U.S. Energy Firms Are Buying Up Abandoned Warehouses—Turning Rust into AI Gold

A warehouse that formerly held auto parts is located on the outskirts of an industrial park outside of Dallas. Weeds are growing through the concrete on its cracked loading docks, and the company’s faded logo is hardly discernible. However, surveyors have just started to show up, walking the perimeter, taking measurements, and using technical language that is quiet. The building was sold in a matter of months, but not to a logistics company, but to an energy developer who had completely different plans. A farm of AI servers.

Energy companies are purchasing abandoned warehouses across the United States at an almost urgent rate, transforming remnants of the manufacturing era into artificial intelligence hubs. Once emblems of industrial decline, these buildings are now strategic assets. This change may reveal more about America’s economic transformation than any official data.

This warehouse rush has a straightforward but startling cause: AI requires massive amounts of electricity. Data centers in the United States used about 183 terawatt-hours of electricity in 2024 alone, accounting for more than 4% of the nation’s total electricity consumption. The scale becomes real as you stand inside one of these repurposed areas, where server racks hum like mechanical insects. Heat is emitted from all directions.

Energy companies are entering the market because they have a basic understanding. Controlling the building, the power source, and the land entails owning the AI’s fundamental infrastructure. Investors appear to think that this is where long-term value can be found in both software and the hardware that supports it.

There’s also geography.

Numerous shuttered warehouses are situated close to fiber-optic cables, railroad tracks, and power lines. These links, which were established for shipping and manufacturing decades ago, now have a new function. The irony is difficult to miss. Digital intelligence is now being fed by infrastructure that was intended for industrial production.

Land prices have sharply increased in Northern Virginia, which is frequently referred to as the data center capital of the world. Data center operators have paid hundreds of millions of dollars for parcels that were originally intended for housing developments. Due to their inability to compete financially, developers who once constructed suburban homes are leaving the market. Whole local economies seem to be changing as a result of AI’s appetite.

Energy companies are not working alone. Giants in technology are supporting these initiatives by securing specialized electricity sources and investing in new power plants. Recent agreements announced by the Trump administration mandate that tech companies contribute $15 billion to the development of new energy. It becomes evident from observing these transactions that the real impediment to the growth of AI may be electricity rather than code.

The transformation within these repurposed warehouses is striking.

Technicians now install cooling systems, running miles of cable beneath raised floors where pallets were once moved by forklifts. The scent of the air has changed. It’s heavier but cleaner. People are being replaced by machines.

This change has ramifications that go beyond technology. Demand from data centers has contributed to an increase in residential electricity costs in a number of states. Officials in Virginia warned that if current trends continue, energy use could increase by 183% over the next 20 years. Whether these expenses will eventually be borne by average households is still up in the air.

Some projects go even farther.

Large AI campuses are being constructed in Texas by developers using gas, solar, and nuclear energy, producing more electricity than some other states use. Almost unreal is the scale. Whole landscapes have been redesigned with computing power in mind.

Additionally, there is risk.

Not all conversions of warehouses are successful. Power outages, regulatory hold-ups, or financial difficulties can cause some projects to stall. By placing billions of dollars on AI infrastructure, investors are placing long-term bets on a technology that is still developing in an unpredictable manner.

But the momentum keeps going.

Speed is provided by abandoned warehouses. It takes years to build new data centers from the ground up. Existing buildings can be repurposed considerably more quickly. Speed is important in a competitive AI race.

There is an odd feeling of transition when passing one of these locations at night with the dim light coming from recently installed windows. The structure’s shape hasn’t altered. However, it serves a completely different function.

Physical goods were once stored in these warehouses. They now possess information, which is less obvious but possibly more valuable.

It’s unclear if this change will ultimately make communities stronger or weaker. One abandoned warehouse at a time, America’s industrial past is subtly laying the groundwork for its AI future.