India should treat artificial intelligence as digital public infra, not proprietary assets: Govt white paper

The paper was released on December 29

India's approach to to artificial intelligence should be hinged on owning less large amounts of hardware and more on treating core AI resources as shared public infrastructure, according to a white paper released by the Office of the Principal Scientific Adviser to the Government of India on December 29.

The paper argues that access to compute, datasets and AI models has become a "critical determinant of innovation, competitiveness, and governance in the digital economy". However, it notes that these remained concentrated in a handful of global firms and major urban hubs.

For India, it says, the policy challenge is to ensure these foundational building blocks are made widely accessible rather than proprietary.

“Democratising access to AI infrastructure means making the AI infrastructure – compute, datasets and model ecosystem available and affordable, such that it reaches a wide set of users,” the paper notes.

At the centre of the proposal is the idea that AI building blocks should be governed as Digital Public Goods (DPGs) and delivered through Digital Public Infrastructure (DPI), drawing on India’s experience with population-scale digital systems such as Aadhaar and UPI.

“For the physical infrastructure to act as an enabler of socio-economic development, the building blocks of the AI system, i.e., data, compute, and an ecosystem of models and algorithms, need to be made widely accessible and usable,” the paper says.

It calls for governance frameworks that treat AI as Digital Public Goods (DPGs) alongside continued investment in physical infrastructure .

Treating AI resources as DPGs, the paper explains, would mean positioning them as “shared public utilities, rather than proprietary assets”.

This could include open or permission-based data repositories, subsidised access to compute through shared cloud platforms and open-source or publicly accessible model hubs .

The paper also revives the idea of an “AI commons”, first proposed in India’s National Strategy for Artificial Intelligence, describing it as a shared digital infrastructure built through public-private collaboration.

Such a commons, it argues, could give India an advantage by enabling access to curated, interoperable datasets tailored to local languages, sectors and public-interest use cases .