India must build AI models in local languages, social norms: Vaishnaw

Business Standard

29-05-2025

India needs to develop artificial intelligence (AI) models that are rooted in domestic languages, culture, conditions, nuances, and social norms, Union Minister for Electronics and Information Technology Ashwini Vaishnaw said.
'We have taken up development of models. One of the first is being developed by Sarvam. There are three or four applications which are in the advanced stages of approvals. We are also taking up common datasets as a part of this entire process,' Vaishnaw said at the CII Annual Summit 2025.
Akin to the advancements made in the semiconductor and electronics industry, AI will bring 'humongous changes' to society and industry, he said.
'I think that kind of change is coming. We should stay prepared for that change. Whichever industry or sector we are in, we must be prepared,' the Union minister said.
In order to ensure access to high-end compute for all, the government has targeted enabling access to 10,000 graphics processing units (GPUs) under the IndiaAI Mission. Against this target, it received bids for more than 18,000 GPUs in the first round of bidding, he said, adding that companies have placed bids to procure nearly 14,000 GPUs in the second round.
The IndiaAI Mission was approved by the Union Cabinet in March 2024, with a capital outlay of ₹10,738 crore. The IndiaAI Compute Portal will provide access to these GPUs as well as AI cloud services for start-ups, application developers, researchers, and students. It will allow end users to access a wide range of high-end GPUs and facilitate AI research and innovation in the country.
The first Made in India chips in the 28–90 nm range will be rolled out by the end of this year, the Union minister said.
'There are many people who criticised us for aiming at semiconductor manufacturing. We targeted a particular segment, which has 60 per cent of market volume, using a targeted approach. Today, we have six units under construction. The first Made in India chip of 28–90 nm will roll out this year. We started manufacturing in 2022,' Vaishnaw said.