Where Should India Build Its AI Data Centres? A Question of Resources
Artificial intelligence (AI) relies on computing data-centre, storage and networking, while data-centres, in turn, depend on critical resources such as electricity, cooling water, land and high-capacity communication networks. As India’s data-centre capacity grows from about 375 MW in 2020 to around 1.5–1.6 GW in 2025–26, the siting of large AI data-centres has become both an engineering and socio-economic question. This paper examines the siting problem as a multi-resource optimisation involving water, energy, connectivity and climate resilience. Using Water Usage Effectiveness (WUE), an illustrative calculation for Chennai shows that a 300 MW data-centre cluster with a WUE of 1 L/kWh would need about 5.76 million litres of water per day, a small fraction of the recycled water available from the city’s utility. The average water availability is less important than drought-year reliability and allocation priority. Further, the electricity, especially the firm power, may be a more binding constraint than water. Treating disaster risks as core design criteria, the submarine (seabed) cable landing station can be connected to an inland siting data-centre through diverse and resilient terrestrial fibre routes. India needs a transparent, multi-criteria siting framework rather than a search for a single ideal location.
Authors
- Neelakantan TR (ORCID: https://orcid.org/0000-0001-5721-3398)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-19
- DOI
- https://doi.org/10.5281/zenodo.24365842
- Primary Topic
- Green IT and Sustainability
- Type
- article
- Field-Weighted Citation Impact
- 0.00