.jpg)
Western Pennsylvania has been here before.
The region knows what it looks like when a massive industrial project arrives promising jobs and economic growth while carrying environmental and health costs that fall disproportionately on the communities closest to the facility. The Shell cracker plant in Beaver County — which we covered on this blog — is the most recent example. But the steel mills, the coke plants, the fracking operations, the natural gas pipelines: this is a region with a long and complicated relationship between industry and environment, between economic need and environmental health.
Now there's a new wave coming. And like the ones before it, it's arriving with promises and questions in roughly equal measure.
Data centers — the physical infrastructure that powers the internet, cloud computing, and increasingly artificial intelligence — are expanding rapidly across the United States, and western Pennsylvania has emerged as a target region for that expansion. Multiple proposals and site evaluations are underway across Beaver, Washington, and Westmoreland counties. The companies involved are some of the largest in the world. The facilities being proposed would be among the largest ever built.
Before the zoning hearings happen, before the permits are issued, before the shovels break ground — communities deserve to understand what they're being asked to host.
A data center is a large facility housing the computer servers, networking equipment, and storage systems that make digital services work. When you stream a video, send an email, use a search engine, or ask an AI chatbot a question, the actual computing happens in a data center somewhere — often many data centers simultaneously, distributed across different locations for reliability.
The scale of modern data centers — particularly the hyperscale facilities being proposed by major tech companies — is difficult to fully grasp. These are not server rooms tucked into office buildings. They are enormous industrial facilities, often spanning hundreds of thousands of square feet, consuming as much electricity as small cities, and requiring infrastructure — power lines, fiber cables, cooling systems, water supply — that fundamentally alters the landscape around them.
The AI boom has dramatically accelerated data center expansion. Training and running large AI models requires vastly more computing power than traditional internet services. A single query to a generative AI system uses roughly ten times the energy of a standard Google search. As AI becomes embedded in more products and services, the demand for data center capacity grows accordingly — and the environmental footprint grows with it.
Water is the question at the heart of data center environmental debates, and the answer comes down to heat.
Servers generate enormous amounts of heat when processing data. That heat has to be removed continuously — if servers overheat, they fail, and the entire facility goes down. Cooling is not optional. It's the fundamental operational requirement that everything else is built around.
Most large data centers use evaporative cooling systems to manage this heat. The basic principle is the same as sweating: water absorbs heat as it evaporates, carrying the heat away from the system. In a data center's cooling towers, hot water from the server halls is pumped to the roof, where it evaporates into the air, cooling before recirculating back through the building.
This process consumes water at a scale that is genuinely staggering.
A large hyperscale data center can consume between one million and five million gallons of water per day. For context: the average American household uses about 300 gallons per day. A single large data center can consume the equivalent of thousands of households' daily water use — every day, indefinitely.
AI-focused data centers use significantly more water than traditional ones because AI processing generates more heat. Microsoft reported that its global water consumption increased 34 percent between 2022 and 2023, a single year, driven largely by AI infrastructure. Google reported a 20 percent increase in the same period for the same reason. These numbers are not anomalies — they're the direction of travel.
The water consumed through evaporative cooling is not returned to the local water system. It evaporates. In regions with water scarcity, this is already a serious concern. In a region like western Pennsylvania, which has historically had relatively abundant water resources, the question is not whether water will be consumed but at what scale, from which sources, and with what effects on local watersheds and communities that depend on them.
Water gets less attention than energy in the national data center conversation, but both matter.
Data centers are among the most energy-intensive facilities ever built. The servers themselves consume electricity constantly. The cooling systems consume additional electricity to run. The power backup systems — which data centers require to maintain uptime guarantees — consume electricity to maintain readiness.
The United States data center industry consumed approximately 200 terawatt-hours of electricity in 2022 — roughly equivalent to the entire annual electricity consumption of some medium-sized countries. Projections suggest that number could double or triple by the end of the decade as AI infrastructure expands.
Where that electricity comes from matters enormously. A data center powered by renewable energy has a fundamentally different carbon footprint than one powered primarily by natural gas or coal. Tech companies have made public commitments to renewable energy, but the reality on the ground is more complicated — the scale of new demand being created by AI data centers is outpacing the expansion of renewable supply, and some facilities are effectively drawing on fossil fuel generation despite corporate sustainability pledges.
Pennsylvania's energy grid is a mix of nuclear, natural gas, coal, and renewables. A major new data center drawing significant power from that grid is not drawing exclusively from renewable sources, whatever the company's public commitments may suggest. The actual energy source for a proposed facility is a legitimate question for communities and regulators to ask.
There's an additional layer specific to Pittsburgh that the data center conversation rarely surfaces: the problem of Duquesne Light.
Duquesne Light, the primary electricity provider for most of the city and surrounding region, has faced sustained and well-documented criticism for infrastructure failures that leave thousands of customers without power for extended periods — often following weather events that neighboring utilities handle without comparable disruption.
The late April and early May 2024 storms are a good example: hundreds of thousands of customers lost power, with many remaining without electricity for up to a week. These were not unprecedented weather events. Yes, they were severe thunderstorms. But thunderstorms, even bad ones, are common in the region. For a weather pattern the regions sees regularly, shouldn't the city's major electricity provider be better prepared? Incidents happen, but it's a situation that was at least somewhat avoidable with better preparation. And the response illustrated a grid that is already not keeping pace with the demands placed on it before supporting a massive data center even enters the equation.
For a data center operator, grid reliability is non-negotiable — which means any major facility in the region would either require significant private backup generation infrastructure, or would become another voice pressuring Duquesne Light to prioritize industrial uptime over residential service reliability. Neither outcome is straightforwardly good for the communities data centers are supposed to benefit. The infrastructure that data centers require to function is not always the same infrastructure that ordinary residents need to live.
