The Thirsty Machine: India's AI Data Centre Race, the Water Crisis, and the Choice Between Argentina's Libertarian Gamble and China's Governance Blueprint

India is building an artificial intelligence data centre network, which is expected to grow to 15 GW by 2031, almost entirely in severely water-stressed cities. This digital growth is coming at a time when we estimate that by 2025 water use has reached 150 billion litres per year, and by 2030 it will increase to 358 billion litres per year, while experiencing aquifer depletion, water scarcity in cities, and no requirement for environmental disclosure Significantly, under the Environment Impact Assessment Notification 2006, gigawatt-scale computing facilities are classified under building construction (Category B2) and hence do not require a public hearing, disclosure of water source, or mandatory Power Usage Effectiveness (PUE) and Water Usage Effectiveness (WUE) benchmarking.. There is no national data centre policy in force since it has been in draft since 2020. This study employs a most-different systems design model (Przeworski & Teune, 1970) to compare three distinct governance paradigms for AI infrastructure: India's democratic vacuum, Argentina's libertarian deregulation, and China's authoritarian spatial planning. This article examines the libertarian vision of the zero regulation, non-liability, AI corporation, governance-free zones for hyperscale infrastructure espoused by Argentina's Milei government as a model for Third World approaches to international law as neo-colonial resource extraction. China's "East Data, West Computing" initiative offers an alternative state-led roadmap and includes the commercial scale deployment of underwater cooling, the spatial movement of data centers to the interior, the imposition of PUE/WUE standards, and the implementation of water pricing. Based on the insights of recent sustainability science, of hydrology, and of governance scholarship, this article calls for a rejection of the race-to-the-bottom deregulatory impulse and of authoritarian centralisation in India, in favour of a sovereign, democratic path, based on mandatory environmental benchmarks, spatial planning, community consent, and institutional accountability. India's freshwater resources constitute a public commons requiring democratic governance, not an unregulated subsidy for expropriation of the global AI infrastructure expansion.