Short Wave What's The Environmental Cost Of AI?
56 snips
May 7, 2025 The podcast dives into the staggering water consumption of AI data centers, highlighting their cooling demands. It reveals that tech giants like Google are using billions of gallons, prompting urgent discussions about sustainability. Insights into the evolution of cooling systems, from air to water, showcase ecological concerns. Additionally, the ambitious Stargate project underscores the energy demands of AI infrastructure. The show stresses the need for the tech industry to adopt greener practices as it grapples with its environmental impact.
AI Snips
Chapters
Transcript
Episode notes
Water Scarcity Shapes Research
- Xia Leiren grew up in a water-scarce coal mining town in northern China, learning to use every drop wisely.
- His early life shaped his water-consciousness and led him to study tech industry water footprints at UC Riverside.
AI's Huge Water Demand
- Data center computers get extremely hot and require effective cooling, often using chilled water.
- Training large AI models and cooling data centers can consume hundreds of thousands of liters of fresh water.
Water Cooling Grows with AI
- Data centers grew from single rooms to massive buildings filled with thousands of computers needing cooling.
- Many use water cooling to reduce electricity demand; water evaporation from cooling towers leads to high water consumption, increasing with AI usage.