Beyond water and energy, data centers carry local impacts that are less often discussed in the broader environmental conversation but matter enormously to the communities where they're built.
Land use is significant. Hyperscale facilities require large footprints — hundreds of acres in some cases — and the surrounding land use changes substantially. Agricultural land, greenspace, and existing neighborhoods can find themselves adjacent to an industrial facility that operates around the clock with significant noise, light, and traffic.
Noise is a persistent issue in communities near data centers. The cooling systems that run continuously generate industrial-level sound that doesn't stop at night or on weekends. Residents near operating data centers have reported significant impacts on quality of life from this ongoing noise exposure, and mitigation measures vary widely in their effectiveness.
Traffic during construction phases is substantial. A major data center project employs thousands of construction workers and requires continuous delivery of equipment and materials. Roads in rural and semi-rural areas where many data center sites are located are often not designed for this volume of heavy truck traffic, and the infrastructure damage and congestion can persist for years.
Employment promises deserve scrutiny. Data center construction creates significant temporary employment — thousands of jobs during the build phase. Permanent operations employment is much smaller, typically in the hundreds, and skews heavily toward specialized technical roles rather than the broader workforce employment that construction numbers suggest. The economic case for a data center is real but more limited than the initial job numbers imply.
Western Pennsylvania is an attractive location for data center development for reasons that are straightforward from a business perspective.
Available land at relatively affordable prices. Proximity to major fiber network infrastructure. Existing electrical transmission capacity from the region's industrial history. Access to water from the Ohio River watershed. A location within reasonable distance of major population centers and financial markets in the northeastern United States.
The region has also seen significant industrial job losses over decades, which creates political conditions favorable to large employers — local officials are often eager to attract investment and employment, and the permitting process can move quickly when economic development is the priority.
This is the same dynamic that made western Pennsylvania attractive to the Shell cracker plant. The economic conditions that make a region eager for investment are often the same conditions that make communities vulnerable to accepting industrial projects without full accounting of their costs.
The communities in Beaver, Washington, and Westmoreland counties that are currently in the path of potential data center development are not uniformly opposed. Some welcome the economic activity. Others have significant concerns. Most are trying to get accurate information in a situation where the companies involved are not always forthcoming and the regulatory process moves faster than community understanding. Even government information (and access to it) remains volatile.
Three days before the publication of this article, the Department of Energy pulled down their page on evaporative cooling. If we're being generous, that means the information is changing and being updated all the time. And if we're being skeptical, it means that access to the research-supported information communities need to make good decisions about their environmental and economic futures has spotty accessibility at best.
If you live in western Pennsylvania and a data center project is being proposed in your area, these are the questions worth demanding answers to before permits are issued.
Where will the water come from? Which water source — municipal supply, groundwater, surface water — will supply the facility's cooling needs? What volume per day? What are the current capacity and stress levels of that source? What happens to the water downstream of the facility?
What is the actual energy source? Not the corporate sustainability pledge — the actual generation mix that will supply this facility from the regional grid. What commitments, if any, are legally binding rather than aspirational?
What noise mitigation is required and enforceable? Not what the company promises, but what is legally required by the permit, how it will be monitored, and what recourse residents have if standards are not met.
What is the tax arrangement? Many data center projects negotiate significant tax abatements that reduce the community's actual fiscal benefit substantially below what the raw investment numbers suggest. Knowing the net tax impact is essential for evaluating the economic case.
What are the permanent employment numbers? Construction employment versus permanent operations employment are very different numbers. The permanent operations figure — not the construction peak — is what matters for long-term community economic benefit.
What environmental review is required? Is an environmental impact assessment being conducted? By whom? Is it publicly available? Can community members submit comments?
Who is the developer and what is their track record? Data center developers vary significantly in how they engage with communities, how they handle complaints, and how their facilities actually perform against their commitments. Researching the developer's record in other communities where they've built is worth doing.
The data center boom driven by AI investment is not unique to western Pennsylvania. It is happening across the United States and globally, and it is raising the same questions everywhere: who bears the environmental costs of digital infrastructure, how are those costs distributed across communities with different levels of political power, and who gets to decide.
These are not abstract questions. They are the same questions that communities across this region have asked about every major industrial project that has arrived with promises and left with consequences. The answers have not always been good ones.
Western Pennsylvania has environmental advocates, labor organizers, local journalists, and community members who have navigated these dynamics before. The organizations and individuals who pushed back on the Shell cracker plant, who have documented the health impacts of fracking in the Marcellus Shale region, who have tracked air and water quality in communities adjacent to industrial facilities — they are the resources worth connecting with as this wave of development arrives.
The data center boom will happen. The question is whether it happens on terms that communities had meaningful input into, or on terms set entirely by the companies doing the building and the officials eager to announce the investment.
Early engagement — before the permits, before the groundbreaking, before it becomes much harder to change anything — is what makes the difference.
If you're in western Pennsylvania and want to get involved or stay informed:
PennEnvironment. Pennsylvania's statewide environmental advocacy organization. Tracks industrial development and environmental policy across the state.
Clean Air Council. Philadelphia-based, but it's Philly's oldest environmental nonprofit and active on regional air quality issues across Pennsylvania. It's been fighting the good fight since even before the EPA was established.
PublicSource. Pittsburgh's nonprofit investigative journalism organization. Has done significant reporting on environmental justice issues in the region.
Beaver County Marcellus Awareness Community. Local organization with experience tracking industrial development in Beaver County specifically.
Pennsylvania Environmental Council. Statewide policy organization focused on environmental protection.
Staying engaged with local municipal and county government meetings — particularly zoning and planning boards — is where the most consequential decisions often get made with the least public awareness.
Don’t forget to follow Kellee for the latest news and music:
add soundcloud, IG
