India’s data centre industry is entering a period of rapid expansion, driven by cloud adoption, 5G, data localisation and the accelerating demand for artificial intelligence. Operational capacity has increased from approximately 375 MW in 2020 to more than 1,575 MW in 2025, with estimates suggesting that capacity could reach 4.5–9 GW by 2030. The sector has attracted an estimated US$95 billion in cumulative investment commitments between 2019 and 2025.
A new discussion paper by the Policy Consensus Centre (PCC), titled Living Next to the Cloud: An Evidence-Based Approach to Data Centres in Residential India, examines how this expansion can be managed alongside communities located near data centre facilities. The paper argues that the question is increasingly one of how data centres and residential communities can coexist, rather than whether digital infrastructure can simply be excluded from urban areas.
Following the paper, PCC has also reviewed environmental clearance filings for data centre projects in Maharashtra and Tamil Nadu, publicly available on the government’s PARIVESH portal. The filings show how operators are designing facilities to coexist with their neighbours.
Globally, data centres have tended to cluster around cities because of access to fibre networks, electricity infrastructure, skilled workers and, importantly, end users requiring low-latency services. India exhibits a similar pattern, with Mumbai, Chennai, Bengaluru, Hyderabad and Delhi NCR accounting for the overwhelming majority of reported operational capacity.
India and international approaches
The regulatory approaches adopted internationally generally focus on measurable impacts rather than an outright prohibition on data centres near residential areas. The paper finds that several international jurisdictions use minimum setback requirements, facility-specific noise limits, electricity and water disclosure requirements, renewable-energy obligations and efficiency standards. For example, Germany requires data centres to progressively increase renewable electricity sourcing, while Ireland has introduced requirements around renewable generation and dispatchable capacity for new data centre connections. The European Union has also moved towards sustainability reporting for larger data centres.
India currently regulates the same issues through a combination of central laws, state-level zoning policies and general environmental and building regulations. Residential areas are subject to ambient noise limits of 55 dB(A) during the day and 45 dB(A) at night.
Electricity consumption is managed through sanctioned-load approvals rather than an absolute consumption cap, while water use is governed principally through wastewater-reuse requirements and groundwater extraction rules. One significant difference is location policy.
Three concerns: what does the evidence show?
1. Noise: Data centre cooling systems, transformers and ventilation typically generate 45–65 dB at the property boundary, while India’s residential limits are 55 dB during the day and 45 dB at night. The paper highlights the need for consistent monitoring and enforcement, noting that a study across 70 monitoring stations in seven major cities found applicable noise standards were exceeded at monitored residential, commercial and silence-zone locations.
Backup diesel generators remain an additional source of noise and are subject to acoustic enclosure and noise-control requirements.
2. Electricity: India’s data centre electricity demand could rise from approximately 1.4 GW in 2024 to 9 GW by 2030, making grid planning, transmission capacity, renewable-energy additions and local infrastructure increasingly important.
In India, clearance filings show operators designing facilities to use less energy and to draw it more responsibly.
3. Water: Water impacts are highly location-specific. A typical 100 MW hyperscale facility using water-based cooling can consume 800,000–2 million litres per day, while India’s data centres are estimated to have consumed around 150 billion litres in 2024–25, potentially rising to 358 billion litres by 2030. The paper emphasises that local water availability, competing demand and cooling technology determine actual impact. Dry and hybrid cooling can reduce water use, although dry cooling may increase electricity demand. Clearance filings show the industry moving away from water-intensive cooling.
Recommendations
The filings reviewed by PCC underline the need for these measures. Clearance conditions for data centres follow a general building template, with no provisions on generator running hours, energy efficiency or cooling water. Six of the seven clearances reviewed fall in a category that requires neither a full environmental impact assessment nor a public hearing.
Compliance reporting is largely self-declared, and only six of 23 compliance reports record a site inspection. The seven projects with equipment details together provide for more than 600 large diesel generator sets, with backup capacity exceeding their combined grid demand.
The Policy Consensus Centre recommends a framework focused on measurable impacts and greater transparency, including district-level mapping of land, power, water and infrastructure capacity; standardised reporting of facility-level electricity, water and noise; differentiated location norms for hyper scale and latency-sensitive facilities; minimum setbacks and clear norms for facilities near homes; a dedicated data centre chapter in the National Building Code; cleaner backup power and greater use of renewable energy; prioritising treated, non-potable water for cooling; mandatory BIS efficiency metrics with facility-specific targets; greater research into low-water cooling technologies suited to India’s climate; and national model guidelines for states covering reporting, planning, power infrastructure and research while retaining state and municipal authority over approvals.
The paper’s central finding is that India’s data centre expansion does not have to be framed as a binary choice between digital infrastructure and residential communities. The more practical policy question is whether location, technology, resource availability and operational performance can be measured upfront and regulated transparently.
