Hidden environmental cost of powering growing digital and AI-world

Hidden environmental cost of powering growing digital and AI-world

Indicating an invisible water footprint behind instant responses generated by systems such as ChatGPT, a research from the University of California, Riverside, estimates that a typical conversation with ChatGPT, involving around 20 to 50 exchanges, can consume as much as half a litre of water — a hidden environmental cost of rising dependence on artificial intelligence (AI).

Evidence and Data

Despite this, some States, which already have difficulties in meeting their water needs, have announced huge data centre projects. Some States have announced incentives to attract companies to set up their data centres. Google, in partnership with AdaniConneX, is coming up with 1GW AI hub in Vishakapatnam and Bharti Airtel’s Nxtra, and Meta’s AI-enabled data centre in partnership with Reliance Industries in Jamnagar. Hyderabad has been officially designated over-exploited by the CGWB. One solution many companies are now looking at is cooling data centres with seawater instead of freshwater. It is now becoming an important part of India’s plan for building data centres along the coast.

The electricity consumption of data centres had a water footprint of 4.5 trillion litres of water in 2025, enough to meet water needs of Chennai for 8.8 years (approximately), according to an UN report. Chennai’s water requirements currently stand at 1,407 million litres of water per day. The report predicts the water footprint could reach 9.3 trillion litres of water by 2030. Andhra Pradesh, under its Data Centre Policy 4.0, offers 100% state GST (Goods and Services Tax) reimbursement on capital goods, 100% stamp duty exemption on first sale, and deemed distribution licenses for direct energy purchase for data center projects with a minimum capacity of 300 MW. The CGWB’s 2024 Dynamic Ground Water Resources assessment puts India’s overall stage of groundwater extraction at 60.47%, with 11.1% of all assessment units nationwide already classified as “over-exploited. Gautam Buddha Nagar in Uttar Pradesh is now home to at least 17 operational or upcoming data centres, and has a stage of groundwater extraction of 104.79%, per the CGWB’s assessment — meaning the district draws more water every year than nature replaces, according to Data Centre Map. Andhra Pradesh looks comfortable at the state level (29.83% extraction), but Visakhapatnam — where Google and Adani are building the flagship AI campus — had just 2.12 thousand million cubic feet of groundwater in reserve as of April 2026, the lowest of any district in the State. Google has used Baltic seawater to cool its data centre in Finland for more than 10 years.

What It Means

However, these projects have been met with resistance as several protests were held in Vishakapattinam, raising concerns on the impact on environment and the pressure on resources such as electricity and water. Similar protests were held in Thane, Maharashtra, to oppose Amazon’s proposed hyperscale project.