India’s AI Data Centre Boom Is Draining Water: Weak Regulation Threatens Long-Term Water Sec…

This is a white paper issued by the Office of the Principal Scientific Adviser to the Government of India, dated December 29, 2025, that examines the policy priorities for democratizing access to AI infrastructure (compute, datasets, and models) in India. It matters to citizens as it outlines the government's strategic approach to making AI resources widely available and affordable, treating them as shared public utilities rather than proprietary assets.

The document concerns the Office of the Principal Scientific Adviser to the Government of India, the broader AI ecosystem stakeholders, innovators, and the general public who are intended to benefit from equitable access to AI infrastructure, datasets, and compute resources.

• The document defines 'democratising access to AI infrastructure' as making compute, datasets, and model ecosystems available and affordable for a wide set of users.
• It identifies three key enablers for this goal: expanding access to high-quality datasets, providing affordable and reliable access to computing resources, and integrating AI with Digital Public Infrastructure (DPI).
• The paper proposes treating the building blocks of AI systems (data, compute, models) as Digital Public Goods (DPGs) or shared public utilities, such as open data repositories and subsidised compute clouds.
• It notes that India hosts nearly 20% of the world's data but only 3% of global data centre capacity, with installed capacity expected to grow from 960 MW to 9.2 GW by 2030.
• The document highlights existing government initiatives including the IndiaAI Mission (which includes a 3,000 GPU cluster and the IndiaAIKosh repository), the National Supercomputing Mission (NSM), and the India Semiconductor Mission (ISM).
• As of December 2025, the IndiaAIKosh platform has onboarded 5,722 datasets and 251 AI models from 54 entities across 20 sectors.

Written from the document by AI, and checked against it. The original below is authoritative.

The original document