The latest water resources news, articles, and practices


The Thirsty Machine: Why AI Is a Water Resources Problem, Not Just a Tech OneAI runs on water, not just code. Every prompt generates heat that data centers shed by evaporating freshwater, and the power feeding those servers consumes still more at the plant. A June 2026 UN University report put global data center water use at 4.5 trillion liters in 2025, on track to reach 9.3 trillion by 2030. This post breaks down where the water actually goes, why the viral "bottle per prompt" figure is a system-boundary problem rather than a fact, how data centers are triggering real resource conflicts in already water-stressed basins, and why "water positive" pledges deserve a hydrologist's skepticism when water, unlike carbon, is always local. The takeaway for water resources engineers: a large data center is a new consumptive demand on a basin and should be evaluated with the same rigor as any other major withdrawal. If you want it shorter for a CMS excerpt field, here is a tighter version: AI is a water problem, not just a tech one. Data centers consumed 4.5 trillion liters of freshwater in 2025 and could reach 9.3 trillion by 2030. A water resources engineer's look at where the water goes, why the local basin is what matters, and why "water positive" pledges need scrutiny. Want me to also draft an SEO title tag and a set of focus keywords to go with it for the Webflow entry? 