World Water Day And The Hidden Water Footprint Of AI

World Water Day And The Hidden Water Footprint Of AI

News ClipYahoo News Australia·Saline, Oscoda County, MI·3/20/2026

A planned $7 billion Stargate data center in Saline, Michigan, is facing strong local opposition over concerns about increased electricity rates and strain on the water supply. This local conflict highlights the broader issue of the growing water footprint of AI and cloud computing infrastructure across the US, prompting discussions among policymakers about sustainable resource management.

waterelectricityopposition
MetaGoogle
Gov: Newton County
Residents in rural Saline, Michigan, are actively protesting against the proposed $7 billion Stargate data center, alleging that DTE Energy is fast-tracking the project, which they fear will lead to higher residential electricity rates and endanger the local water supply. This local opposition underscores a national and global concern regarding the significant and increasing water consumption of the digital economy, particularly driven by cloud computing and generative AI. The article highlights that every AI prompt has a water footprint, with data centers requiring substantial electricity and cooling infrastructure. For instance, a Meta data center in Newton County, Georgia, consumes an estimated 500,000 gallons of water daily, representing about 10% of the county's total daily usage. Nationally, US data center water consumption rose from 21 billion liters in 2014 to 66 billion liters in 2023. Data centers typically use evaporative cooling towers, drawing from municipal freshwater, to manage the heat generated by servers. While some companies are exploring treated wastewater or closed-loop systems, these alternatives are not widely adopted. Projections indicate that annual US AI server water use could reach between 731 and 1,125 billion liters by 2030, with a significant portion attributed to electricity generation rather than on-site cooling. Analysts foresee a doubling or quadrupling of US data center water consumption by 2028, potentially exposing nearly half of existing facilities to high water stress by the 2050s. Community concerns are emerging in various countries, including the United States, as residents and policymakers question whether the economic benefits of hyperscale facilities outweigh long-term freshwater withdrawals. The environmental impact extends to the intensive water requirements for training large AI models. As global policymakers begin to integrate discussions about AI into water governance, there is a growing call for greater transparency, planning, and accountability to manage the digital economy's tangible water footprint responsibly.