CHENNAI: In a bid to establish itself as a regional hub for artificial intelligence (AI), the state is planning to develop large language models (LLMs) tailored for Tamil computing, while ramping up investments in graphics processing infrastructure and overseas academic partnerships. In this regard, the government is looking to tie up with many organisations, including the National University of Singapore.

Pradeep Yadav, additional chief secretary and head of the state’s newly expanded artificial intelligence, information technology and digital services department, told TNIE the government is working towards building Tamil-language AI tools, datasets and digital public infrastructure, an initiative that comes amid rising global interest in vernacular AI ecosystems and concern among Indian policymakers about dependence on English-centric models dominated by global technology giants.

Yadav said Tamil Nadu already possesses GPU capacity and is exploring further investments to scale up its computing backbone. Discussions are also under way with AI start-up Sarvam AI as part of the state’s broader strategy to deepen indigenous capabilities.

GPUs (graphics processing units) are specialised chips capable of handling the vast parallel computations that training and running large language models demand, and access to them has become a flashpoint in the global AI race.

He added the state-run Tamil Virtual Academy, which already runs Tamil computing projects, would play a central role in expanding Tamil-language AI resources. The push represents one of the clearest signals yet from a state government of ambitions to build sovereign language AI capabilities.

Tamil Nadu, however, is yet to implement the union government’s National AI Mission, approved in 2024 with an outlay of about `10,300 crore. The centrally-sponsored programme is designed to build large-scale GPU infrastructure, support Indian AI start-ups and develop indigenous large language models.

This story is reported by C Shivakumar