Policy Update
Prisha Sachdeva
Background
GPU infrastructure is a critical enabler of modern AI development because it provides the computational power required to train large models, process vast datasets, and support high-volume experimentation. This makes it the backbone of any serious AI ecosystem, and its expansion has direct implications for the scale and speed of AI market growth.
The IndiaAI Mission was launched in March 2024 to address existing gaps in data, research, and skills, enabling the benefits of artificial intelligence (AI) to contribute to the country’s growth. The Ministry of Electronics and Information Technology (MeitY) identified seven key pillars of the mission: IndiaAI Compute, Foundation Models, AIKosh, the IndiaAI Application Development Initiative, FutureSkills, Startup Financing, and Safe and Trusted AI.
The mission’s overarching goal is “AI for All.” With an outlay of ₹10,371 crore over five years, it aims to build a comprehensive AI ecosystem in India. The AI market is currently valued at approximately $120 billion, growing by more than 20% annually, and is expected to reach $1.5 trillion by 2030. In addition, the global market for AI-specialised hardware is projected to grow ninefold, reaching $90 billion by 2030.
The IndiaAI Compute pillar was officially launched on 6 March 2025 by Union Minister Ashwini Vaishnaw under MeitY. It aims to make high-performance computing resources more accessible and promote self-reliance in artificial intelligence.
This pillar seeks to eliminate the financial and geographical barriers that hinder domestic AI innovation.
Functioning
The IndiaAI Compute pillar aims to build a strong national development environment by stimulating AI innovation through strategic public–private partnerships. By widening access to cloud-based compute, storage, networking, and platform services, improving data quality, and promoting homegrown tools, the pillar supports faster development of AI capabilities. The IndiaAI Compute Portal offers eligible users subsidised access to these services via a unified interface: https://compute.indiaai.gov.in/login.
Providers are selected through a public empanelment route. Responding firms — including global chip and cloud vendors as well as Indian data-centre operators — are evaluated under a Request for Empanelment (RFE) and contracted to supply GPUs and associated cloud services. Competitive rounds of empanelment have steadily increased the total available capacity; suppliers include multinational chipmakers and domestic providers such as Yotta.
How to apply for GPUs
1)Sign in to the IndiaAI Compute Portal using DigiLocker, e-Pramaan, or Jan Parichay.
2)Complete and submit the end‑user registration form with required documents for verification.
3)After approval, apply for compute services via the portal. Provide your requested resources and a draft bill calculated from the published rate card. You may also request a subsidy.
4)The Project Management and Evaluation Committee (PMEC) reviews requests against service‑level agreements (SLAs) and approves allocations. Large or complex requests may require a detailed project proposal and a bill of materials.
Beneficiaries
The programme serves a wide range of users, including academic institutions and researchers,students and IndiaAI Fellowship awardees,startups and MSMEs,early‑stage researchers and entrepreneurs,government bodies and agencies
Subsidised pricing (around ₹65/hour) lowers the cost of experimentation and model development, making advanced compute more accessible to organisations that previously lacked the budget for high-end resources.
Performance
The Compute pillar has outperformed its initial targets and rapidly scaled national compute capacity.
| Timeline | Number of GPU`s |
| Launch(Target) | 10,000 |
| Before may 2025 | 18,417 |
| 30 May 2025 | 34,000 |
| December 2025 | 38,000 |
(Compiled by the author)
India’s national AI compute capacity has grown substantially since the inception of the IndiaAI Compute pillar. Starting with an initial target of 10,000 GPUs, the capacity had already reached 18,417 GPUs before exceeding 34,333 GPUs by May 2025, following the addition of 15,916 new units. By December 2025, the capacity had expanded to more than 38,000 GPUs—nearly 3.8 times the original target—with continued availability confirmed as of February 2026.
This infrastructure is made available to Indian startups, academic institutions, and researchers at a subsidised rate of approximately ₹65 per hour, making high-end AI computing significantly more affordable than standard market rates.
Impact
The Compute pillar’s vision of developing a robust AI infrastructure appears to be progressing strongly. Through a public–private empanelment process, the initiative has brought together global technology companies, cloud service providers, and Indian firms to expand access to GPU capacity and other essential computing resources. This collaborative approach has helped reduce both the financial and geographical barriers that previously limited participation in AI development.
The pillar has exceeded its original target by nearly four times and has directly enabled progress in other areas of the IndiaAI Mission, most notably the Foundation Models pillar. More than 500 proposals—including Sarvam AI’s 120-billion-parameter sovereign model—rely on this expanded computing capacity.
This demonstrates that the initiative is not merely establishing GPU infrastructure; it is also ensuring that these resources are actively used to develop advanced AI models, tools, and applications. By making high-performance computing available to startups, academic institutions, researchers, MSMEs, and government agencies, the Compute pillar is helping create a broader and more inclusive AI innovation ecosystem in India.
The subsidised pricing of approximately ₹65 per hour has further improved access to high-end AI computing for startups and academic researchers who previously could not afford such infrastructure. In addition, the application and approval process enables eligible users to request compute services according to their specific project requirements and apply for subsidies where applicable. By reducing the cost of experimentation, research, and model development, the initiative is supporting innovation across the wider IndiaAI Mission and strengthening India’s capacity to build indigenous AI technologies.
