A water-positive framework for data centers in India
Watershed accounting. Liquid cooling with dry heat rejection. Why PUE alone is not enough.
A conventional gigawatt-class AI campus, cooled by evaporating water in cooling towers, moves through four to fifteen billion litres of water a year. The number depends on the cooling architecture and on how honest the accounting is. PUE has been the industry default metric for twenty years. It says nothing about that water. It also stays silent on the water embedded in the electricity the facility buys. This paper proposes a more honest accounting. It separates three layers of water responsibility: on-site use, water embedded in electricity, and watershed-scale impact in the geography of operation. It also introduces the Nagmati Programme, which is the commitment HyperNext made on World Environment Day 2025 to restore more watershed capacity than the company consumes. Under this accounting HyperNext's own on-site cooling water is effectively zero, because every campus uses a sealed glycol cooling loop with dry heat rejection rather than evaporative cooling, and recycles building-services water on site. The framework is non-proprietary. We hope other Indian operators pick it up.
The complete paper, including all figures, tables, references, and citations, is available as PDF. Enter your details to receive it.
Request paper · HN-RP-001.pdfWatershed restoration is measured on the ground, not on a slide. The images below document the first phase of the Nagmati Programme in Kutch, Gujarat: from the ₹5 crore CSR commitment that funds it, to the recharge structures built across the seasonal channel and the site reviews carried out with the local community ahead of the monsoon.
HyperNext Research. (05 June 2025). The Nagmati Programme: A water-positive framework for data centers in India. HyperNext Data Center Limited. HN-RP-001. Retrieved from https://www.hypernxt.com/research/hn-rp-001
@techreport{hypernext_hn_rp_001,
title = {The Nagmati Programme: A water-positive framework for data centers in India},
author = {HyperNext Research},
institution = {HyperNext Data Center Limited},
number = {HN-RP-001},
year = {2025},
url = {https://www.hypernxt.com/research/hn-rp-001}
}