The world is riding an artificial intelligence (AI) wave. Governments are racing to expand their AI capabilities, technology companies are investing in its development, and experts are warning about the risks of increasingly powerful systems. India, too, wants a bigger role in this race. But behind the technology lies a physical infrastructure that needs electricity, water and land: data centres.
India’s installed data-centre capacity has grown from about 375 megawatts (MW) in 2020 to approximately 1.5 gigawatts (GW) in 2026, according to the Union government. On October 14, NITI Aayog’s Frontier Tech Hub released a Data Centre Evaluation Framework to Help States Plan for Sustainable Growth, urging states to assess infrastructure and resource availability before committing support to projects. The framework comes as demand for AI, cloud computing and other digital services drives the need for greater computing capacity.
Behind every cloud service and AI tool is a data centre, a facility housing servers, storage systems and networking equipment that process, store and manage vast amounts of digital information.
AI leans on these machines heavily. Models need enormous computing power to be trained, and even more to answer each user’s request. As adoption grows, the facilities need more advanced hardware and a steady, dependable power supply. All that equipment runs hot, so cooling is a necessity, and some cooling methods can use a lot of water. The International Energy Agency (IEA) estimates that data centres used about 415 terawatt-hours (TWh) of electricity globally in 2024, roughly 1.5 per cent of the world’s total. Their electricity use has grown by about 12 per cent annually since 2017. These are global figures, not estimates of India’s consumption, but they illustrate the growing energy demands of digital infrastructure.
As data centres expand, states must consider whether existing electricity grids can support their requirements. New projects may require grid upgrades and additional power infrastructure, while facilities also need reliable supplies to avoid disruptions.
NITI Aayog’s framework recommends assessing grid readiness, estimating how much electricity a project will draw from the grid and considering renewable energy and storage-backed power. It also calls for allocating shared infrastructure costs according to who benefits from it, the contracted load and actual utilisation. Assets built exclusively for a project may be funded by its developer.
Water is another consideration. Data centres use different cooling technologies, each with its own resource requirements. A location with limited water availability may face competing demands from households, agriculture and industry.
The framework recommends assessing water availability at the basin level, accounting for seasonal changes and the impact on existing users. Rainwater harvesting, on-site water treatment and storage are among the measures it proposes. The aim is to identify constraints before a project is approved rather than after a facility begins operating.
For states, the draw is clear. Data centres can build up India’s digital backbone, bring in investment and give a base to AI and cloud services. The Union government has already sweetened the offer, granting data centres infrastructure status and announcing a tax holiday running to 2047 for eligible foreign cloud providers that use facilities in India, subject to the applicable conditions.
However, investment alone does not establish a project’s wider economic value. States must also consider the infrastructure required, employment opportunities and benefits for domestic businesses.
NITI Aayog recommends assessing the economic and strategic value of projects and linking state incentives to measurable outcomes. It also calls for suitable site selection, time-bound approvals and monitoring of developers’ commitments.
The framework is intended to guide state-level decisions, not introduce an additional layer of regulation. Its effectiveness will depend on how states use its recommendations when evaluating projects and allocating public resources.
As India expands its AI capabilities, the challenge is not simply to build more data centres. It is to ensure that the infrastructure supporting this growth is planned responsibly, delivers lasting economic benefits and does not place avoidable pressure on electricity, water and land. Who ultimately pays for that growth will depend on the decisions being made now.