TLDR: The India AI Mission is providing Graphics Processing Units (GPUs) at a subsidized rate of approximately Rs 65-67 per hour, significantly lower than market rates. This initiative, backed by a substantial budget, aims to democratize AI development by making high-end computing accessible to startups, researchers, and academic institutions through a network of empaneled service providers.
India’s ambitious AI Mission is making significant strides in democratizing access to high-performance computing, offering Graphics Processing Units (GPUs) at highly subsidized rates. Under this initiative, the cost per GPU hour can be as low as Rs 65-67, a substantial reduction from the average market rate of Rs 115 per GPU hour and global benchmarks of $2.5-$3 per hour. This affordability is achieved through a government subsidy that can cover up to 40% of compute costs for eligible users, including researchers, startups, MSMEs, and academic institutions.
The mission, approved with a budgetary outlay of Rs 10,371.92 crore (approximately $1.25 billion) in March 2024, has earmarked 44% of its budget, or Rs 4,563.36 crore, specifically for compute capacity expansion over the next five years. The government has opted to empanel private service providers to invest in and provide the compute infrastructure, rather than building its own data centers. This strategy has already yielded competitive bids, with initial rates more than 40% below prevailing market prices even before the application of subsidies.
As of May 2025, the IndiaAI Compute Portal, which became operational in March, has made 14,000 GPUs accessible, with a total of 18,417 GPUs empaneled in the first tranche. Union Minister for Electronics and Information Technology (MeitY) Ashwini Vaishnaw announced in May 2025 that the total GPU capacity in the IndiaAI compute portal has crossed 34,000 units, with seven new companies joining the initiative. These companies will provide additional GPU capacity at the subsidized rates.
Key service providers involved in the mission include Yotta Data Services, E2E Networks, Tata Communications, AWS, CtrlS Datacenters, and Jio Platforms. Notably, Yotta Data Services holds the largest stake with 9,216 GPUs, including 8,192 Nvidia H100s. AWS is supplying 1,200 low-end GPUs through its managed service providers, while Jio Platforms has acquired Nvidia H200 and AMD MI300X GPUs.
The IndiaAI Compute Portal streamlines the allocation of AI resources, allowing users to request compute capacity. As of June 2025, 4,423 GPUs with a total subsidy of Rs 111.86 crore have been allocated to 21 applicants from various sectors. Bengaluru-based startup Sarvam AI, for instance, has already been allocated 4,000 GPUs to build an indigenous foundational model.
Also Read:
- India Accelerates Indigenous Large Language Model Development, Aims for Year-End Launch
- Indian AI Startups Prioritize Open-Source Small Models Amidst Evolving Landscape
While the subsidized pricing is highly advantageous for users, some service providers have expressed concerns about a potential “price bloodbath” in the nascent industry due to the significantly reduced L1 rates. However, the mission’s chief executive, Abhishek Singh, stated that existing empaneled players will be asked to match the new L1 prices, ensuring continued savings for users. The IndiaAI Mission is poised to significantly accelerate AI development and innovation across the country by making advanced computing resources widely accessible and affordable.
