
Pennsylvania confronts AI data center growth amid electricity and water concerns
The growing demand for AI is creating a significant physical footprint through data centers, especially in Pennsylvania, raising concerns about electricity and water consumption. Bucknell University professors highlight the strain on the power grid and local aquatic environments. Governor Josh Shapiro has issued an executive order requiring developers to cover new infrastructure costs and ensure environmental protections.
Artificial intelligence's rapid expansion is creating a substantial physical footprint through data centers, with numerous proposals emerging across Pennsylvania. This development is sparking significant debate about the enormous electricity and water demands these facilities require, along with their associated development costs and local impacts, making it a key political issue.
Three experts from Bucknell University emphasize that AI's capabilities are reliant on massive amounts of electricity, computing equipment, and potentially water. Professor Brian King of computer science notes Pennsylvania's appeal to developers due to available land and energy resources. However, he warns that AI is dramatically increasing computing demands, projecting data centers could consume up to 12% of all U.S. electricity by 2028, up from 4.4% in 2023, potentially stressing the state's robust grid. King suggests that using smaller, specialized AI models could substantially reduce power usage.
Beyond electricity, Benjamin Hayes, program director of watershed sciences & engineering at The Ecology Center, points to the environmental impact of water consumption. Modern data centers require continuous cooling, using large quantities of water. While national usage might be small, Hayes states that local impacts can be severe, with some facilities demanding up to a million gallons daily, potentially exceeding available local resources. This concern is compounded by the fact that both data centers and the power plants generating their electricity require significant water for cooling.
Professor Lily Li, who studies electrical-grid demand response, explores the potential for more distributed computing loads to alleviate infrastructure costs and time, though this raises coordination challenges. In response to these growing concerns, Governor Josh Shapiro has issued an executive order mandating that developers cover new generation and grid costs, rather than shifting them to other ratepayers. The order also requires environmental protections, transparency, and community involvement for data center projects. The overarching challenge, as King describes, is to balance AI's transformative capabilities with its significant, often unseen, environmental and infrastructure costs.