Across the world, cities are facing a historic challenge: they need powerful artificial intelligence computers called data centers to power the future, but building these facilities is creating unexpected problems for electricity grids and communities. This weekly update examines how city planners are learning to work with AI infrastructure in smarter, more efficient ways.

The Data Center Boom is Reshaping American Cities

The race to build AI computing power has reached unprecedented levels. In 2025, just four major technology companies—Google, Amazon, Microsoft, and Meta—invested $465 billion in building data centers and computer networks. This money is flowing into unexpected places. Southern California's Vernon, a tiny industrial town with just a little more than 200 residents, has become a hub for massive AI data centers. Inside one three-story building called LAX01, advanced computer chips hum continuously, using enough electricity to power more than 26,400 homes for a year.

Why is Vernon so attractive to tech companies? The town has several advantages. First, it operates its own public electric company that charges lower rates—sometimes as little as half the price of other Southern California electricity. Second, with so few people living there, residents cannot organize to protest data centers the way they do in other neighborhoods. Some towns, worried about wasted water and higher electricity bills, have blocked billions of dollars of data center projects through community opposition.

The Power Problem Gets Serious

Data centers are now one of the biggest reasons electricity demand is growing in the United States. California, which already struggles with power costs and environmental rules, is becoming the country's third-largest data center hub after Texas and Virginia. One analysis showed that electricity costs near data centers have jumped as much as 267 percent over five years.

This creates real problems. A proposed data center in Westfield, Massachusetts would need four to five times more power than the entire city uses. Another project in Texas wants to build a massive nuclear power plant just to run AI computers. These enormous power demands are pushing electrical grids to their limits. Old equipment, like power substations built in 1975, now runs at 90 percent capacity and cannot handle more.

Cities Are Learning to Build Smarter AI Systems

Instead of just accepting this situation, some cities and government agencies are redesigning how they use AI to waste less power. Leaders have discovered that better design matters more than just buying bigger computers.

One important strategy is called pushing intelligence to the edge. Instead of sending all information from city sensors—like traffic cameras and bridge monitors—far away to giant cloud computers, smaller AI computers now process the information right where the sensors are located. A vibration sensor on a bridge, for example, no longer needs to stream gigabytes of raw data to distant servers; it can simply send an alert when something is wrong. This cuts down on wasted energy and makes cities faster and more reliable.

Cities are also getting serious about choosing the right computer hardware for each job. Not every AI task needs the biggest, most powerful graphics processing units that burn enormous amounts of electricity. Many jobs work well on smaller, more efficient computer chips designed to use less power. Some agencies are building local simulation labs where they can test and adjust their AI systems before putting them to work.

The Abandoned Project That Says Something Important

Westfield's abandoned data center project tells an interesting story. In 2021, the town gave a real estate company called Servistar huge tax breaks—worth $352 million over 40 years—to build a massive data center. Officials promised it would bring jobs and money to the community. But five years later, the project has gone nowhere. The developer never filed the required environmental permits and appears to have abandoned the plan.

One reason might be practical: New England's electricity costs about 24 cents per kilowatt-hour, but data center companies want to operate where power costs only 5 to 10 cents. The region simply does not have enough cheap power or enough grid capacity.

What Cities Should Do First

Experts now agree that mission planning comes before technology. Cities that first decide what specific problem they want to solve, who will benefit, and what success looks like—they make smarter decisions about AI and infrastructure. Cities that just buy expensive equipment without a clear plan often end up wasting money on tools that do not actually help.

The lesson for communities worldwide is clear: the AI infrastructure boom is not stopping, but thoughtful planning, efficient design, and local control can help cities benefit from AI while protecting their electricity grids, water supplies, and communities.