The Infrastructure Behind Intelligence: Rethinking India’s Data Centre Expansion

India is positioning itself as a major global hub for data centres, driven by the rapid expansion of its digital economy, growing demand for cloud services and the race to build AI infrastructure. At COP31 in Antalya, the Turkish presidency and the International Telecommunication Union launched the voluntary Antalya AI Pledge, calling on governments and companies to power data centres with low-emission electricity and disclose their energy consumption.

The scale of this ambition is reflected in the more than 80 data-centre facilities operating or under development across the country, alongside a pipeline that accounts for nearly a tenth of global construction activity. Installed capacity has increased from approximately 375 MW in 2020 to around 1.8 GW by mid-2026, while committed investments crossed US$126 billion by the end of 2025. Major technology companies, including Google, Microsoft and Amazon, have announced investments of at least US$65 billion in states like Andhra Pradesh, Gujarat and Uttar Pradesh. The Union Budget’s extension of tax incentives for foreign cloud service providers until 2047 depicts the government’s commitment to establishing India as a global digital infrastructure hub. Yet, behind these investment commitments and capacity targets lies a fundamental question of resource sustainability: how will India mobilise the natural resources and ecosystem services required to support this expansion, including reliable energy, freshwater, land and cooling infrastructure, without intensifying pressure on already stressed ecosystems and competing livelihood demands?

AI-generated illustration by the author

The growing electricity demand from data centres will add further pressure on India’s power grid. The consumption is expected to rise from 13 TWh in 2024 to 57 TWh by 2030. Since data centres require uninterrupted electricity, coal is likely to remain an important power asset. This is particularly concerning given India’s plans to add around 97 GW of coal capacity by 2035. At the current grid emission intensity reported by the Central Electricity Authority, meeting a demand of 57 TWh could generate approximately 41 million tonnes of CO₂ annually. Renewable energy procurement is often presented as the industry’s solution, but renewable contracts do not necessarily translate into additional clean electricity, particularly during evening peak hours when solar generation is unavailable.  Nuclear power, including emerging Small Modular Reactors (SMRs), could complement renewables by providing firm, low-carbon electricity to meet the continuous energy demands of data centres, although commercial viability and deployment timelines remain critical constraints. Buying certificates or sourcing power from existing renewable projects does little to address this gap unless it supports additional generation and storage. Andhra Pradesh and Gujarat have introduced a 51% clean-energy requirement for data centres, although no operating facility reportedly meets this threshold until now. Further, exemptions from certain cross-subsidy charges under deemed distribution licences raise concerns about whether the costs of supplying large data-centre premises could ultimately be shifted to other electricity consumers. The central issue is whether India’s data-centre expansion will drive additional clean-energy capacity or deepen coal dependence while placing a greater burden on the existing power system.

Water availability poses the next major challenge. Data centres also depend heavily on cooling systems, and the choice of technology determines whether their operations place greater pressure on electricity or freshwater resources. Evaporative cooling consumes significant quantities of water, while air-cooling systems, favoured by companies such as Amazon and Google, reduce on-site water use but can increase electricity demand. The scale of demand is already significant: Indian data centres are estimated to consume around 150 billion litres of water annually, while nearly 50 facilities are located in areas facing extremely high water stress. Visakhapatnam highlights the challenge of meeting data-centre demand in a water-stressed region. The city receives around 410 million litres of water daily, against an estimated requirement of 480 million litres. To address some of these concerns, Andhra Pradesh has announced a requirement for data centres to source 70% of their energy from renewable sources, alongside measures to address water consumption. Even by the state’s more conservative estimates, data centres would account for 15% of this allocation, bringing into sharper focus the competing demands on an already finite resource. The question, therefore, extends beyond whether the state can accommodate this new industrial demand to how it intends to reconcile it with the water needs of households, agriculture and other essential uses.

