India’s operational data-centre capacity stands at roughly 1.4–1.6 GW (2025) and is projected to reach about 5 GW by 2030. This expansion is backed by close to $30 billion in committed investment. The current footprint is spread across 300 facilities in 33 markets, with very large, new hyperscale clusters announced in Andhra Pradesh and Maharashtra. Many cities where these data centres are located are already water-stressed, which is raising concerns. AI data centres tend to rely on water-intensive evaporative systems in cooling towers. The quantum of water needed depends on ambient temperature and humidity.

Data centres in India typically consume between 1.5 to 2.5 liters of water per kWh of IT load. A mid-sized 20 MW centre can use approximately 1-1.2 million litres of water daily (MLD), depending on cooling technology and climate conditions. This is not actually a large volume of water, but the demand from multiple hyperscale data centres (>100 MW) in a single city can quickly add up. Importantly, demand will be highest in the hottest and driest months. At a city scale, a data-centre cluster could eventually account for 5-10% of municipal water demand, albeit only 1-2% of basin water use. While the basin-level or national footprint looks trivial, at the level of a single stressed city, the demand would compete with supply for domestic use. Conflicts are likely to arise in drought years.

Also Read | From Michigan to Mumbai, communities are asking: How much water do data centres use?

As water........

© hindustantimes