The Indian government is considering mandating local storage of AI models. This initiative aims to reduce risks associated with data leaks and prevent sensitive information from leaving the country.

As the rules for the Digital Personal Data Protection (DPDP) Act are expected to be finalized by April, S Krishnan, Secretary of the Ministry of Electronics and Information Technology (MeitY), said that the act will help mitigate personal data breaches.

The DPDP Act includes provisions to restrict cross-border data transfers and mandates strong measures for handling personal information.

Krishnan expressed concerns about large language models (LLMs) developed by Chinese entities, citing potential risks of data misuse. He noted that hosting AI models in India could significantly lower the risk of data being sent abroad.

🔗 Source: Inc42

🧠 Food for thought

A. Data localization joins a decade-long regulatory evolution in India

The proposed AI model localization is part of India’s broader and longer-term data sovereignty strategy, not a sudden shift.

India began this journey years ago when the Reserve Bank of India mandated in 2019 that payment firms store data on Indian servers, giving them six months to comply—a move that immediately raised operational costs for foreign companies1.

This historical context shows the government has been consistently expanding data localization requirements across sectors, from financial services to now artificial intelligence.

The current push comes after data breach concerns have escalated, with studies showing the average cost of a data breach in India rose to $1.7 million in 2017, marking an 8% year-over-year increase2.

The evolution from the original draft data protection law led by Justice B.N. Srikrishna (which recommended at least one copy of personal data be stored in India) to the current comprehensive approach under the Digital Personal Data Protection Act demonstrates India’s maturing regulatory framework2.

This pattern of gradually increasing localization requirements suggests that the AI model localization is part of an ongoing expansion of data sovereignty measures.

B. AI localization represents both national security and economic strategy

The government’s push for AI model localization serves dual strategic goals beyond the stated privacy and security concerns.

India has allocated ₹10,300 crore (approximately $1.25 billion) over five years specifically to develop AI computing infrastructure, including the acquisition of 18,693 GPUs to power local AI development3.