Principal Scientific Advisor’s white paper says open access to AI infra key in India

The office of the Principal Scientific Advisor to the government said in a working paper that making Artificial Intelligence infrastructure available to all players would be essential. In the paper, released on Monday (December 30, 2025), the PSA said that it would be important to make “foundational AI resources, such as compute capacity, high-quality datasets, and enabling tools, available beyond a limited set of large firms and major urban hubs, so that a wider range of actors can build, test, and deploy AI responsibly.”

As India gears up to host the AI Impact Summit in February 2026, the government has set democratising access as a key plank, as much of the world’s AI infrastructure is being built by Western tech giants. Officials have reacted with concern around the concentration of power that this situation could lead to. Under the IndiaAI Mission, for instance, the government has provided a pool of thousands of graphics processing units (GPUs), which are essential components of AI infrastructure, for researchers and local startups to access. 

The working paper echoes those concerns and suggests that India should work to integrate its own digital public infrastructure (DPI) into AI systems and open the gates for small players to participate. DPIs are projects like Aadhaar and UPI, which have been referred to under this umbrella term as a way to push other developing economies to come up with digitised identity, payments and governance systems. 

“A significant challenge arising from AI development is resource efficiency,” the paper says, pushing for India to contribute to a less energy-intensive future for AI infrastructure. “Scaling AI data centres will require an additional 45–50 million square feet of real estate by 2030, underscoring the need to integrate sustainability planning with compute expansion. Data centres currently account for roughly 0.5% of India’s total electricity consumption, a share that could rise to nearly 3% by 2030 as capacity and workloads expand.”

The report further said that “ecosystem-wide efforts are needed for expanding access to data and computing for scaling AI development and adoption” in sectors such as agriculture and education, which have been relatively slower than sectors like pharma in taking up AI, the report said.