Amazon faces major allegations after a recent Rolling Stone report linked clusters of rare cancers and miscarriages near its U.S. data centres to enormous use of water and nitrate-tainted wastewater. The company denies that its AWS (Amazon Web Services) facilities are responsible for long-standing local groundwater issues.
Meanwhile, India’s southern states are seeing a surge of deals worth billions of dollars as global tech giants race to expand cloud infrastructure. Google signed an ₹87,500 crore agreement to build an AI hub in Visakhapatnam, its largest such investment in India, slated for completion between 2026 and 2030. OpenAI, the maker of ChatGPT, is exploring local partnerships to set up a data centre of at least 1 gigawatt (GW).
Domestic firms such as Yotta, CtrlS are also tapping AI-focused capacity across Chennai and Bengaluru. According to the National Real Estate Development Council (NAREDCO) report, South India’s data centre market could grow by 65% by 2030, supported by government incentives, strategic infrastructure investments, and rising demand for digital services.
What are AI data centres?
A data centre is a facility that stores I.T. (Information Technology) equipment such as servers, switches, storage systems, and routers. AI data centres are more advanced, handling high-performance computing that allows data to be processed and analysed at high speed. They rely on large numbers of CPUs (Central Processing Units), GPUs (Graphics Processing Units), and memory chips to power operations.
Why south India is the hotspot for AI data centres
A knowledge paper by The Associated Chambers of Commerce & Industry of India (ASSOCHAM)–PwC on Tamil Nadu’s data centre and cloud vision underscores Chennai’s emergence as a prominent and successful data-centre destination. The city benefits from six major submarine cable landing stations, ensuring high-speed, reliable connectivity.
Key Chennai suburbs such as Ambattur and Siruseri are supported by a regular and uninterrupted power supply and an influx of skilled workforce. Chennai is expected to contribute about a quarter of India’s new data centre capacity over the next five years, reaching approximately 551 MW by 2030, with a Compound Annual Growth Rate (CAGR) of over 22%.
“Government incentives, such as subsidies on land acquisition, tax exemptions play a major role in selecting locations for a data centre,” said Manjul.S, Deputy Vice-President of Kotak Mahindra Bank. Mr. Manjul is also a DevOps professional, specialising in cloud and platform engineering.
The immense water demands of the ‘cloud’
Data centres generate enormous heat from constant server operations, making water-intensive cooling essential. The International Energy Agency (IEA) estimates AI data centres’ water consumption at over 560 billion litres. With AI workloads increasing, this figure could hit 1,200 billion litres by 2030.
“Outages can happen just because of high temperatures. Keeping these systems cool is critical to any data centre,” Mr. Manjul said.
“Most of the opex (Operational Expense) of the data centres go towards cooling of these systems (machines),” he added.
‘Water-positive goal’
Tech giants claim to pursue water-positive goals through replenishment projects. Microsoft aims for 100% renewal by 2030 via watershed restoration, while Google invests in recycling and desalination.
Amazon has committed to water neutrality by 2030, through aquifer recharge and more efficient infrastructure at select sites. AWS estimates these measures will preserve more than 530 million gallons of drinking-water supply.
“‘Sustainable’ and ‘green’ are just fancy words of the decade. Setting up rainwater harvesting and then sucking out gallons of water from reserves is not a water-positive move,” said G. Sundarrajan of Poovulagin Nanbargal, a Chennai-based NGO working on socio-environmental issues.
Replenishment often falls short in water-stressed regions, as upstream projects rarely offset local depletion near facilities.
The silent toll on communities
India’s rush to build AI data centres comes at a time when a significant portion of India’s groundwater is already under severe stress. A 2024 report by Central Groundwater Board (CGWB) highlights that groundwater samples with uranium concentrations exceeding 30 ppb (parts per billion) are predominantly clustered in areas classified as over-exploited, critical, and semi-Critical groundwater stress zones, including regions of Tamil Nadu, Andhra Pradesh and Karnataka among others.
However, there are no official data quantifying how much of this critically stressed groundwater is used solely by AI data centres.
India’s 2025 digital push plan offers substantial incentives for data centre developers, including a 20-year tax holiday and Input Tax Credits (ITC) on GST for capital assets such as construction materials, Heating, Ventilation, and Air Conditioning (HVAC) systems, cooling systems, and electrical equipment. These measures lower the upfront costs of building and operating data centres.
“AI data centres are usually built in rural areas, where land and labour are cheaper,” said Mr. Sundarrajan. “These centres are often located in already water-stressed regions, which raises serious concerns among local communities. Neither the government nor companies disclose accurate data on the annual water usage of these AI centres,” he added.
As India opens its doors to more data centres, a pressing question remains: how will the country meet the enormous water demands of this rapidly growing digital infrastructure, when so much of its groundwater is already over-extracted or critically stressed?
Published – December 19, 2025 04:36 pm IST
