Connect with us

Headlines of the Day

Centre allocates ₹5,000 crore for GPUs and compute capacity in India AI Mission

The Centre will spend about ₹5,000 crore of the ₹10,372 crore fund under the India AI Mission to procure Graphics Processing Units (GPUs) and provide subsidised compute capacity to Indian startups, a top IT ministry official said on Thursday.

GPUs are specialised chips optimised for accelerating artificial intelligence (AI) and machine learning tasks through faster processing. Players like Nvidia, Intel, and AMD are among the world’s biggest producers of GPUs.

“We have almost 5,000 crore earmarked for providing more than 10,000 GPUs that are required to support the creation of compute capacity under the mission,” said Abhishek Singh, Additional Secretary, Ministry of Electronics and Information Technology (MeitY) at the Global IndiaAI Summit held in New Delhi.

“We have almost 5,000 crore earmarked for providing more than 10,000 GPUs that are required to support the creation of compute capacity under the mission,” said Abhishek Singh, Additional Secretary, Ministry of Electronics and Information Technology (MeitY) at the Global IndiaAI Summit held in New Delhi.

He said the government was in the process of floating tenders for procuring the chips but the approach would not be to buy them directly.

“It is not that we will buy chipsets and build up a public-owned, public-run infrastructure. We will want the investments to come from the private players but part of the cost of access to that compute will be subsidised by the government. So that those who need the compute, those who are building models or training models or doing inferencing or doing research or working on algorithms, they will be able to access compute at a cost which is much less than what they are incurring today,” he explained.

Singh said the reason behind this approach was to let the users, enterprises and startups in this case, decide on which GPUs would be fit for their use.

“The way we are building the procurement model is that the users will decide what they want. So, if anybody wants H100, or Gaudi 2, they will be able to access that. If anybody wants to use CPUs to build an AI solution, that will also be possible. Sitting in the government, we will not sit into judgment about what the end users want. Ultimately, the startups and those who are building the solutions, they are the best judge with regard to how they want to train their models, which infrastructure to use,” said Singh.

“We would only do some checks and balances to ensure that it’s not misused, and we don’t start getting into voucher trade, that people get subsidised compute and give it to somebody else,” he added.

Earlier, MeitY Secretary S Krishnan had said that for faster and wider availability of GPU resources in the country, the government was looking at the viability gap funding approach or a voucher-based mechanism for the AI sector. Business Standard

Click to comment

You must be logged in to post a comment Login

Leave a Reply

Copyright © 2024 Communications Today

error: Content is protected !!