The AI data center boom is colliding with cities scarred by big industry

Philadelphia officials are eyeing a neighborhood already damaged by a defunct oil refinery as a potential site for new AI data centers, according to reporting that's part of a growing national pattern. It's a move that perfectly captures the collision between AI's massive infrastructure needs and environmental justice concerns.

The irony is hard to miss. We're building the infrastructure for technology that's supposed to make everything smarter and more efficient by potentially placing heavy burdens on communities that have already absorbed decades of industrial pollution. Data centers require enormous amounts of power and cooling, generating heat, noise, and strain on local electrical grids, and city planners are treating them like any other industrial facility by routing them to areas with existing industrial zoning.

This isn't just a Philadelphia story. From Northern Virginia to rural Georgia, communities are pushing back against data center construction, concerned about water usage for cooling systems, increased energy demand, and the transformation of their neighborhoods into industrial corridors. The AI boom needs physical space, and it needs it fast, but the traditional playbook of placing infrastructure in lower-income areas is running into organized resistance.

The scale of demand is staggering. Training a single large language model can consume as much electricity as hundreds of homes use in a year, and inference at scale requires always-on server farms. Tech companies are scrambling for power purchase agreements and suitable land, often promising economic benefits while underplaying environmental impacts.

What makes this moment different from past industrial booms is the speed and the awareness. Communities have seen this movie before with refineries, power plants, and factories. They're organizing earlier, armed with better data about environmental and health impacts, and they're questioning why the profits from AI flow to Silicon Valley while the infrastructure costs land in their backyards.

The fundamental tension won't resolve itself. AI companies need massive compute infrastructure, and that infrastructure has to go somewhere. But the current approach, which treats data centers as just another industrial use and routes them to communities of least resistance, is generating the kind of opposition that could slow the entire industry.

What this means for you: If you're building AI-dependent products or services, factor in that infrastructure costs and timelines may increase as siting becomes more contentious. The compute you rely on has a physical footprint, and that's becoming a constraint. When planning projects, try this prompt with your AI assistant: "Help me estimate the compute requirements for [your project] and identify three ways to optimize for efficiency without sacrificing core functionality." Building lighter from the start may become a competitive advantage as infrastructure gets harder to scale.