India is rapidly emerging as a global data centre hub. By the end of 2025, billions of dollars had been committed to AI and data infrastructure, with major technology firms announcing large-scale investments. At the same time, India generates nearly 20 percent of the world’s data but hosts only around 3 percent of global data centre capacity. Expanding domestic infrastructure is therefore seen as both strategic and inevitable. But beneath the digital narrative lies a more fundamental reality: AI infrastructure is water infrastructure in disguise.
A small 1 MW data centre using traditional cooling systems can consume an estimated 26 million litres of water annually – roughly equivalent to the yearly domestic needs of over five hundred people. With over 1.4 GW operational, another 1.4 GW under construction, and 5 GW in the pipeline, India’s AI expansion is not merely an electricity challenge but a hydrological one.
The central question is not whether India should build AI capacity. It is whether the country can do so without treating water as an unconstrained input into digital growth.
A Strategy Racing Ahead of Governance
India does not yet have a unified national AI policy. Regulation remains fragmented across sectors, and data centres are typically treated as industrial or IT infrastructure rather than as resource-intensive utilities. Environmental impact assessments are not systematically mandated on the basis of water intensity, nor is there a binding national framework for disclosure of water and energy footprints.
This institutional gap matters.
Water governance in India is divided between central environmental oversight and state-level groundwater regulation, while digital policy is driven primarily by the Ministry of Electronics and Information Technology. The result is a structural disconnect: the ministry promoting AI capacity expansion is not accountable for groundwater stress or environmental sustainability.
Such fragmentation creates regulatory blind spots. Infrastructure decisions made today will lock in cooling technologies, sourcing models, and site locations for decades. Retrofitting water-efficient systems later will be far more expensive than embedding sustainability standards at the outset.
The Economics of Mispriced Water
The deeper risk is not only environmental – it is economic.
India has 18 percent of the world’s population but only about 4 percent of its freshwater resources. Urban water supply norms already strain infrastructure in many cities. Yet industrial water abstraction often remains underpriced or weakly monitored, especially in groundwater-stressed regions.
When water is mispriced or weakly regulated, data centre expansion effectively externalises its true costs. What appears as rapid digital growth may in fact be the silent depletion of aquifers. In extreme summer conditions, rising temperatures and water scarcity could force facilities to scale down operations, directly undermining the AI ambitions they were built to support.
Infrastructure that cannot reliably operate at full capacity due to water stress risks becoming stranded capital. For a country positioning itself as a global AI destination, that is not just an environmental liability but a macroeconomic one.
From Voluntary Sustainability to Binding Standards
Policy solutions are not conceptually complex. What is missing is political prioritisation.
First, environmental impact assessments should be mandatory for large data centres above a defined capacity threshold, with explicit accounting of water intensity and cooling methods.
Second, projects located in over-exploited groundwater blocks should be required to meet a minimum percentage of cooling demand through treated wastewater rather than freshwater abstraction.
Third, differential water tariffs should penalise unsustainable freshwater extraction while incentivising recycled water use.
Finally, mandatory public disclosure of water intensity metrics – subject to third-party audits – would align data centre growth with transparent ESG standards and prevent greenwashing.
These measures would not slow innovation; they would make it durable.
Governing AI as an Ecosystem
India’s AI moment is real. But digital infrastructure cannot be governed as a series of isolated industrial projects. It must be treated as part of a broader ecological and economic ecosystem.
The choice is not between growth and sustainability. It is between building AI capacity on transparent, priced, and regulated resource foundations – or allowing a fragmented governance model to lock in unsustainable patterns that future governments will struggle to correct.
India still has a window. Much of its projected data centre capacity remains under construction or in the pipeline. Regulatory architecture designed now will be cheaper, fairer, and more effective than crisis management later.
If AI is to power India’s future, water governance will determine whether that future is stable or fragile.
