Science & Technology19 November 20266 min read
Petitions challenge proposed USD 15 billion Adani–Google data-centre campus
Petitions before the National Green Tribunal (NGT) challenge environmental clearances and land allocation for a proposed USD 15 billion Adani Enterprises–Google AI data-centre campus in Visakhapatnam, Andhra Pradesh. Opposition to unchecked expansion also featured in the 2026 US midterm elections, while local resistance has surfaced in Finland and Kenya.
The dispute centres on the resources AI facilities require as India seeks to expand its digital and computing capacity.
How AI data centres differ from conventional facilities
A data centre houses servers, storage drives, networking equipment and cooling systems to store, process and manage digital data. AI data centres use dense clusters of Graphics Processing Units (GPUs) to train Large Language Models (LLMs) and perform real-time inference, or generate responses to user prompts.
These facilities require more computing power, and in turn more land, electricity and water. They are also often located near user hubs to reduce latency, the delay experienced by digital signals over long distances or congested networks.
Why India is expanding domestic data-centre capacity
India’s data-centre capacity rose from 520 MW in 2020 to 1.5–1.57 GW in 2026. The Economic Survey 2025–26 noted that India generates nearly 20% of the world’s data but hosts about 3% of its data centres. The main demand and policy drivers include:
- Data sovereignty and security: Domestic facilities can strengthen control over sensitive data and support compliance with the Digital Personal Data Protection Act, 2023.
- Digital public services: Data centres support real-time banking, UPI transactions, e-governance and cloud-based services such as MeghRaj. The digital economy is projected to contribute 20% of national income by 2030.
- AI and semiconductor ambitions: Local computing capacity supports the IndiaAI Mission and Semicon India Programme 2.0.
- Investment incentives: The Union Budget 2022–23 gave data centres infrastructure status by including them in the Harmonized List of Infrastructure. The Union Budget 2026–27 introduced a tax holiday until 2047 for eligible foreign cloud providers using India-based facilities.
- Private investment: Google, Amazon and Microsoft have committed more than USD 67.5 billion to AI and cloud infrastructure in India, including subsea cables, clean energy and power-grid resilience.
What environmental pressures do data centres create?
The Ministry of Power has projected that AI and data centres could add 26.3 GW of electricity demand to India’s grid by 2031–32. The International Energy Agency (IEA) has compared a typical AI data centre’s electricity use with that of 100,000 households. The largest upcoming facilities could use 20 times as much.
The pressures include demands on power and water, heat and noise, and the risk of higher emissions if renewable supply does not keep pace.
| Pressure | Reported impacts and figures |
|---|---|
| Electricity demand and grid strain | The projected additional demand is 26.3 GW by 2031–32. A typical AI data centre uses as much electricity as 100,000 households; the largest upcoming facilities could use 20 times as much. Heavy, round-the-clock loads could also lead to higher tariffs for households. |
| Heat generation and cooling limits | Almost all electricity used by IT equipment ultimately becomes heat. A 1-GW facility produces nearly 1 GW of thermal load. AI racks operating at 120–150 kW can exceed the limits of air cooling, known as the “Thermal Wall”, requiring specialised liquid cooling. Cooling equipment also consumes electricity. |
| Water use and competition | Evaporative cooling draws large volumes of freshwater, while direct-to-chip liquid cooling reduces but does not eliminate water use. Estimates cited by the Council on Energy, Environment and Water (CEEW) put data-centre water use at about 150 billion litres in 2025, projected to reach 358 billion litres by 2030. This could intensify competition with households and agriculture, particularly in water-stressed regions. |
| Noise | Mechanical equipment used to air-cool a 1-GW facility can generate up to 100 decibels of noise, which can travel several kilometres and disrupt nearby communities. |
| Urban heat | More than half of India’s data-centre capacity is concentrated in Mumbai and Navi Mumbai, with other major hubs in Chennai, Hyderabad, Bengaluru, Delhi-NCR and Jamnagar. Waste heat from concentrated facilities can intensify the urban heat-island effect and increase air-conditioning demand. |
| Emissions | If renewable energy does not scale with demand, greater reliance on coal-based power could raise emissions and complicate India’s Net Zero 2070 commitment. |
Why are communities and regulators challenging projects?
A frontier-scale 150 MW facility can employ thousands of people temporarily, but automated operations typically retain few permanent jobs. Users far from a facility may receive the benefits of its digital services, while host communities bear pressures on land and water, as well as heat and noise. This distribution has contributed to NIMBY, or “Not In My Backyard”, opposition.
The absence of a binding national data-centre policy has left states competing for investment through fast-track clearances and subsidies, sometimes at the expense of environmental and community safeguards. Under the Environmental Impact Assessment Notification, 2006, AI data centres do not have a separate project classification. This could allow some projects to avoid comprehensive environmental assessment and public consultation.
In Visakhapatnam, fast-tracked clearances have raised concerns about scrutiny of water and ecological impacts near the Mudasarlova reservoir catchment and Kambalakonda Wildlife Sanctuary.
What measures could make expansion more sustainable?
Proposals to reduce the environmental and social costs of expansion include:
- Establish a binding national framework, drawing on NITI Aayog’s Data Centre Evaluation Framework (2026), that links approvals and incentives to grid readiness, water availability, environmental safeguards and site selection.
- Prioritise smaller, task-specific AI models and decentralised compute networks to reduce dependence on resource-intensive hyperscale facilities.
- Promote additional renewable energy, closed-loop direct-to-chip liquid cooling and water recycling. The proposed approach calls for 100% additional renewable energy.
- Provide local benefits through grid upgrades, water-recharging infrastructure, skills development and fair land compensation.
- Strengthen e-waste and hazardous-waste management, including the safe disposal of cooling fluids and batteries, and monitor diesel-generator emissions.
- Require public disclosure of Power Usage Effectiveness, Water Usage Effectiveness, renewable-energy use and waste generation, backed by independent environmental audits.
The stated aim is to expand data-hosting capacity while protecting ecological safeguards, providing community benefits and keeping growth aligned with Net Zero 2070.