India’s data centre boom is colliding with its climate reality

Why in the News

India’s data centre capacity is set to grow from about 1.5 gigawatts (GW) today to 6.5 GW by 2030, a fourfold expansion in four years, on investment the government estimates at nearly $200 billion over the coming decade. Google, Meta, Amazon and Microsoft have committed billions to build facilities, and States are competing for them with tax exemptions, cheap land and duty waivers. No policy document at the Central or State level has assessed what guaranteed power costs a grid already strained under 45 degree Celsius heat, where cooling water will come from, or what the thermal load of thousands of servers does to regions already near dangerous temperatures. The tension is that the facilities are clustering in exactly the regions where water and grid stress are most acute, and State policies attach almost no conditions to the incentives they offer.

Why is the scale of the build-out itself the problem?

1. A fourfold expansion in four years: Capacity nearly tripled from 520 megawatts (MW) in 2020 to nearly 1.5 GW today. The 6.5 GW projected for 2030 compresses the next round of growth into four years.
2. Electricity demand more than quadruples: Demand from data centres is expected to rise from about 13 terawatt-hours (TWh) in 2024 to roughly 57 TWh by 2030. The Union Ministry of Power estimates that artificial intelligence alone will add 26.3 GW of new demand by 2031-32.
3. States compete on incentives, not conditions: Maharashtra wants to be the data centre capital. Telangana has declared data centres “essential services”. Karnataka is reviewing its policy to attract more, and Rajasthan is offering tax exemptions and cheap land.

Why does data centre water use collide with groundwater stress?

1. The volume per facility: A 100 MW data centre consumes about 2 million litres of water daily, equal to the daily use of roughly 6,500 households. India’s data centres consumed an estimated 150 billion litres in 2024-25, projected to more than double to 358 billion litres annually by 2030.
2. Clustering in the most stressed regions: Rajasthan extracts 147.11 per cent of its annual groundwater recharge, the second highest rate in the country. Several groundwater assessment units in Maharashtra are classified as semi-critical. Telangana’s Irrigation Minister confirmed in May 2026 that 16 districts were under groundwater stress.
3. Cities already rationing: Hyderabad’s surface water supply dropped 20 per cent in the summer of 2024 on poor monsoon recharge, forcing the water board to ration supply. Mumbai’s reservoirs stood at 44.5 per cent of capacity in March 2026.
4. Unaccounted in every State policy: No State policy requires public disclosure of daily water consumption. None requires a hydrogeological assessment before approval. None mandates that water sourcing must not compete with agriculture or municipal supply.

Why can the grid not deliver the power the boom needs?

1. Record peaks and interconnection queues: Maharashtra’s peak demand hit 27,230 MW in April 2026, the highest ever handled by the State utility. Two-year waits for 220 kV grid interconnections across the country are lengthening project timelines.
2. Renewable power is being thrown away: India curtailed 300 gigawatt-hours of renewable energy in the first quarter of 2026 alone because the grid could not carry it, per an Ember Energy analysis of Central Electricity Authority data. Over five years India has met only about 80 per cent of its annual transmission targets, one in four major transmission schemes runs a year or more behind schedule, and 20 GW of renewable capacity faces connectivity delays of more than four months.
3. Wires lag panels: Rajasthan and Gujarat house the bulk of utility-scale solar and wind. Both face the longest queues at pooling stations (substations where several renewable plants aggregate output before it enters the transmission grid). Solar projects are being built faster than the lines to carry their power.
4. The coal default: Data centres need reliable, uninterrupted electricity. If renewable power cannot reach them, the power comes from coal, adding to the emissions India is trying to curtail.

How do data centres make their surroundings hotter?

1. The satellite evidence: A March 2026 University of Cambridge study of 20 years of NASA satellite data found that data centres raise land surface temperatures by an average of 2 degree Celsius within a 10 km radius, with extreme cases reaching 9.1 degree Celsius. About 340 million people globally live within these affected zones.
2. Indian cities are already at the edge: Mumbai’s land surface temperature rose from 40.9 degree Celsius in 2003 to 47.3 degree Celsius in 2023, driven largely by urbanisation and heat-trapping infrastructure. Hyderabad’s urban heat island intensity ranges from 5.74 to 6.82 degree Celsius, its urban area doubled between 2001 and 2020, and it recorded temperatures above 43 degree Celsius in the summer of 2024.
3. The feedback loop: Data centres generate heat, and that heat raises ambient temperature. Higher ambient temperature increases cooling demand. Higher cooling demand raises electricity consumption. Unless that electricity is fully renewable, emissions rise and feed the climate change that is making India hotter.

Do State policies ask for anything in return for their incentives?