Emerging issues
Although a substantial number of GPUs are being established, the empanelment process appears to be lengthy and complicated. This may discourage individual users and create an additional barrier for small startups and researchers, particularly when compared with larger industries that have greater financial and administrative resources.
Another important concern is the limited availability of data on the actual utilisation of the GPUs. While detailed figures on the total GPU capacity have been reported, there is little information on how frequently these resources are being used, by whom, or for which projects. This raises an important question: is the available GPU capacity being effectively utilised, or does a significant portion of it remain idle?
The distribution of participating public–private partners may also create regional barriers. If access is concentrated among companies located primarily in Tier 2 and Tier 3 cities, users from other regions may face difficulties in accessing the necessary infrastructure and support. Although expanding access beyond major technology hubs can promote inclusivity, uneven distribution and limited availability in certain areas could restrict participation in the AI ecosystem.
Furthermore, a significant proportion of the GPUs is supplied by foreign companies. While partnerships with global technology firms have helped India rapidly expand its computing capacity, heavy dependence on imported hardware raises concerns about the mission’s long-term goal of creating a self-reliant AI ecosystem. To strengthen this objective, India would need to invest further in domestic semiconductor manufacturing, AI hardware development, and indigenous computing technologies.
Therefore, the success of the IndiaAI Compute pillar should not be measured solely by the number of GPUs installed. It should also be assessed on the basis of actual utilisation, regional accessibility, participation by smaller organisations, transparency in allocation, and progress towards reducing dependence on foreign hardware.
Way forward
In the next phase, the Government should publish detailed data on how the GPUs are being used. Greater transparency regarding GPU allocation, utilisation rates, participating organisations, and the projects being supported would make it easier to assess the programme’s effectiveness and track its progress.
The Government should also focus on financing Indian companies that develop AI hardware products. Investment in domestic GPU-related technologies, semiconductor manufacturing, and other essential components would help reduce dependence on foreign suppliers and support the long-term goal of achieving self-reliance in AI infrastructure.
In addition, efforts should be made to extend AI model development and computing facilities to rural and underserved areas. Taking these opportunities beyond major cities and established technology hubs would make the IndiaAI Mission more inclusive and enable students, researchers, startups, and institutions from diverse regions to participate in the country’s AI ecosystem.
The empanelment process should also be simplified and streamlined. Clearer eligibility criteria, a faster application process, reduced documentation, and time-bound approvals would make it easier for small startups, individual researchers, and academic institutions to access computing resources. These measures would ensure that the expansion of GPU capacity translates into wider and more meaningful participation in India’s AI development.
References
Ministry of Electronics and Information Technology. (2025, May 30). India’s common compute capacity crosses 34,000 GPUs [Press release]. Press Information Bureau. https://www.pib.gov.in/PressReleasePage.aspx?PRID=2132817®=3&lang=1
Ministry of Electronics and Information Technology. (2025, December 30). Transforming India with AI: Over ₹10,300 crore investment & 38,000 GPUs powering inclusive innovation [Fact sheet]. Press Information Bureau. https://static.pib.gov.in/WriteReadData/specificdocs/documents/2025/dec/doc20251230747901.pdf
Ministry of Electronics and Information Technology. (n.d.). [Press release]. Press Information Bureau. https://www.pib.gov.in/PressReleasePage.aspx?PRID=2227612®=48&lang=2
IndiaAI. (n.d.). Union Minister of Electronics & IT, Railways and I&B announces the availability of 18,000 affordable AI compute units. https://indiaai.gov.in/article/union-minister-of-electronics-it-railways-and-i-b-announces-the-availability-of-18-000-affordable-ai-compute-units
Ministry of Electronics and Information Technology. (n.d.). IndiaAI Mission expands AI ecosystem with affordable compute and startup support [Press release]. Digital India. https://www.digitalindia.gov.in/press_release/indiaai-mission-expands-ai-ecosystem-with-affordable-compute-and-startup-support/
IndiaAI. (n.d.). IndiaAI Compute Capacity. https://indiaai.gov.in/hub/indiaai-compute-capacity
Contributor:
Prisha Sachdeva is a Research & Editorial intern at IMPRI. She’s pursuing Bachelors in Psychology (Honours) from the University of Delhi. Her interest lies in cognitive science and human behavior with a focus on evidence-based policy and behavioral research.
Acknowledgement:
The author sincerely expresses gratitude to the reviewers and the editorial team for their valuable comments, constructive suggestions, and continuous guidance throughout the preparation of this article. Their insightful feedback significantly enhanced the clarity, organisation, and analytical quality of the manuscript. The author also acknowledges the support and encouragement received during the research and writing process, which contributed to the successful completion of this work.
Reviewers: CB Kavin Adithya and Vyomini Nathwani
Disclaimer:
All views expressed in the article belong solely to the author and not necessarily to the organisation.
Read More at IMPRI:
India–UK Comprehensive Economic and Trade Agreement 2026: Trade and Investment Cooperation
