10K GPUs under India AI mission available in 18-24 months, MeitY

The graphics processing units approved under the IndiaAI Mission will be made available in the next 18-24 months, a top government official said on Friday.

Speaking on the sidelines of Tie-Con Delhi-NCR, Electronics and IT Secretary S Krishnan said that the government will invite bids from the industry under the mission and provide viability gap funding for GPU-based computing infrastructure.

Our target is that the GPUs under the AI machine should be available between 18 and 24 months and will not be directly available under this mission, Krishnan said.

The Cabinet has approved the India AI Mission with an outlay of Rs 10,372 crore for five years to encourage AI development in the country.

Under the mission. supercomputing capacity, comprising over 10,000 GPUs (graphics processing units), will be made available to various stakeholders for creating an AI ecosystem.

The demand for GPU-based servers has increased as they can process data at a higher speed compared to CPU-based servers.

Krishnan said that 675 GPUs are available through CDAC’s AIRAWAT which is ranked 75th in the world in terms of supercomputing power.

He said that there is a need to quickly ramp up the AI supercomputing capacity in the country which will be done through public-private partnerships.

In order to make sure that this compute capacity is provided very rapidly and quickly, we intend to do it on a public-private partnership through a bidding transparent process whereby private entities will be selected on the basis of the viability gap funding to establish the compute facility which the government would require is made available to innovators, to startups and to research and for research purposes, at reasonable rates, Krishnan said.

Under IndiaAI Mission, startups, academia, researchers and industry will be given access to the AI supercomputing infrastructure established under the India AI Mission.

The corpus has provision to provide financial support to start-ups to access computing resources for the development of AI.

Start-up funding will not be directly available under this mission. What will be available will be supporting startups accessing AI resources- AI compute, AI foundation model. For direct start-up funding there are many other programs, Krishnan said.

According to industry sources, firms in India are required to pay almost double the cost for accessing GPU or high-performance computers on cloud compared to countries like the US and China due to a shortage of computing infrastructure for AI in the country.

Krishnan said that another important element to the IndiaAI Mission is to prepare India-specific foundation models.

If we look at the kinds of foundation models that have been created globally, many of those are based on Western data. As a result, there are inherent biases and prejudices which they carry, which make it sometimes difficult for them to operate in the Indian environment, he said.

Recently, Google’s AI tool Gemini earned sharp criticism from the government over its objectionable response and bias to a question on PM Narendra Modi. The company has apologised over the response of Gemini AI and conceded that the chatbot “may not always be reliable” in responding to certain prompts related to current events and political topics. PTI