India’s data centre buildout has become a showcase of its digital ambition, but it is increasingly colliding with the hard limits of urban water and power systems. Hyperscale facilities that keep cloud, AI and OTT services running are being clustered in already stressed metros such as Mumbai, Chennai, Hyderabad, Bengaluru and Delhi-NCR, where utilities are struggling to balance residential demand, industrial growth and climate-driven variability in supply.
Over the past few years, India’s installed data centre capacity has more than doubled, with operational IT load crossing roughly 1.2–1.3 GW by mid-2025 and additions of about 387 MW recorded in 2025 alone. Mumbai remains the epicentre, accounting for nearly half of national absorption, with Chennai and Delhi-NCR following as secondary hubs, while Hyderabad is rapidly positioning itself for the next wave of hyperscale investments. Policy incentives, submarine cable landings, and proximity to financial and digital services fuel this growth—but they also concentrate water and power-hungry infrastructure in some of India’s most contested urban geographies.
The scale of the “digital thirst”
Cooling is at the heart of the problem. A single 1 MW data centre can consume about 26 million litres of water annually, primarily for evaporative cooling and maintaining humidity thresholds. Extrapolated to a large 30 MW facility, that translates to nearly 780 million litres a year—roughly equivalent to the domestic water needs of more than 15,000 urban residents, based on a benchmark of 135 litres per capita per day. At full build-out, some hyperscale parks planned at 150–160 MW could use more than 4 billion litres a year, rivaling the consumption of a mid-sized Indian town.
Market studies estimate that data centre water use in India will reach about 150 billion litres in 2025 and more than double to nearly 360 billion litres by 2030, driven by AI workloads and rising digital penetration. This trajectory is unfolding in a country where nearly two-thirds of districts already experience varying degrees of water stress and where urban utilities face chronic leakage, seasonal shortages and rising demand from households and manufacturing.
When cloud clusters meet parched cities
The risks are most visible where data centre parks come up in water-stressed micro-regions. In western Uttar Pradesh’s Greater Noida belt, for instance, mega facilities promoted as anchors for “data centre cities” have triggered concerns over falling groundwater tables, even as nearby villages grapple with irregular municipal supply. Similar tensions are emerging around planned facilities in Visakhapatnam, where civil society groups have flagged that the city already struggles with summer shortages yet is being asked to support additional bulk industrial demand from data infrastructure.
In Maharashtra and Tamil Nadu, where monsoon variability and heatwaves are intensifying, data centres compete directly with domestic and agricultural needs for surface and groundwater. Many small and mid-sized facilities still rely heavily on municipal pipelines or private tanker supplies, transferring the burden to already strained urban grids and informal extraction from peri-urban aquifers. As the number of AI-ready sites grows, this silent shift of water from people to servers becomes an increasingly political question rather than a purely technical one.
Energy decarbonisation vs. water efficiency
Operators often point to aggressive renewable power procurement and PUE (power usage effectiveness) improvements as evidence that India’s data centre boom can align with net-zero commitments. However, global studies show a clear trade-off between energy and water efficiency: many of the most energy-efficient cooling systems, such as evaporative and adiabatic setups, tend to be far more water-intensive than traditional chiller-based designs. As a result, pushing facilities to the cutting edge of energy performance can inadvertently increase their draw on local water resources unless explicitly constrained.
This trade-off is particularly acute for AI-heavy workloads. High-density racks running GPUs generate far more heat than conventional server farms, nudging operators toward advanced cooling architectures that may either consume more water or demand new capital-intensive technologies such as liquid immersion or rear-door heat exchangers. Without clear national benchmarks for both energy and water use, operators are free to optimise for whichever metric enhances their sustainability narrative, even if it shifts other environmental burdens onto local communities.
Transparency gaps and “green” claims under scrutiny
One of the most contentious issues is the lack of transparent, independently verified reporting on water usage and sourcing. Several large operators in India have announced “water-neutral” or “water-positive” ambitions, often citing rainwater harvesting, wastewater recycling or offset projects. Yet investigative reporting has highlighted that many of these claims are based on internal calculations, with limited third-party validation or granular disclosure on how much freshwater is actually drawn, how much is recycled on-site, and how much is offset elsewhere.
This opacity is compounded by the fragmented regulatory landscape. Water supply contracts may sit with state utilities, industrial development corporations or municipal bodies; impact assessments are often limited to project clearances rather than ongoing performance; and there is no common disclosure standard on water use comparable to energy reporting frameworks. As climate risks intensify, governments could come under pressure to tighten permitting norms, enforce water accounting and, in extreme cases, revoke or restrict licenses for facilities in high-stress basins.
Searching for a sustainable model
The industry is not without options. Closed-loop cooling systems, higher reliance on treated wastewater instead of potable supplies, and strategic siting of new facilities in water-abundant regions are among the measures being explored in India and globally. Some state governments are beginning to condition approvals on commitments to use recycled water, integrate rooftop and surface harvesting, or co-locate with industrial clusters where treated effluent is available for non-potable applications.
At the same time, national digital infrastructure plans, semiconductor and AI missions, and state data centre policies have yet to fully internalise water risk as a central design parameter rather than a secondary environmental clearance issue. As India races to triple or even quadruple capacity by 2030, the question is shifting from whether the country can attract cloud and AI investments to whether it can do so without undermining the liveability of the very cities that anchor its digital economy.