Photo by Geoffrey Moffett on Unsplash

The pressures extend beyond natural resources to the land on which this infrastructure is being built. At Tarluvada, more than 50 Dalit families reportedly lost land allocated to them by the state in the 1970s, despite a 1977 state law prohibiting its transfer. Promised alternative plots and jobs have yet to materialise, though the state maintains that no land was taken by force. The economic returns also warrant scrutiny: Google’s proposal projects just 1,725 permanent jobs, against the state’s estimate of 1.88 lakh direct and indirect jobs. Meanwhile, data centres attracted 23.2% of private-equity inflows into Indian real estate in 2025, entirely from foreign investors. The disparity raises a fundamental question: while the costs of land, water and energy remain local, who ultimately captures the economic gains?

Data infrastructures do not require environmental clearance unless they cross building thresholds measured in floor area, rather than their electricity demand or water consumption. Consequently, projects with a combined capacity of 2.51 GW received clearance as building projects in just nine days. The irony is evident: substantial energy and resource footprints can escape closer scrutiny on a technicality. Meanwhile, NITI Aayog’s new framework remains voluntary, and India’s national data centre policy has been in draft since 2020.

These regulatory gaps are mirrored in the way capital is channelled into the sector. Infrastructure status, capital subsidies, land concessions and power-related incentives are helping de-risk private investment, yet these public concessions are not consistently tied to domestic value addition or local economic outcomes. States largely compete through investment incentives, with limited emphasis on measurable performance outcomes. NITI Aayog has called for assessing domestic procurement and linking incentives to employment, local vendor development and capability-building milestones. Equally important is the question of who captures the economic value of these investments. While data centres primarily serve the digital operations of large multinational corporations, their resource and environmental costs are borne locally. What do host communities gain in return for committing their land, water and energy resources? Project financing and regulatory approvals must therefore incorporate enforceable local economic benefit provisions, prioritising domestic procurement, local enterprises, workforce development and greater participation in the data-centre supply chain. Without such commitments, India risks subsidising global digital infrastructure without building corresponding domestic capabilities or ensuring that communities hosting these facilities share in their economic returns. As India races to build the infrastructure of the digital future, its environmental and economic safeguards remain several steps behind.

The answer lies in planning at the scale of the impacts, guided by the precautionary principle. States should assess cumulative water, energy, land and climate risks before designating data-centre zones, with public consultation and environmental clearance triggered by capacity, not merely floor area. Treated wastewater should become the default for cooling, renewable energy claims should reflect genuinely additional supply, and operators should bear the cost of grid upgrades. Transparent resource-use disclosures restored consent for assigned lands and the inclusion of data centres in India’s climate finance taxonomy would further strengthen accountability.

At COP31, the Antalya initiative has brought renewed attention to disclosure and cleaner power, with a first stocktake expected at COP32. India need not wait. Its data-centre ambitions are still measured largely in gigawatts announced and billions invested. But the more consequential question is what these vast structures will leave behind: resilient infrastructure or overstretched grids, depleted water resources and deeper local inequalities? India undoubtedly needs data centres. What it cannot afford is to build the digital foundations of its future while overlooking the environmental and social costs that may outlast the infrastructure itself.

(Views expressed are the author’s own and do not reflect those of ICRIER)

Author

  • Ritika Verma

    I am an environmentalist leveraging technology to better understand and protect our planet. My experience spans diverse fields, including ecology, waste management, and geography, where I have addressed pressing challenges such as climate change monitoring and waste management. My core interests center on climate change mitigation & adaptation, wetland conservation, and the innovative application of GIS and remote sensing to advance environmental research and support informed decision-making.

Published by Ritika Verma

I am an environmentalist leveraging technology to better understand and protect our planet. My experience spans diverse fields, including ecology, waste management, and geography, where I have addressed pressing challenges such as climate change monitoring and waste management. My core interests center on climate change mitigation & adaptation, wetland conservation, and the innovative application of GIS and remote sensing to advance environmental research and support informed decision-making.

Leave a Reply