DCEI: A Data Center Environmental Impact Index for India — Quantifying Water, Carbon, Community, and Thermal Risk for AI Infrastructure Siting Across 50 Indian Cities

 India is building data centers at breakneck speed — $32 billion committed since 2023. But nobody is asking a basic question: should we be building them WHERE we're building them?

Mumbai has 70+ data centers and an 80 million litre/day water deficit. Bengaluru's groundwater is officially "Over-Exploited" and the city ran dry in 2024. Chennai's reservoirs collapse almost every summer. Delhi's water table drops 1-2 metres every year. And we're putting the most water-hungry industrial facilities on the planet right on top of these cities.

This paper introduces DCEI — a scoring framework that evaluates 50 Indian cities across four dimensions: water stress (using CGWB groundwater data), carbon intensity (CEA grid emission factors), community impact (population density, existing DC concentration), and thermal cooling load (IMD climate data, wet-bulb temperatures). Every data point comes from public Indian government sources.

The results are uncomfortable but not surprising: India's most popular data center locations are also its most environmentally risky. Delhi scores 66/100. Chennai 64.8. Mumbai 59.2. Meanwhile, cities like Kochi (27.6), Guwahati (27.9), and Dehradun (30.1) offer dramatically lower risk — but host almost no data center infrastructure.

An open-source interactive dashboard accompanies this paper at dcei.vercel.app, letting anyone explore scores, compare cities, adjust weights, and find lower-impact alternatives.

This is v0.1 — a working paper, not a final word. The framework has limitations (documented honestly in the paper) and I actively welcome feedback from domain experts in hydrology, energy systems, and urban planning.

Code and data: github.com/rudra6155/DCEI