Key Takeaways:
- Principal Scientific Advisor calls for open access to foundational compute, datasets and tools to broaden AI development.
- IndiaAI Mission already provides GPU pools to researchers and startups to reduce concentration of power.
- Report urges integration of digital public infrastructure such as Aadhaar and UPI to support inclusive AI adoption.
- Emphasis on sustainability as data centre demand could rise to nearly 3% of electricity use by 2030.
India urged to open AI infrastructure to wider users
The office of the Principal Scientific Advisor (PSA) to the Indian government has set out a case for widening access to core artificial intelligence resources so that a larger set of organisations can develop and deploy models responsibly. The working paper, released on Monday 30 December 2025, recommends making compute capacity, high-quality datasets and enabling tools available beyond a small group of large firms and major urban centres.
AI infrastructure in India
The paper argues that open access to foundational resources will allow universities, regional startups and public-interest groups to build and test AI systems without being excluded by cost or geography. It notes that much of the world’s cutting-edge AI infrastructure has been developed by Western technology companies, a concentration that has prompted concern among policymakers about market power and equitable outcomes.
As India prepares to host the AI Impact Summit in February 2026, democratising access is a central plank of the government’s approach. Under the IndiaAI Mission, officials have already made a pool of thousands of graphics processing units available to researchers and local startups. The PSA paper recommends extending such measures and integrating India’s digital public infrastructure into AI development to widen participation.
Projects such as Aadhaar and the Unified Payments Interface (UPI) are cited as examples of digital public infrastructure that could be linked to AI systems to support a broader set of use cases. The report suggests that doing so would not only expand the ecosystem domestically but could serve as a model for other developing economies seeking digitised identity, payment and governance tools.
The working paper also addresses the environmental footprint of scaling AI. It warns that meeting future compute demand will require significant additional data centre space — an estimated 45 to 50 million square feet by 2030 — and that data centres could rise from around 0.5% of India’s electricity consumption today to nearly 3% by 2030 unless energy efficiency is prioritised. The PSA calls for planning that couples compute expansion with sustainability measures.
Sectoral adoption is another focus. While pharmaceuticals and healthcare have been relatively quick to trial AI solutions, the paper highlights that agriculture and education have lagged. It recommends ecosystem-wide initiatives to expand access to data and computing for these sectors, enabling locally relevant models and services that address on-the-ground needs.
Officials familiar with the paper said the emphasis is on creating pathways for smaller players and regional centres to contribute to innovation, rather than centralising capabilities in a few large firms. The approach would include shared compute facilities, public datasets, tools for responsible deployment and capacity-building programmes for researchers across India.
Analysts say the move aligns with broader strategies to reduce dependence on a handful of global providers and to foster domestic talent and infrastructure. If implemented, the measures could help India export models of inclusive digital governance and collaboration to partner nations in the BRICS and beyond.
The PSA’s working paper stops short of prescribing a single model, instead urging a mix of public and private initiatives to ensure access, affordability and sustainability. With the AI Impact Summit on the horizon, policymakers will face choices about how quickly to scale compute capacity and how to balance innovation with energy and governance considerations.
