A widely shared claim that a single ChatGPT email prompt uses a bottle of water is false; a simple request uses 6mL, or about 1% of a standard bottle. While AI does require water, its footprint per query is small, though calculating total usage is challenging due to a lack of transparency from AI companies. Independent researcher Andy Masley, a skeptic of the “one water bottle per prompt” figure, argues that AI’s water use receives outsized attention compared to other industries. He contrasts data centers with livestock facilities, noting one uses roughly 1,000 times more water. Producing one kilogram of pork takes about 1,078 liters of water, and a kilogram of beef consumes 925 liters from limited resources. Masley questions why society cannot think seriously about these everyday food choices. The article notes that AI’s water use remains a concern, especially at an industrialized scale by large corporations, but should not overshadow other water-draining decisions like food production. Data centers use water to cool heat-generating equipment, but cooling efficiency depends on several factors, which partly explains why precise water usage figures are difficult to determine.