Editorial note by Aarthi Padmanabhan, Northern News Associate Editor, October 28, 2025
Paid access
A paid subscription to APA’s Zoning Practice publication is required to access the full article.
While artificial intelligence holds genuine promise for addressing environmental challenges—from monitoring methane emissions to tracking biodiversity loss—we must recognize that AI itself carries substantial environmental costs that often remain invisible to users.
AI’s infrastructure depends on data centers that consume massive amounts of electricity and water, rely on unsustainably mined rare earth minerals, and generate significant electronic waste. By 2028, AI could consume as much electricity as 22% of U.S. households, with inference operations (everyday AI use) now accounting for 80-90% of total computing power. The built environment bears much of this burden: data centers’ power density requirements can be seven or eight times higher than typical computing workloads, and the number of data centers has surged from 500,000 in 2012 to 8 million today. This rapid expansion of physical infrastructure – requiring construction materials, land use, and ongoing resource extraction – occurs largely in communities already facing environmental stress.
By raising awareness of APA’s recent Zoning Practice publication article, The Physical Footprint of Artificial Intelligence, Northern seeks to promote an evidence-based, constructive discussion of the net effects of AI on the planet as the technology begins to deploy at scale. This means honestly accounting for both AI’s potential benefits and its material costs so that users can make informed decisions about when and how to use it.