Drinking AI: AI/Water Consumption Comparisons to Beverages (maxghenis.com)

A new study has revealed that using AI models like ChatGPT for 30 queries a day over a year consumes about as much water as a single beer, highlighting the significant water usage associated with AI technologies. The research indicates that while AI systems are predicted to consume between 312.5 billion to 764.6 billion liters of water in 2025, the total water footprint of beverage production in the U.S. alone is estimated at 75.6 trillion liters annually, showcasing that conventional beverages still have a much larger environmental impact. The models consider factors such as cooling requirements for data centers and water usage from power plants, which are crucial for understanding AI's broader ecological footprint. This analysis emphasizes the growing environmental concerns surrounding AI and its power consumption as the technology scales. As AI usage is projected to increase significantly in the coming years, with estimates suggesting water consumption could reach 1.1 trillion liters by 2026, there are implications for sustainable practices in technology and environmental management. The study calls attention to the need for data-driven insights, as various measures of water consumption in AI range widely, depending on the definitions used, highlighting a critical gap in understanding the ecological ramifications of increasingly powerful AI systems.