Govt to place Rs 10,000 crore supercomputing hub plan before Cabinet soon: Rajeev Chandrasekhar

Union minister of state for electronics and information technology Rajeev Chandrasekhar

The Central government will be approaching the Union Cabinet for approval of setting up Rs 10,000 crore- worth supercomputing and quantum hubs in public-private partnership mode, Minister of State for Electronics and Information Technology Rajeev Chandrasekhar said on July 23.

Speaking to reporters on the sidelines of the inauguration of the Synopsys’ Chip Design Centre in Noida, Chandrasekhar said, "The quantum of funds is more than Rs 10,000 crore. It will be presented before the cabinet for approval soon. It will include setting up of graphic processing units (GPUs) in PPP mode with data centres in the private space and public data centres under C-DAC (Centre for Development of Advanced Computing)."

Chandrasekhar also said that C-DAC's PARAM supercomputers will also be used under this plan to boost the artificial intelligence computing capacity.

In October 2023, a working group of AI experts recommended that the Indian government create computing infrastructure of 24,500 GPUs at 17 centres to enable innovation. Under the proposal, about 14,500 GPUs have been recommended for training of AI models and high performance storage, while the rest 10,000 is for AI inferencing.

Apart from that, Chandrasekhar also said that the Information Technology Rules will be notified soon.

Chandrasekhar additionally mentioned that the rules will have provisions on ensuring algorithmic accountability, preventing misinformation (deepfakes).

"No platform should have algorithmic bias or use models that are prone to errors or bias. The onus will be on platforms to ensure such biases do not exist," Chandrasekhar said.

Moneycontrol had earlier reported that the now-delayed Digital India Bill would have had provisions for ensuring algorithmic accountability. Sources had told the publication that the bill would have empowered citizens to opt from being subjected to an algorithm's decisions.