India's race to become a global AI powerhouse is creating an environmental crisis that policymakers are only beginning to address. As the country projects its data center capacity to reach 8 gigawatts (GW) by 2030, water consumption from AI infrastructure could more than double in just five years, jumping from roughly 150 billion liters in 2025 to 358 billion liters by 2030. This explosive growth threatens water security in already drought-stressed cities like Bengaluru, Hyderabad, and Gurugram, even as India positions itself as a leader in "Applied AI" and a driver for the Global South.

The numbers are staggering. India's AI market was valued at $9.51 billion in 2024 and is projected to reach approximately $130 billion by 2032. The Indian government has backed this ambition with over Rs 10,300 crore (roughly $1.25 billion) through the India AI Mission, launched in March 2024. Yet this growth comes with a hidden cost that extends far beyond electricity bills.

Why Is Water Consumption Such a Critical Problem for AI Data Centers?

Data centers require enormous quantities of fresh water, primarily for cooling high-performance processing devices. Large cooling towers evaporate this water, depleting nearby water supplies. In water-stressed regions, this concentration of infrastructure can threaten local water security by taxing municipal supplies that communities depend on for drinking and agriculture. The problem is compounded by the fact that AI's computational demands are growing exponentially, with electricity demand from data centers expected to increase nearly fivefold by 2030.

Beyond water, AI infrastructure creates a cascade of environmental pressures. Data centers are projected to consume nearly 5 percent of India's total electricity by 2030. This surge in power demand often forces reliance on backup diesel generators when local grids cannot keep pace, increasing carbon emissions. The physical expansion of these facilities also fragments forests, with high-voltage transmission lines and road access creating what environmental experts call "linear intrusions" that split large habitats into thousands of small, vulnerable patches.

What Are the Major Environmental Impacts Beyond Water and Energy?

The environmental toll extends to hardware waste and supply chain damage. India is already the world's third-largest producer of electronic waste, and AI hardware accelerates this problem significantly. High-performance graphics processing units (GPUs) typically become obsolete within 2 to 3 years, yet India currently lacks advanced recycling infrastructure to handle the toxic materials and rare minerals found in these specialized chips. When improperly processed, discarded hardware containing lead and mercury can contaminate soil and water. The extraction of raw materials and rare-earth elements required for AI hardware often involves environmentally damaging mining practices.

Additionally, AI applications in agriculture risk unintended environmental consequences. Algorithms optimized for yield increases may inadvertently encourage monoculture farming, reducing biodiversity and soil quality. Increased pesticide and fertilizer use driven by AI optimization can contaminate land and water supplies. More broadly, AI systems can accelerate environmentally harmful activities by optimizing fossil fuel production, promoting overconsumption through supply chain efficiency, or spreading misinformation about climate change that erodes public support for sustainability initiatives.

How Can India Build "Green AI" Infrastructure?

• Renewable Energy Transition: Data centers must shift from coal-heavy grids to renewable energy sources like solar, wind, and hydropower. This requires both policy mandates and infrastructure investment to ensure reliable renewable supply at the scale AI demands.
• Water-Efficient Cooling Systems: Facilities should eliminate cooling towers and transition to zero-water cooling designs or use only non-potable water (treated wastewater) for cooling. Closed-loop liquid cooling systems can dramatically reduce consumption.
• Strategic Location Selection: New data centers should avoid high-temperature and water-deficient regions. Indian coastal cities offer advantages due to lower temperatures and abundant water availability, reducing cooling demands.
• Environmental Impact Assessment Integration: Data centers and AI compute farms should be included in India's Environmental Impact Assessment (EIA) Notification 2006. A proposed "Green-B" category would require environmental clearance for facilities with aggregate IT loads of 5 megawatts or greater, or annual energy consumption of 20 gigawatt-hours or more.
• Operational Transparency Requirements: Facilities should report key performance metrics like Power Usage Effectiveness (PUE) and Water Usage Effectiveness (WUE) to enable accountability and drive continuous improvement.
• Extended Producer Responsibility: AI hardware importers should be required to recycle 70 to 80 percent of specialized chips by 2026, reducing e-waste and recovering valuable materials.

The challenge is urgent. Data center capacity in India is expected to grow by 77 percent to 1.8 gigawatts by 2027, according to global real estate adviser JLL. Without regulatory frameworks in place now, this "explosive growth" will lock in unsustainable practices for years to come.

Cooling technology is also advancing to address these challenges. Johnson Controls, a global leader in thermal management and building systems, released its second AI Factory Reference Design Guide in May 2026, focused on air-cooled chillers for gigawatt-scale data centers. The blueprint demonstrates that high-efficiency air-cooled designs can return up to 50 megawatts of power to the AI facility and improve annual energy consumption by 32 percent while eliminating cooling towers entirely, saving over 12 million gallons of water per day.

"At gigawatt scale, AI factories require a fundamentally different way of thinking about infrastructure," stated Austin Domenici, president of Johnson Controls Global Data Center Solutions. "The future requires designing integrated systems that can scale predictably, perform efficiently and adapt as technology evolves."

Austin Domenici, President, Johnson Controls Global Data Center Solutions

These design innovations show that sustainable AI infrastructure is technically feasible. The question now is whether India's policymakers will mandate their adoption before the window for prevention closes. The stakes are high: India's decarbonization objectives and long-term water security depend on getting this balance right between AI innovation and environmental protection.