AI, Data Centres and India’s Emerging Water Footprint

Syllabus: GS3/Science & Technology

Context

• The rapid expansion of AI and data centres has highlighted their hidden water footprint, raising concerns over groundwater depletion, cooling requirements and sustainability in water-stressed regions.

Emerging Water Footprint of AI

• Artificial Intelligence (AI) is often discussed in terms of computing power and electricity consumption, but water is an equally important resource for data-centre operations, particularly for cooling high-density computing systems.
• The University of California has estimated that a typical ChatGPT conversation of around 20–50 exchanges can have a water footprint of up to about 0.5 litre, depending on the location, cooling technology and electricity mix.
  • It illustrates how seemingly intangible digital services have a physical environmental footprint.
• According to a UN assessment, data centres could have a water footprint of around 4.5 trillion litres in 2025, potentially rising to 9.3 trillion litres by 2030.
  • It arises both from direct cooling requirements and indirectly through water used in electricity generation.

India’s Data-Centre Expansion

• India is emerging as a major global data-centre hub owing to rapid digitalisation, cloud computing, AI adoption and increasing data generation. 
• Data-centre capacity reportedly increased from around 375 MW in 2020 to nearly 1,500 MW in 2025, with projections of 13.56 GW by 2031–32.
  • Andhra Pradesh’s Data Centre Policy 4.0, for instance, provides incentives such as GST reimbursement and stamp-duty exemptions.
  • Uttar Pradesh’s 2026 policy targets more than 2 GW of additional capacity.
  • Gujarat’s 2026–29 policy aims at 7.5 GW and investments of around ₹6 lakh crore.

Key Issues & Concerns

• Groundwater stress: The Central Ground Water Board (CGWB)’s 2024 assessment puts India’s stage of groundwater extraction at 60.47%, with 11.1% of assessment units classified as over-exploited.
  • The challenge becomes sharper when data centres are concentrated in already water-stressed urban clusters.
  • For Example:
    • Gautam Buddha Nagar has numerous existing and upcoming data centres, while its groundwater extraction level has been assessed at 104.79%.
    • Hyderabad is also classified as over-exploited. Visakhapatnam, despite Andhra Pradesh’s relatively comfortable state-level groundwater position, faces local resource constraints.
• Lack of transparency: Many data centres do not publicly disclose their monthly or peak water consumption, water sources or proportion of potable, groundwater and reclaimed water used.
  • It makes meaningful assessment of their local environmental impact difficult.
• Water Energy Nexus: Data centres are highly electricity-intensive. Electricity generation itself can have a water footprint, creating an interconnected water-energy challenge.
  • Excessive dependence on water-intensive cooling can also intensify competition between industrial and domestic requirements.
• Coastal risks: Seawater cooling can reduce dependence on freshwater. However, the discharge of concentrated brine can affect marine ecosystems if inadequately managed.

Efforts & Initiatives

• Andhra Pradesh Data Centre Policy 4.0 provides fiscal and energy-related incentives for large data-centre projects.
• UP Data Centre Policy, 2026 seeks to expand capacity by over 2 GW.
• Gujarat Data Centre Policy, 2026–29 targets 7.5 GW of capacity.
• Companies are exploring seawater cooling, as demonstrated by Google’s data centre in Finland.
• Greater adoption of reclaimed wastewater, closed-loop cooling and energy-efficient cooling technologies can reduce freshwater dependence.

Way Forward

• India needs to pursue ‘digital growth with water security’ rather than treating water availability as an externality.
• Mandatory disclosure: Data centres should report monthly and peak water consumption, source-wise use and water intensity per unit of computing.
• Water-stress-based regulation: New projects in over-exploited or stressed areas should face stricter environmental and groundwater permissions.
• Prioritise non-potable water: Mandate greater use of treated wastewater and reclaimed water for cooling wherever technically feasible.
• Promote water-efficient cooling: Encourage closed-loop, liquid, dry and hybrid cooling technologies according to local climatic conditions.
• Integrated planning: Data-centre approvals should consider the combined water-energy-land and ecological carrying capacity of the region.
• Responsible coastal cooling: Seawater systems should incorporate stringent brine-treatment and marine-impact monitoring.

Source: TH