Plan Water First Before Expanding Artificial Intelligence Infrastructure

To
Prime Minister Narendra Modi
Global Leaders and Policy Makers

The Issue

India is rapidly expanding its Artificial Intelligence (AI) infrastructure. Massive data centres are being planned and built to power AI systems, cloud computing and digital services.

But AI does not operate in the cloud alone.

It runs inside high-density facilities that generate intense heat 24 hours a day. To prevent overheating, many data centres use evaporative cooling systems that consume significant quantities of water — often permanently lost through evaporation.

Peer-reviewed research (npj Clean Water, Nature portfolio) confirms that data centres can consume substantial volumes of freshwater, including potable water, for cooling. As AI workloads scale up, cooling demand increases proportionately.

The International Energy Agency (IEA) has warned that electricity demand from AI and data centres will grow sharply this decade. More computing means more heat. More heat often means more water.

Yet at the recent Global AI Summit, the environmental cost — especially water — received little attention.

Water cannot remain invisible in the AI revolution.

 
Facts Speak for Themselves

• Industry projections estimate India’s data centre water consumption could rise from roughly 150 billion litres annually to around 358 billion litres by 2030.

• A 1-megawatt (MW) data centre can consume approximately 20–30 million litres of water per year (55,000–80,000 litres per day).

• A 100-MW AI data centre campus — increasingly common for hyperscale operations — could require 2–3 billion litres annually.

• India is the world’s largest extractor of groundwater (FAO, World Bank).

• Approximately 70–80% of India’s freshwater withdrawals support agriculture (FAO AQUASTAT).

• Over 60% of irrigation in many states depends on groundwater.

These are institutional and scientific assessments — not conjecture.

 
Water vs Wheat: The Real Trade-Off

India is a farming nation. Water sustains crops, rural livelihoods and food security.

Typical irrigation needs:

• Wheat: ~4–6 million litres per hectare per season
• Paddy: ~10–15 million litres per hectare per season

Now please compare:

AI Infrastructure Water Use
Agricultural Equivalent
1 MW Data Centre
Irrigates ~4–7 hectares of wheat
100 MW Campus
Irrigates ~300–700 hectares of wheat
Projected 2030 DC Water Use
Tens of thousands of hectares
Evaporative cooling represents consumptive water loss. It does not recharge aquifers.

In water-stressed basins, this becomes a zero-sum equation.

If billions of litres are allocated to cooling servers, they are not available for irrigation, drinking water or ecosystem needs.

This is hydrological arithmetic.

 
The Impending Crisis

India holds 18% of the world’s population but only about 4% of global freshwater resources.

Nearly 600 million citizens face high to extreme water stress.

States such as Punjab, Haryana, Rajasthan, Karnataka, Tamil Nadu and Maharashtra report falling groundwater levels.

Climate change is intensifying heatwaves and rainfall variability — increasing irrigation demand and increasing AI cooling demand simultaneously.

Unchecked AI infrastructure expansion could accelerate:

• Aquifer depletion
• Rising pumping energy costs
• Rural-urban water conflict
• Agricultural instability

In a country where farmer distress is often linked directly to water availability, this trade-off demands serious national debate.

Sustainable development cannot mean digital expansion at the cost of food and water security.

India’s Opportunity: Lead in Sustainable AI Infrastructure

This is not at all a call to halt AI progress. It is a call to innovate responsibly.

India should not simply replicate water-intensive global data centre models. We should lead the world in low-water AI infrastructure design.

Research and pilot projects globally show that:

• Direct-to-chip liquid cooling
• Single-phase and two-phase immersion cooling
• Closed-loop cooling systems
• AI workload optimisation
• Waste heat recovery systems

can reduce water consumption by 30–50% compared to traditional evaporative cooling systems.

India has the scientific capacity to pioneer water-responsible AI infrastructure suited to hot climates.

 
The Ask: A National “Plan Water First” Framework

We respectfully urge the Government of India and global leaders to adopt the following measures:

1. Mandatory Water Disclosure

Require public reporting of annual water consumption (WUE), source of water, and basin-level impact for all data centres.

2. Aggressive Rainwater Harvesting

Mandate large-scale rainwater harvesting and groundwater recharge infrastructure in every AI campus.

3. Mandatory Use of Treated Wastewater

Cooling systems must prioritise treated municipal wastewater in water-stressed regions.

4. Invest in Low-Water Cooling R&D

Launch a National Mission on Sustainable Data Centres under MeitY, NITI Aayog or Jal Shakti.

Provide grants to IITs, IISc and engineering institutions to develop:

• Water-free cooling systems
• Climate-adapted cooling technologies
• Net-positive water data centre models
• Heat reuse systems for agriculture or district needs

5. Incentivise Water-Neutral Infrastructure

Offer fiscal benefits to facilities achieving near-zero potable water use and demonstrable aquifer recharge.

6. Comprehensive Basin-Level Impact Assessments

No hyperscale AI campus should be approved without cumulative hydrological impact studies.

7. Desalination as a Last Resort

In coastal regions, desalination may be considered only after conservation and recycling measures are exhausted, and with strict environmental safeguards.

 
A Responsible National Question

India has led the world in digital innovation, renewable energy expansion and sustainable development advocacy.

If irrigation wells run dry while data centres operate uninterrupted, what model of development are we choosing?

This is not AI versus agriculture.

It is about ensuring that technological progress does not quietly undermine the foundation of rural livelihoods and food security.

India is a farming civilisation, not merely a digital marketplace.

Water sustains life.
Water sustains crops.
Water sustains social stability.

Let India lead not only in AI capability — but in AI responsibility.

We urge immediate action before infrastructure decisions lock the nation into irreversible water stress.

Yours sincerely,

NatConnect Foundation
and other environmentally concerned citizens groups