NewsBite analysis · AI, Cloud, GreenTech, Hardware, Regulation
Toms Hardware
AI data center project secretly sucked 29 million gallons of water over 15 months before detected by residents complaining about low water pressure published 10 May 26
An AI data center in Georgia, part of the growing infrastructure for artificial intelligence, consumed 29 million gallons of water over a 15-month period without public disclosure. The excessive water usage was only detected when nearby residents reported low water pressure, leading to an investigation. The incident underscores the often-overlooked environmental footprint of data centers, particularly water consumption for cooling systems. This case may prompt stricter regulations on resource usage and disclosure requirements for tech infrastructure projects, especially those tied to AI expansion.
MonitorConfidence: MediumShort term horizonReputational risk: High
Published 10 May 2026, 12:46 UTC1 min read
Visit original article ↗Lead image — Toms Hardware
Summary
TL;DR
- 29 million gallons of water consumed by a single data center in Georgia over 15 months.
- Detection occurred only after residents complained about low water pressure.
- The data center is described as an 'AI data center,' indicating high-performance computing usage.
- No prior public disclosure of water usage or environmental impact was made.
- The event raises questions about oversight of resource consumption in the tech industry.
Interesting facts
- Water consumption from data centers is often not disclosed in public environmental reports.
- AI workloads, especially training large models, generate significant heat requiring cooling, often via water evaporation.