India Eyeing Deal With NVIDIA To Procure GPUs For AI Mission

India is looking to strike a deal with US-based chipmaker NVIDIA to procure graphics processing units (GPUs) for its artificial intelligence (AI) mission.

The government aims to provide these GPUs to local startups, researchers, academic institutions and other users at a subsidised rate under the mission worth INR 10,000 Cr, Economic Times reported.

GPU is a specialised chip that renders graphics and images by performing rapid mathematical calculations that are needed in processing AI.

India is exploring two potential methods to furnish AI computing infrastructure to its companies. The first option is a “rent-and-sublet” model, wherein the Ministry of Electronics and Information Technology directly procures GPUs from NVIDIA.

The second option is a marketplace model where the government will encourage companies to strike a renting or subletting deal with the supplier. The government will then provide incentives based on the incremental productivity achieved through these GPUs, under production-linked incentive (PLI) schemes.

NVIDIA has declined to comment on the development.

Last month, NVIDIA India’s top executive Vishal Dhupar said that the company is keen on partnering with more Indian companies to deploy its AI solutions. 

“We’re going to also work with more business houses to bring it (GPUs) here and cater for the sensitivities of this country, helping you to innovate and build here so that you can maximise the disruption that is needed in our country,” NVIDIA Asia-South’s managing director Vishal Dhupar had said.

This move seems important as GPUs like NVIDIA’s Blackwell can be very expensive, up to $40,000 each. Moreover, globally countries like China and the US are making AI compute infrastructure and are spending billions of dollars to secure compute capacity for their companies as well as for national security needs.

China and the US have even started acquiring various NVIDIA GPUs, including the powerful H100 chips.

As creating computing power usually needs lots of GPUs in one location, small businesses and startups can’t afford it. As a result, government subsidies can come in handy here. 

As of now, India lacks adequate GPUs and even the AI startups and companies which are considering using computing power from other countries, are finding it difficult to do so because it costs them a lot. 

It is pertinent to note that in March the Cabinet approved the IndiaAI Mission with an allocation of INR 10,372 Cr over the next five years.

The funds will be used to foster innovation in the homegrown AI ecosystem and implement the Mission’s vision via a public-private partnership (PPP) model. 

The US-based chip maker has already made its first major AI bet in India via Yotta. 

Real estate mogul Niranjan Hiranandani-backed Yotta Data Services has invested capital to acquire the GPU manufacturer’s 16,000 chips.

NVIDIA’s products are highly sought after as they’re essential for the development of AI and are essential for training large language models (LLMs) and building apps such as OpenAI’s ChatGPT.