IT and Railway Minister Ashwini Vaishnaw addressing a session during the CII Annual General Meeting & Business Summit 2025, in New Delhi on Thursday
| Photo Credit:
PTI
The Centre on Thursday announced that it is set to procure an additional 14,000 Graphics Processing Units (GPUs) under the India AI mission, in addition to the earlier procurement of 18,000 GPUs.
Speaking at the CII Business Summit 2025, Ashwini Vaishnaw, Minister for Electronics and Information Technology, said, “This is a big change, and AI is here for good.”
“In the first round, our target was to make 10,000 GPUs available to everybody… so against the target of 10,000 GPUs, in the first round itself, we secured 18,000 GPUs and now we are about to get another 14,000 GPUs. So, that’s a substantial number which is now compute facilities available as a common compute facility,” he added.
Model development
Simultaneously, the country will have models which are designed and trained on Indian conditions — Indian culture, languages, nuances and Indian social norms, he noted.
“So, we have taken up development of models, one of the first ones is being developed by Sarvam – there are 3-4 applications which are in the advanced stage of approvals. We are also taking up common datasets as part of this entire process so that you can develop your own applications, which are useful for you, whether it is the agriculture sector, healthcare, or industrial, whichever way. We can develop the models that are suitable for our own needs,” Vaishnaw explained.
The Minister said the government’s intent is to “democratise” access to AI and prepare the country for rapid technological transformation.
In March last year, the Cabinet had approved the IndiaAI mission with a budget outlay of ₹10,371.92 crore to establish a comprehensive ecosystem catalysing AI innovation through strategic programmes and partnerships across the public and private sectors.
The IndiaAI Innovation Centre will undertake the development and deployment of indigenous Large Multimodal Models (LMMs) and domain-specific foundational models in critical sectors.
In January this year, the Centre empanelled bidders, including CMS Computers, Jio Platforms, Tata Communications, E2E Networks, Yotta Data Services, and others for the common compute facility.
Vaishnaw had assured that this computing facility will be the ‘most affordable’ in the world, coming significantly less than $1 (per hour) after 40 per cent cost borne by the government.
The average rate that has come for AI compute is ₹115.85 per GPU hour, compared to a global benchmark of $2.5-$3 per GPU hour (for accessing low-end GPUs). And, for the high-end compute, the average rate is ₹150 per GPU hour, he had said,
As a part of the Mission, the government will be giving a 40 per cent subsidy on this compute price, which means the compute will be available in less than ₹100 per GPU hour.
More Like This
Published on May 29, 2025