1. Generous on incentives: State policies offer electricity duty exemptions, transmission charge waivers, stamp duty relief and fast-track clearances.
2. Silent on conditions: None of the major State policies requires a grid impact assessment before commissioning, mandatory renewable energy sourcing, or a thermal load assessment for surrounding communities.
3. Telangana guarantees power in a stressed State: The “essential services” classification guarantees data centres uninterrupted power even during shortages, in a State where 16 districts face groundwater stress and temperatures have reached 47 degree Celsius.
4. Maharashtra diluted its own mandate: The State’s policy originally required 100 per cent renewable energy for core operations. In June 2026 it cut the requirement to 51 per cent, framed as improving project viability.
5. The exceptions, and their limit: Gujarat’s Data Centre Policy 2026-29 mandates at least 51 per cent green energy sourcing. Karnataka’s IT Minister told the Assembly in March 2026 that the State was reviewing its policy over water and energy concerns, and Tamil Nadu has linked incentives to renewable compliance. Even where mandates exist, enforcement and verification remain weak.

Why is the constraint market design rather than generation capacity?

1. Capacity is not the bottleneck: The Union Ministry of Power holds that India’s generation pipeline can absorb the additional demand from data centres. The constraint is market design and transmission infrastructure.
2. Price grid services separately: The Council on Energy, Environment and Water (CEEW) argues for climate-intelligent power markets in which short-term markets pay separately for ramping, storage and demand response.
3. Storage at the pooling station: Ember calculates that roughly 3 to 4 GW of two-hour battery storage at renewable pooling stations could have absorbed most of the generation curtailed in early 2026. The technical pieces exist; the gap is regulatory and commercial.

What should a national sustainability framework contain?

1. Enforceable use standards: CEEW proposes phased power and water use standards with enforceable benchmarks, and a national AI Energy Star rating that lets buyers and regulators compare how energy efficient a facility or model actually is.
2. Who pays for the grid: The Institute for Energy Economics and Financial Analysis (IEEFA) warns that a significant part of the associated infrastructure cost could be socialised. Where wider grid infrastructure is required, government support or dedicated financing should stop the cost being passed to consumers through higher tariffs.
3. Siting away from stressed hubs: IEEFA points to coastal locations. These offer proximity to near-shore wind and solar, and seawater cooling without desalination. Most facilities instead cluster around Mumbai, Hyderabad, Bengaluru and the National Capital Region, where water and grid stress are most acute.
4. Four minimum standards: A national framework would set minimum standards for renewable energy sourcing, water consumption disclosure, grid impact assessment and thermal load evaluation. It would give investors one set of expectations and consumers one set of protections, on the premise that environmental constraints are economic constraints.

Challenges to a national data centre framework

1. Split jurisdiction: Water is a State List subject and electricity sits on the Concurrent List, so a Central standard on water sourcing or grid impact binds no State unless the State adopts it. Eg. The Ministry of Electronics and Information Technology’s draft National Data Centre Policy of 2020 addressed infrastructure status and single-window clearance, not resource standards.
   The Fix: Route the standards through the Central Electricity Authority’s technical standards and the Bureau of Energy Efficiency, which already bind connected consumers, and tie Central incentive money to State adoption.
2. Mandates without metering: A renewable sourcing mandate is only as good as the verification behind it, and discoms already miss the obligations they carry. Eg. State distribution companies have missed Renewable Purchase Obligation targets for years, forcing the Ministry of New and Renewable Energy to renotify trajectories.
   The Fix: Require third-party audited reporting of power usage effectiveness and water usage effectiveness (ratios of total facility energy and water to that used by computing equipment) as a condition of every incentive.
3. Cooling technology is a trade-off, not a free fix: Liquid and immersion cooling cut water use but raise capital cost and still dump heat locally. Eg. Evaporative cooling, the cheapest option at 45 degree Celsius, is also the most water intensive.
   The Fix: Set the water standard by climate zone rather than one national number, so a coastal seawater-cooled site and an inland Rajasthan site face different limits.

Conclusion

The data centres will be built, and the only open question is on whose terms. The unresolved tension is between States competing on incentives and a resource base that no State policy has been made to account for. What to watch is whether the Centre converts the four standards, renewable sourcing, water disclosure, grid impact and thermal load, into an enforceable national framework before the projected capacity is locked in. The nearer marker is whether Karnataka’s policy review produces conditions or only more incentives.

Back2Basics

1. Urban heat island: An urban heat island is the difference in temperature between a built-up city and its rural surroundings, caused by concrete, asphalt and roofs absorbing and re-emitting heat that vegetation and soil would have released through evaporation.
2. Intensity: Its intensity is that temperature gap in degrees, so Hyderabad’s 5.74 to 6.82 degree Celsius means the city runs that much hotter than its surroundings at the same hour.
3. Why data centres add to it: Servers convert almost all the electricity they draw into heat, and cooling systems reject that heat into the surrounding air or water, so a large facility acts as a fixed heat source inside the island.

Matching Previous Year Question

“[2026] Which of the following statements with regard to Green Hydrogen is/are correct? 1. It is decarbonized hydrogen obtained from natural gas reforming combined with carbon capture and storage (CCS). 2. It is produced using electrolysis of water with electricity generated by renewable energy. 3. National Green Hydrogen Mission of India aims for abatement of nearly 50 MMT of annual greenhouse gas emissions by 2030. (a) 1 only (b) 2 and 3 only (c) 2 only (d) 1, 2 and 3 Answer: B”