Festival greetings now reach millions of screens in an instant, but behind every image, video, and voice note, data is processed and routed through data centres before it reaches a phone. What most people never consider is the environmental cost of that digital convenience: the electricity drawn, water consumed, and heat released into already warming cities. Data centres are the foundational layer of the modern digital economy. They process everything from billions of UPI transactions each year to video streaming to management of smart grids and Internet of Things (IoT) devices. As artificial intelligence (AI) becomes more common, data centres grow in importance, as does their environmental impact.
Environmental impact of data centres
The energy challenge: India’s data centres consumed approximately 13 TWh of electricity in 2024, i.e., roughly 0.8% of the total national demand. That sounds reasonable until we look at the trajectory. By 2030, S&P forecasts consumption will reach 57 TWh, approximately 2.6% of the national electricity demand. This amount is comparable to what a mid-sized European country uses in an entire year (Kang & Chauhan, 2025). Furthermore, because coal still fuels a significant portion of India’s electricity generation, every new megawatt of data centre capacity carries a carbon penalty.
The water strain: In 2025, Indian data centres consumed over 150 billion litres of water, a figure projected to reach 359 billion litres by 2030 (The Business Standard, 2025). This amount is equivalent to the annual water needs of roughly.[1] In Bengaluru, data centres collectively consume roughly 8 million litres per day, i.e., roughly 0.3% of the city’s total demand. While this is a relatively small share, it reflects rapidly growing industrial demand in a city already facing a daily water deficit of nearly 500 million litres and experiencing one of its most severe water shortages in recent decades (The Indian Express, 2024). In Devanahalli, groundwater extraction stands at 169% of permissible limits, yet eight data centres have been approved for construction in the area (Krishnamurthy, 2025).
The heat problem: Almost all the electricity used by servers is converted into heat—a basic rule of thermodynamics. As data centres are located in areas such as Navi Mumbai and the Chennai IT corridor, the extra heat contributes to warming and exacerbates the urban heat island effect.
Solutions that already work at scale
Renewable energy integration: Integrating renewable energy into data centre operations is necessary to reduce the sector’s carbon footprint. Globally, companies have moved beyond basic annual renewable targets and are now matching clean energy generation to hourly electricity demand by using a mix of solar, wind, and battery storage. In India, solar and wind capacity are growing quickly enough to make this shift possible. However, most facilities still rely mainly on the grid, which faces challenges such as unreliable open-access rules, restrictions on buying power across states, and concerns about inconsistent supply. These issues are real but can be addressed. Combining solar and wind in hybrid portfolios can help smooth out generation, and adding battery storage and direct corporate power purchase agreements can serve as practical solutions. Still, combining renewables with battery storage alone may not provide cost-effective, round-the-clock clean power, especially for large AI data centres that need steady baseload energy. Small modular reactors (SMRs) can help fill this gap. Amazon, Microsoft, and Google have all announced plans to include SMRs in their future energy mix. India is also making progress: the 2025–26 Union Budget set aside INR 20,000 crore under the Nuclear Energy Mission to bring at least five locally designed SMRs online by 2033. For data centre operators, SMRs offer a reliable way to secure firm, dispatchable, clean power at scale, as long as regulatory and licensing processes keep up.
Efficient cooling infrastructure: What is still missing from a policy standpoint is a mandatory PUE disclosure and benchmarking framework. Other markets have moved ahead. For instance, the EU requires annual PUE reporting for any facility above 500 kW, and Singapore now approves new data centres only if they obtain a PUE of 1.25. Without such policy measures, there is no market mechanism to distinguish efficient operators from inefficient ones, and no basis for regulators to set performance benchmarks.
Strategic siting: The choice of location also determines the bulk of a facility’s lifetime environmental footprint. Today, more than two-thirds of India’s data centres are located in four cities—Mumbai, Chennai, Hyderabad, and Bengaluru—that all face strained power grids, falling water tables, or both (CBRE, 2025). Strategic site selection can therefore play a decisive role in addressing the issue. For instance, in Europe, enormous quantities of low-grade heat from data centres are treated as a resource rather than vented into the atmosphere: Meta’s Odense campus in Denmark channels roughly 165,000 MWh of waste heat annually into the local district heating network—enough to warm approximately 11,000 homes. India’s warm climate rules out residential heating as a use case, but industrial applications warrant exploration. Policymakers should treat data centres not as standalone IT assets but as anchor tenants within eco-industrial zones by co-locating them with food-processing, textile, and pharmaceutical facilities that can absorb low-grade waste heat. They should also link siting approvals to renewable-energy procurement, water-recycling standards, and land-use frameworks that account for thermal synergies.
The way forward
India’s data centre expansion is inevitable and, overall, a positive development. Building domestic capacity is crucial for data sovereignty and digital infrastructure. Incentives such as infrastructure status since 2022, tax holidays until 2047, and concessions on state-level land and power are rapidly attracting investment. However, the sustainability framework has not kept pace. There are still no binding rules for renewable-energy use, energy efficiency, or water limits for private operators. While incentives exist, environmental protections are missing.
Endnotes
[1] Estimated by dividing projected data-centre water consumption in 2030 (359 billion litres) by the annual domestic water requirement of an average Indian household. Assuming 135 litres per capita per day (Government of India urban water supply norm) and an average household size of four persons, annual household consumption equals approximately 197,100 litres (135 × 4 × 365). On this basis, 359 billion litres is equivalent to the annual domestic water needs of roughly 1.8 million households.
By Kareena Jaisinghani, Senior Analyst, Climate and Sustainability Initiative (CSI). Views expressed are personal.
