Urgent.News

What's breaking now, across thousands of outlets.

AI

Sam Altman on ChatGPT's Water Footprint: How Much Does It Use?

OpenAI CEO Sam Altman said that 38,000 ChatGPT queries use as much water as growing a single almond in California.

OpenAI's CEO Sam Altman has addressed doubts surrounding ChatGPT's water usage, disputing claims that each AI query consumes a significant amount of water. On the Sources podcast, Altman stated that around 38,000 ChatGPT queries equal the water usage of producing a single almond in California. While he couldn't provide the exact calculation, Altman felt the figure was close to accurate.

The public's growing awareness of AI's water consumption led to viral videos and headlines claiming that nearly every AI query requires a 500mL bottle of water. However, Altman argued that such comparisons do not accurately represent how modern data centers function. In the past, data centers relied more on evaporative cooling, which required substantial water. But he noted that today's large-scale data centers have shifted away from this method, making their water consumption more similar to that of an office building.

Altman compared AI's water use to that of an office, emphasizing that water is primarily used for everyday activities like sinks and toilets. By drawing a parallel between thousands of ChatGPT queries and the production of a single almond, Altman aimed to clarify that the water footprint of an individual AI interaction may be far smaller than commonly believed. He did not downplay the environmental impact of AI altogether, but sought to challenge the notion that each AI query carries a substantial water footprint.

Written by urgent.news from Tempo.co English's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at en.tempo.co →

More in AI

Give Your Coding Agents Proof Obligations, Not Instructions

Six AI agents wrote a complete screen of my cockpit in two hours. That is not the interesting part. TL;DR : code generation is no longer the bottleneck. Proof is.

  • Give coding agents proof obligations, not free instructions
  • Six specialized AI agents orchestrated for backend, design, verification, testing
  • Enforce environment with written rules like idempotent seeding and Docker restarts

More from Friday 4 September →