1 week Ago
Bengaluru: India has lagged behind in building its own foundation AI models, compared to China and the US, said Abhishek Singh, additional secretary (MeitY) and CEO of India AI Mission . "China has developed DeepSeek, and the US has multiple companies that have built models like GPT and Gemini. India, however, is playing catch-up," Singh said and added: "We are now providing funding support to startups.
We called for proposals to build Indian foundation models and received a very enthusiastic response. Infosys is also participating in one such proposal." He was speaking at Bharat Intelligence Summit organised by IIT Kanpur alumni at IIT Alumni Centre in Bengaluru Saturday.
You Can Also Check: Bengaluru AQI | Weather in Bengaluru | Bank Holidays in Bengaluru | Public Holidays in Bengaluru He added teams capable of building foundation models to apply for govt support can submit their proposals until April 30. Singh also detailed the India AI Mission, for which the Union cabinet approved an outlay of Rs 10,372 crore, to be implemented across seven verticals. "One key focus is AI compute.
It was felt that India needs at least 10,000 GPUs to help developers and researchers train models and build applications," he said. "In the past year, we empanelled 10 service providers and made over 14,000 GPUs (graphics processing unit)available at a cost much lower than the market rate. These are accessible for around Rs 67 per GPU per hour.
Another 10,000 GPUs will be added soon," Singh noted, emphasizing the importance of affordable compute power. Explaining why innovation has been faster in the US and China, Rajeeva Parasar, founder and CEO, Tricon, said, "The US has investors with deep pockets, who back startups early. In India, investments usually happen later, once startups prove something.
Our startups are often focused on hitting unicorn status fast. In China, maybe because of their communist economy, they can afford to stay invested longer without immediate pressure for returns." The discussion also touched upon the need to introduce AI education at an early stage, given that China is introducing AI in schools at the age of six, and the US is moving in the same direction.
"The ministry of school education is working with NCERT to bring AI and cybersecurity into the curriculum."It will be introduced as a subject for classes 10 and 12, and chapters are being added in classes six and seven," Singh said. He also spoke about YuvAI, the govt's programme for AI literacy among school students.
"Last year, under YuvAI, 50,000 students from govt schools completed an AI 101 course, followed by mentorship from industry partners.".
