AI environmental impact and resource consumption

The rapid expansion of artificial intelligence has sparked debates regarding its environmental footprint, specifically concerning the water required to cool data centers. While some public discourse has utilized humor to highlight these concerns, experts emphasize that treated wastewater serves as a viable industrial cooling solution.

As the conversation has evolved, scrutiny has turned toward the precision of water consumption statistics. While some claims suggest a single query consumes an entire bottle of water, research indicates the actual footprint is much smaller, such as approximately 6mL for writing an email. However, calculating exact usage remains difficult due to a lack of transparency from AI companies regarding model energy requirements. Some researchers have also noted that the intense focus on AI water usage may overshadow other high-consumption industries, such as meat production.

OpenAI CEO Sam Altman has recently pushed back against viral narratives, labeling concerns about the water footprint of data centers as an ‘unfounded meme’. Speaking on the Sources Podcast, Altman challenged the idea that a single ChatGPT query consumes as much water as a six-hour shower, suggesting instead that 38,000 queries consume an amount comparable to the water required to produce a single almond in California. He argued that modern cooling technologies have reduced reliance on evaporation, making facility consumption levels similar to standard office building plumbing.

Despite these defenses, critics continue to highlight the lack of official electricity and water usage figures from OpenAI. This ongoing opacity has prompted legislative efforts in certain regions to mandate that data centers publicly disclose their resource consumption.