The resolution exposes how the world has rushed headlong into an AI-driven future without fully accounting for ecological consequences. Behind every generative AI model lies a hectic physical reality — data centres, specialised chips, water-intensive cooling systems and immense energy demand. What appears digital and weightless is, in fact, deeply material.
What AI Doesn’t Show
India has embraced AI and data-driven development with enthusiasm. By some estimates, nearly 20 per cent of the world’s data originate here, yet accounts for only about 3 per cent of global data centre capacity. This imbalance is now rapidly correcting. Data centres form the physical backbone of AI. They store data, train models and power real-time inference. Currently, India’s installed data centre capacity, estimated at around 950 megawatts, is expected to double by 2026 and reach several gigawatts within the decade. Google has announced a USD 15 billion AI Hub in Visakhapatnam, its largest investment in India, while Amazon Web Services (AWS) is setting up an USD 8.3 billion data centre in Maharashtra.
These data centres are also voracious consumers of energy and water. Globally, the electricity consumption of data centres rose to 460 terawatt-hours in 2022. According to the Organisation for Economic Co-operation and Development, this would have made data centres the world’s eleventh-largest consumer of electricity (between Saudi Arabia (371 terawatt-hours) and France (463 terawatt-hours). Data centres are predicted to consume close to 1050 terawatt-hours of electricity by 2026, moving them up to fifth place on the global list between Russia and Japan.
India is not insulated from these trends. Research suggests that data centres could account for nearly 5 per cent of national electricity consumption by 2030. However, one silver lining remains that data centres indirectly help reduce emissions by lowering the need for paper, travel and physical infrastructure, especially with the rise of remote work.
Full Cost…
Globally, AI-related infrastructure may soon consume six times more water than Denmark, according to one estimate. That is a problem when a quarter of humanity already lacks access to clean water and sanitation. It is particularly concerning given that major data centre hubs such as Mumbai, Bengaluru and Chennai are already classified as water stressed. However, this captures only part of AI’s environmental footprint. Its lifecycle includes the mining of rare earth minerals for chips, energy-intensive manufacturing processes and the growing burden of electronic waste as it becomes obsolete within years rather than decades. These upstream and downstream impacts are routinely excluded from sustainability discussions. Recent analyses show that companies deploying AI systems do not measure their environmental impact in any systematic way.
Not reliant on a single resource
There is, however, room for cautious optimism. Data centres powered by solar, wind or hybrid energy systems can significantly reduce emissions. Advances in cooling technologies, closed-loop water systems and AI-driven energy optimisation can lower both energy and water intensity, turning sustainability into a competitive advantage rather than a constraint. Green financing mechanisms can further ensure that capital flows towards infrastructure that meets measurable environmental performance.
