re-industrialize in a hurry—CHIPS Act, Infrastructure Investment and Jobs Act, tariff policy—all pushing manufacturing back home. Third, the power grid is under growing strain, and AI is a major reason why.
The supply side is structurally constrained. We didn’t just outsource manufacturing; we stopped training electricians and building transformer factories. The lead time to bring new copper production online is 8-12 years. The skilled trades workforce has been aging out for two decades with no pipeline to replace it.
The Bottom Line
The investors who made fortunes in the original AI trade bought Nvidia when it was a gaming graphics company. The investors who make fortunes in the Re-Materialization will have bought copper miners, power equipment manufacturers, and industrial constructors before the mainstream media figured out that AI runs on copper.
The first wave of AI rewarded the obvious winners. The next wave will reward the companies positioned closest to where value actually concentrates.
