Govt unveils mega AI plan to rival global models

India on Thursday outlined global AI ambitions with plans to build its own foundational model that could take on the might of ChatGPT, DeepSeek R1 and others, as it lined up the “most affordable” common compute facility powered by 18,693 graphics processing units (GPUs) to be used by startups and researchers.

Most affordable

Advertisement

The common compute facility, powered by 18,693 graphics processing units, will be made available at a fraction of the global cost benchmarks. Ashwini Vaishnaw, IT Minister

The move came less than a week after a Chinese artificial intelligence (AI) lab launched a low-cost foundational model, DeepSeek. The government said it had decided to build a domestic large language model of its own as part of the Rs 10,370 crore AI mission. Minister for Information and Technology Ashwini Vaishnaw said the government had selected 10 companies, which would supply 18,693 graphics processing units (GPUs) — high-end chips needed to develop machine learning tools that can go into developing a foundational model of this large language model.

He said roughly 10,000 of the 18,693 GPUs approved to be empanelled were ready to be installed. Besides, the government had also selected 18 application-level AI solutions for the first round of funding.

The common compute facility (powered by 18,693 GPUs) would be made available at a fraction of global cost benchmarks, Vaishnaw assured. The compute facility would be “most affordable” coming significantly less than one dollar (per GPU hour) after 40 per cent cost borne by the government. “Making modern technology accessible to everyone, that is the economic thinking of our PM... Ours is the most affordable compute facility, at this point of time,” Vaishnaw claimed.

“Over the last one and a half years, our teams have been closely working with startups, researchers and professors. Today, we are calling for proposals to develop our own foundational model. The model will take care of the Indian context, languages and culture, devoid of biases,” he added.

The minister further said the government was in touch with at least six developers for building the foundational model, which could take anywhere between four to eight months. He also announced an AI Safety Institute, which would enable stakeholders to develop tools, frameworks and processes for AI safety. Eight projects had been undertaken under “Indigenous Governance of AI” in this regard.