As data centers expand across Pennsylvania, highlighted by Amazon Web Services’ $20 billion investment in the Commonwealth, local communities must answer real questions about water and energy use, infrastructure, and long-term planning.

Training and running large AI models in data centers consumes substantial amounts of energy, which turns into heat. To keep servers from overheating and to increase power efficiency, many data center facilities use water for cooling.

The electricity they use is drawn from the grid, which relies on different sources depending on where they’re located. When that power comes from non-renewable sources, it often carries a hidden water cost, since extracting and generating power from these fuels requires water.

Water demand and effect

Data centers currently account for a small share of U.S. freshwater withdrawals. According to a 2024 Data Centers report to the Governor of Virginia, “most data centers use about the same amount of water (or less) as an average large office building (6.7 million gallons per year), although a few require substantially more, and some require less than a typical household.

Advertise with us

Even when overall water demand seems small, data centers can significantly affect local watersheds, depending on several factors: where the facility is located, local water availability, the cooling technologies used, the energy mix, and day-to-day operational practices.

The Water Center at Penn, which I direct, is co-leading the new Water-AI Nexus Center of Excellence with the Water Environment Federation, Amazon, and the global network called Leading Utilities of the World. The initiative focuses on two goals: making AI infrastructure water-efficient while harnessing AI to address critical water challenges.

A recently released report from The Water and the AI Nexus Center of Excellence, “Principles for Sustainable Water Use by Data Centers,” outlines four core strategies Pennsylvania should adopt.

Four strategies

Smarter design and siting. The impact of water use differs based on location, and the choice of data center cooling technology should reflect the relative scarcity of water resources and the capacity of water infrastructure to handle an increase in water use in any given location.

Data center locations should be sited where water resources can support them without harming watershed health or overly burdening aging water infrastructure — favoring air or hybrid cooling systems in water-stressed areas. Where power is constrained and local water resources are not stressed, water-efficient evaporative cooling designs help reduce both power demand and water consumed in the power generation process.

Advertise with us

Cut water use through transparency and technology. Require reporting of Water Usage Effectiveness (WUE) which measures liters of water used per kilowatt-hour of IT energy. WUE varies by cooling design and by location, with warmer climates requiring more water to maintain lower air temperatures in data centers.

Even in warm climates, deploying AI-enabled monitoring systems with thousands of sensors that monitor temperature, airflow, and water quality can determine water needs and volumes at minute-level precision to continuously improve efficiency. Leading operators have cut water intensity by 40% in four years through such approaches.

Prioritize water reuse and sustainable sourcing. Incorporate closed-loop liquid cooling systems, maximize reuse of cycles of concentration (CoC) water in evaporative cooling systems, and alternatives like recycled water or on-site rainwater catchment to reduce reliance on potable supplies.

Engage communities early. Collaboration with utilities, municipalities, and watershed groups ensures better planning and local benefits. Resources like PennFuture’s Data Centers series give municipalities the information and tools they need to make informed decisions —  including a model zoning ordinance with suggested provisions addressing water, power, noise, and aesthetic concerns.

In addition, water replenishment initiatives such as watershed restoration, WASH investments, irrigation optimization, and utility-scale water efficiency measures should be implemented through careful project selection, use of established quantitative methods, and project monitoring to lower regional water stress.

Advertise with us

Pennsylvania’s turning point

Pennsylvania is at a turning point. With smart planning, data centers can drive innovative water infrastructure development that meets industrial needs without raising costs for local residents.

Every $1 million invested in water infrastructure creates about 15 jobs, generating ripple effects across communities. Strategic zoning and land use can protect neighborhoods while building capacity for broader economic growth beyond technology.

By incorporating these principles into state permitting, incentivizing water reuse, and supporting AI adoption by water utilities, Pennsylvania can leverage its abundant freshwater resources to attract data centers that bring thousands of jobs to communities while also creating modernized water systems that benefit entire regions for decades.

Howard Neukrug is Executive Director of The Water Center at the University of Pennsylvania and former CEO and Commissioner of Philadelphia Water.