Data centers are migrating to rural America, and the power grid is the reason

For decades, data centers clustered around major metro areas, close to the customers and fiber networks they served. That playbook is being rewritten. The new gold rush for AI infrastructure is heading somewhere most tech executives wouldn’t have considered five years ago: rural America.

According to a Pew Research Center analysis, 67% of the more than 1,500 data centers currently in development across the US are planned for rural locations. That’s a near-complete inversion of the existing footprint, where 87% of roughly 3,000 operational data centers sit in urban areas.

Why the countryside?

The answer is straightforward: electricity. AI workloads are extraordinarily power-hungry, and urban grids in traditional data center corridors are running out of room. Rural counties offer cheaper land, fewer permitting bottlenecks, and, critically, more available power capacity.

Nearly 40% of the new facilities are targeting counties that don’t currently host a single data center. The rural South is absorbing the largest share, with 754 new facilities slated for construction. The Midwest follows with 419. Texas, Virginia, and Georgia are emerging as the primary hotspots.

The scale of these projects is growing, too. Nearly 100 hyperscale data centers, defined as those drawing at least 100 megawatts at peak load, are already operational in the US. Another 120 are under construction, and 460 more are in the planning stages. Some of the newest AI-focused campuses are being designed at gigawatt scale, enough to power a small city.

To put that in perspective, 100 megawatts can serve roughly 80,000 homes. A gigawatt-class facility would consume more electricity than some entire counties generate.

The grid math is getting uncomfortable

Data centers currently consume approximately 4.5% to 4.7% of all US electricity, around 192 terawatt-hours in 2024. That number is climbing fast. Projections suggest data center power demand could reach 9% to 17% of total US electricity consumption by 2030.

To avoid becoming hostage to local grid constraints, many developers are building their own power sources on-site. Natural gas turbines, battery storage systems, and renewable energy installations are being integrated directly into data center campuses.

Amazon Web Services, Microsoft, and Google are among the major players driving this expansion. All three have made significant commitments to AI infrastructure, and all three are actively securing rural land and power agreements across the South and Midwest.

Small towns, big tradeoffs

For rural communities, the arrival of a hyperscale data center is a complicated proposition. On the upside, these projects bring construction jobs, permanent operations positions, and substantial property tax revenue to counties that often desperately need all three.

On the other side of the ledger, residents are raising legitimate concerns. Data centers require enormous amounts of water for cooling, which can strain local supplies. The facilities generate constant low-frequency noise from cooling systems and backup generators. And a single campus can consume enough electricity to meaningfully push up rates for everyone else on the local grid.

Land use is another friction point. Data center campuses can span hundreds of acres, converting agricultural or undeveloped land into industrial sites.

These tensions are playing out county by county across the rural South and Midwest. Some local governments are embracing the development, offering tax incentives and fast-tracking permits. Others are pushing back, imposing moratoriums or demanding environmental impact studies before breaking ground.

The integration of on-site energy generation is partly a response to these community concerns. By reducing dependence on local utilities, developers can argue their facilities won’t raise rates for existing customers.

With 460 hyperscale facilities in planning alone, the next several years will determine whether rural communities become thriving tech corridors or cautionary tales about industrial development outpacing local capacity to absorb it.