NEW DELHI (AP) — Leaders from governments, technology companies and multilateral agencies converged at India’s first Artificial Intelligence Impact Summit this week to redefine AI deployment for the Global South. The event spotlights India’s chance to lead on AI that drives inclusive growth without ravaging resources, as data centers guzzle electricity rivaling aluminum smelters and 150 billion liters of water in 2024 alone.

Speakers highlighted projections from the Council on Energy, Environment and Water (CEEW) and SYSTEMIQ showing data center water use doubling by 2030. Only five of 15 state-level policies currently mandate sustainability measures, the groups found. Summit chair and columnist outlined risks in a report from the AI and Climate Expert Engagement Group, drawing on CEEW, Dalberg analysis and input from 20 global experts.

“AI ambition and climate responsibility is a false binary,” the report states, detailing applications across models, infrastructure, governance and ecosystems. AI already improves power grids for renewables, delivers hyperlocal flood forecasts, tracks methane leaks and bolsters farm resilience. Yet unchecked expansion threatens social acceptance, especially in water-stressed India with 18% of world population but 4% of its water.

The gathering called for ‘frugal, Gandhian AI’ — efficient, public-purpose systems distinct from resource-heavy models in water-rich nations. India’s data center capacity has tripled since 2020, backed by $95 billion in investments and the IndiaAI Mission for socially responsive tech. Digital public infrastructure proves scalable open systems can deliver public goods.

Experts urged four key shifts. First, enforce environmental disclosures like power usage effectiveness, water-use effectiveness and carbon intensity at data centers, verified independently. An AI Energy and Water Star rating system could guide choices, akin to appliance labels, building on EU AI Act requirements for training data transparency.

Second, revamp incentives in state policies to prioritize efficiency over sheer volume. Rajasthan’s 2025 policy stands out, requiring zero liquid discharge, rainwater harvesting and green standards. Integrated planning must assess grid capacity, water stress and climate risks before site approvals.

Third, designate climate data as digital public infrastructure. India’s Energy Stack and AgriStack show the model: open datasets on soil moisture, groundwater, wind and extreme weather, managed by data trusts for public-interest AI. Satellite launches rose 45% yearly over five years, but converting data to policy lags — AI can close that gap while securing data sovereignty for the Global South.

Fourth, launch massive reskilling for 48 million jobs in 36 green value chains by 2047, generating $1.1 trillion, per CEEW. AI-driven skills matching demands public investment in taxonomies, learning paths and institutional frameworks.

Summit participants drew parallels to India’s successes with the International Solar Alliance, Coalition for Disaster Resilient Infrastructure and Green Development Pact, all born in New Delhi. Organizers said translating summit principles into laws, procurement and enduring systems starts now, rejecting growth-versus-sustainability tradeoffs. With the next billion users in Asia and Africa, India’s blueprint could lock in or liberate AI’s global trajectory.