150 billion litres water a year: India wants to be AI superpower, but it comes with a bill

Published 14 June 2026 · tech

Artificial intelligence (AI) is rapidly emerging as one of the most transformative technologies of the 21st century, reshaping industries, economies and everyday life at an

Artificial intelligence (AI) is rapidly emerging as one of the most transformative technologies of the 21st century, reshaping industries, economies and everyday life at an unprecedented pace. From drafting simple emails and generating lifelike images to powering advanced chatbots, deep research, and decision-making systems, AI is increasingly influencing how people work, learn, communicate and consume information. But, it has a cost, and this one isn't simply in dollars or rupees. Every AI prompt costs electricity, land, minerals, and enormous quantities of water. Read Full Story A recent report by the United Nations University Institute for Water, Environment and Health (UNU-INWEH) warns that the environmental footprint of AI is growing rapidly as the technology expands across the world. A man walks through an AI data centre. (Photo: Reuters) While much attention has focused on energy use and carbon emissions, another resource is increasingly coming under pressure: water. Water scarcity is already one of the biggest emerging concerns in the world's most populous country, and yet India is setting itself up to become one of the world's busiest AI hubs. But it also needs to consider the water cost that will come along with the rapid expansion of the AI infrastructure. WHY DOES AI NEED WATER? Every AI query begins inside a data centre, which is essentially a vast warehouse packed with servers that store, process and transmit information. Every prompt you enter into an AI chatbot to ask something, write something, create an image, or research, the system generates enormous heat to meet your demand and must be cooled continuously to keep functioning. Many data centres use evaporative cooling systems, where water absorbs heat from servers and is then released into the atmosphere. A substantial portion of that water is lost through evaporation during the cooling process.

Servers, networking devices and cables are seen in a data center. (Photo: Pexels) As AI models become larger and more powerful, the amount of cooling required also rises sharply. The scale of that increase is rather alarming. According to the UNU-INWEH report, training ChatGPT-4 likely required around 592 million litres of water. While training large AI models is resource-intensive, experts note that much of AI’s environmental footprint ultimately comes from the billions of interactions that take place after a model is deployed. According to a 2026 report by the Council on Energy, Environment and Water (CEEW), a typical 100-megawatt (MW) hyperscale data centre, which has thousands to hundreds of thousands of servers, can consume around 20 lakh litres of water per day for cooling, although actual consumption varies depending on cooling technology and local conditions. "The current industry standard for large-scale AI infrastructure is the 100-MW “AI factory” — hyperscale facilities designed specifically to support intensive AI workloads," said Daswin De Silva, Professor of Analytics and AI at La Trobe University, Australia. A technician works at an AI data center. (Photo: AP) As the industry shifts towards even larger MW-scale AI hubs, demand for both water and energy is likely to rise significantly, De Silva added. Estimates also show that India’s data centres consumed roughly 150 billion litres of water in 2024-2025 and could require around 358 billion litres annually by 2030 as digital infrastructure expands. "Water demand varies widely depending on design, cooling architecture and local climate conditions," noted Guillaume Dourdin, CEO of Veolia India, a multinational company specialising in waste and water management. Citing Deloitte estimates, he said that a 1-MW data centre can consume approximately 68,500 litres of water per day.