Calling data a strategic resource in the age of AI, the Economic Survey 2025-26 pitched a new framework to govern how Indian data is used, shared and monetised, without hurting startups and innovation.

The Survey, prepared by chief economic advisor V Anantha Nageswaran’s office and tabled by finance minister Nirmala Sitharaman in the Parliament today, said India should allow data to move across borders but ensure that the value created from Indian data is not lost overseas.

At the same time, it said that smaller startups, research labs and firms building Indian or sovereign AI models should face lighter compliance so innovation is not stifled.

According to the Survey, India has over 100 Cr people with internet access, making it one of the most digitally intensive markets in the world. This large user base also means India generates vast amounts of data.

This scale and diversity of data is a major advantage for companies building AI products for the Indian market.

However, the Survey also warned that global digital services and the concentration of AI development among a few large multinational firms have created imbalances. Much of the value from data generated in India is captured outside the country. This makes it necessary to rethink data governance, it added.

India has so far avoided strict data localisation rules to support ease of doing business, attract investment and benefit from global data flows. But the economic document said that this approach now needs to be expanded so that India can retain more value from its own data.

The document also flagged a coming shortage of high-quality human-generated data. As AI models struggle when trained only on synthetic data, companies will increasingly look for fresh human data.

In this context, Indian data becomes even more valuable, and policy needs to reflect this reality.

Instead of forcing all data to stay within India, the Survey proposed a shift towards what it calls “accountable portability”. Under this approach, data can move across borders, but companies processing Indian data at scale, especially for training large or general-purpose AI models, must remain accountable to Indian regulators.

Such companies would need to ensure Indian data is auditable, traceable and retrievable, even if it is processed outside the country.

(The story will be updated soon)