New York commentary urges contextual analysis of AI and data center resource consumption
This commentary argues against public panic regarding AI and data center resource consumption, particularly water and electricity, by providing context to usage statistics. It highlights that while local impacts are valid concerns, national-level numbers are often presented without proper scale, leading to misleading fear. The author urges a more analytical approach to discussions on regulating these technologies.
Potsdam-based mechanical engineer Layton Colbert authored a commentary in the Times Union, urging a contextual approach to the public debate surrounding artificial intelligence and data center resource consumption. Colbert argues that while concerns about electricity, water, land, and grid capacity are legitimate, discussions often present large usage statistics without comparing them to overall national consumption or other industries.
He cites, for example, U.S. data center water consumption at approximately 449 million gallons per day, which, while seemingly large, represents only about one-seventh of one percent of total U.S. water withdrawals, dwarfed by irrigation usage. Similarly, data centers' 2023 electricity use of 176 billion kilowatt-hours (4.4% of national use) is less than residential and commercial air conditioning combined.
Colbert acknowledges that local impacts, such as strain on aquifers or municipal systems, are valid concerns that communities should address. However, he contends that broad statements like "AI uses too much water" lack analytical depth. He concludes that regulatory efforts should focus on the realities of software, electricity, and engineering systems rather than "science fiction."