New Delhi: The Ministry of Electronics and IT (Meity) will seek cabinet approval for the India AI program, which includes the setting up of GPU-based servers in public-private partnership mode with an outlay of over Rs 10,000 crore, Minister of Electronics and IT Rajeev Chandrasekhar said on Tuesday.
Working groups formed by the government on Artificial Intelligence (AI) on Friday recommended setting up a three-tier compute infrastructure comprising 24,500 Graphics Processing Units (GPUs).
“India AI program is worth over Rs 10,000 crore. It (the proposal) will be taken with cabinet for approval,” Chandrasekhar said on the sidelines of the inauguration of a semiconductor-focussed software company Synopsys. The minister said the AI program will be based on public-private partnership under which computer infrastructure capacity will be built both in private data centres as well as in data centres that are run by public firm CDAC.
The demand for GPU-based servers has increased as it can process data at higher speed compared to CPU-based servers. At present, the US and China lead in computing infrastructure required for the development of AI technology.
According to Top 500 website, top performing supercomputers are located in the US, Europe, Japan, China, South Korea, and Saudi Arabia where purchasing power parity is high. As per the Top 500 list issued in November 2022, China had 162 supercomputers, followed by the US with 127, while India has three. According to industry estimates, NVIDIA dominates the GPU market with about 88 per cent market share and there is a lag of 12-18 months in getting GPUs from the company due to its high demand across the globe.
The AI working groups have recommended setting up best-in-class AI compute infrastructure at five locations with 3,000 AI Petaflops computing power, with 15 times more capacity than the highest capacity installed at present.
The groups have recommended setting up an Inference Farm (2,500 AI PF) and Edge Compute (500 AI PF) systems.
The government has already spent Rs 1,218.14 crore in the last eight years to set up 24 PetaFlops compute capacity under the National Supercomputing Mission.
