Turn Post Offices into Data Centers

Postal property can be used in a far better way.

Photo by Scott Rodgerson on Unsplash

Data centers need land, and a lot of it. According to a recent analysis by real-estate investment management firm Hines, the world will need about 40,000 new acres of land—or almost 2 billion square feet—to meet growing demand for artificial intelligence (AI) services. To maintain its global share of output capacity, the U.S. will need to contribute about 25,000 new acres worth of land for these power-intensive operations. In other words, America will need to find almost two Manhattan’s worth of land to continue to be a global leader in AI.

Enter the U.S. Postal Service (USPS). America’s mail carrier owns 8,500 properties (including about 7,200 post offices) totaling an astounding 20,000 acres of land. What the USPS does not have is a lot of money. The struggling agency lostmore than $9.5 billion in fiscal year 2024, and has accumulated more than $100 billion in red ink over the past fifteen years. Selling underused land to high-valued operations such as data centers would cement American AI leadership while delivering the USPS out of debt.

The USPS is sitting on thousands of acres of land it does not need. Post offices of course provide a valuable function to the millions of Americans that need their mail delivered every day. But it doesn’t follow that there needs to be dedicated brick-and-mortar post offices with an army of federal employees serving consumers. The USPS can sell its real estate and save a significant sum on labor and maintenance costs by outsourcing “window” operations (e.g., stamp sales, package hand-offs) to private retailers. The USPS currently has about 2,500 “alternative retail access points” in which private retailers take the lead on postal operations.

The Inspector General (IG) notes though that “these locations have declined by nearly 20 percent in recent years” and closures have been especially significant recently (possibly because of DOGE). The USPS can save money by reversing course and expanding retail contracts, but it needs to tweak its current model. According to the IG, most of these locations are paid fixed sums to provide services, and payouts aren’t closely monitoredto ensure that the locations are serving consumers. Instead, the USPS should switch to a commission-based model to make sure that taxpayer dollars aren’t being wasted on seldom-used locations.

The USPS used to be more ambitious at contracting out at scale. In 2013, the agency launched a pilot to put postal retail operations in more than eighty Staples locations. The idea was to expand this partnership if everything went smoothly. Despite the convenience and reasonableness of this idea, unions (predictably) killed the partnership shortly after it began. A similarly large effort today would probably require Congress to override these unions and mandate that the USPS contract out a large percentage of its retail operations.

Allowing this to proceed and selling off redundant post offices would result in a large windfall for the struggling agency. According to a 2025 Wells Fargo analysis, the USPS could rake in $12 billion selling its post offices. Even assuming it kept half of its retail properties, the agency could use sale proceeds to erase most of its current year net losses. While that’s just a drop in the bucket compared to accumulated losses over the decades, it’s a significant start toward solvency.

To avoid confusion and red tape, lawmakers should step in and exempt any such mass sales from a lengthy National Environmental Policy Act review process. Assuming regulations don’t get in the way, USPS property sales to data center operations could be a rare all-around win for the government, innovators, taxpayers, and consumers. It’s time for a new and bold approach that drives American AI and digital leadership forward.