A Pennsylvania Town of 7,000 Is Being Reshaped by the AI Data Center Boom

Archbald, Pennsylvania is not the kind of place that usually makes national headlines. A borough of roughly 7,000 people tucked into Lackawanna County, it has the quiet, unhurried character of a post-industrial Northeastern town that long ago made its peace with modest ambitions. That peace is now over. A data center development is set to consume approximately 14 percent of the borough's total land area, displace a trailer park community, and press up against the edges of residential neighborhoods in ways that have left many residents feeling blindsided by a future they never voted for.

The project sits at the intersection of two enormous forces that are reshaping American geography right now: the insatiable appetite of artificial intelligence for physical infrastructure, and the desperate hunger of small municipalities for tax revenue and economic activity. Data centers are the unglamorous backbone of the AI economy. Every query processed by a large language model, every image generated, every recommendation served requires vast arrays of servers drawing enormous amounts of power and generating enormous amounts of heat. Those servers have to live somewhere, and increasingly, they are living in places like Archbald, where land is cheaper, local governments are more accommodating, and the political resistance is easier to manage than in larger metros.

What makes the Archbald situation particularly sharp is the human cost that tends to get abstracted away in discussions about data center siting. The trailer park residents facing eviction are not an abstraction. Mobile home communities are among the most economically vulnerable housing populations in the country, often composed of retirees and working-class families who own their homes but rent the land beneath them, leaving them with few legal protections when a developer or municipality decides the land has a better use. When a trailer park disappears, its residents rarely land somewhere better. Research consistently shows that displaced manufactured housing residents face significant difficulty finding comparable affordable housing, and many end up paying substantially more for worse conditions.

The proximity of the planned facility to residential properties raises a separate set of concerns that go beyond displacement. Data centers are not quiet neighbors. They require continuous cooling systems that generate persistent low-frequency noise, they draw on electrical infrastructure in ways that can affect grid reliability for surrounding areas, and they generate significant truck traffic for equipment delivery and maintenance. Residents living near existing large-scale data centers in Virginia's Loudoun County, long known as "Data Center Alley," have documented these quality-of-life impacts for years, and Loudoun has itself begun pushing back on unchecked expansion after decades of welcoming the industry with open arms.

The question of blame in Archbald is genuinely complicated, and that complexity is worth sitting with rather than resolving too quickly. Local officials who approved or facilitated the development were almost certainly responding to real fiscal pressures. Pennsylvania's small boroughs operate on thin margins, and the property tax revenue and construction activity associated with a large data center can look transformative on a municipal budget spreadsheet. The incentive structure is not irrational, even if the outcomes feel unjust to the people bearing the costs.

What the Archbald situation illustrates is a systems-level failure of planning that is playing out in dozens of communities across the country simultaneously. The AI infrastructure buildout is happening at a pace that outstrips the capacity of local zoning and land-use frameworks to respond thoughtfully. Those frameworks were designed for a world where industrial development moved slowly enough that communities could deliberate, negotiate, and adapt. The data center boom is not moving slowly. Microsoft, Google, Amazon, and a constellation of smaller operators are committing billions of dollars to new facilities on timelines measured in months, and local governments are being asked to make decisions with generational consequences under significant time pressure and with significant informational asymmetry.

The second-order effect worth watching here is political. Communities that feel steamrolled by development they did not meaningfully consent to tend to generate lasting civic anger, and that anger has a way of reshaping local politics in unpredictable directions. Archbald may be a preview of a broader backlash against data center siting that could, if it reaches sufficient scale, begin to complicate the infrastructure timelines that the AI industry is currently treating as essentially guaranteed. The companies building the AI economy have largely assumed that the physical world will accommodate their ambitions. Archbald suggests that assumption deserves a second look.