The IndiaAI Mission has built a large public pool of computing power, but without fair access rules, ethics review and support for state universities, most of India’s academic system will be left out.
In October, data from the IndiaAI Compute Portal showed how the government’s AI computing subsidy has been shared so far. Nine organisations building foundation models had received about ₹1,058 crore. Researchers, students and fellows together had received roughly ₹10 crore, or 0.8 per cent of the subsidy.
An IIT faculty member told The Print that registering on the portal took more than a month, even for an IIT. If a premier institute finds the system hard to use, it is not difficult to guess where a state university in a smaller city stands. The numbers matter because computing power is the backbone of the mission. Of the ₹10,371.92 crore the Union Cabinet approved in March 2024, about ₹4,563 crore is earmarked for compute capacity, and more than 38,000 GPUs have been onboarded on the portal.
The aim is sound: in today’s AI economy, access to computing power decides who builds models and who only uses models built elsewhere. But hardware alone does not produce research. Without institutions able to use it, subsidised compute risks repeating the digital divides of earlier technology rollouts, with the best-known institutes absorbing the capacity while state universities and regional colleges are left behind.
Who gets the computers?
India’s research computing has long been concentrated. Supercomputers built under the National Supercomputing Mission, such as PARAM Ganga at IIT Roorkee and PARAM Pravega at the Indian Institute of Science, sit mostly in centrally funded institutions. Yet most of India’s students study in state universities and the colleges affiliated to them, and these remain at the edge of the country’s computing networks. The IndiaAI Mission could deepen this gap unless its access rules stop rewarding prestige.
Training or adapting an AI model is not simply a matter of logging in. It needs fast campus networks, a reliable power supply and technical staff who can manage the work. Most state university departments run on fixed annual maintenance budgets, with little money to keep paying for cloud services once a grant ends. The portal’s design adds to the burden: researchers must pay a share of the cost themselves, which the same IIT faculty member described as difficult for younger academics. Left as it is, the subsidised pool will flow to the institutions that need help least.
No one to ask the hard questions
The second gap is governance. Many universities abroad have begun adapting their research ethics boards to review AI projects, looking at where training data comes from and whether models carry bias. Most Indian universities still rely on ethics committees designed for clinical trials or social surveys. These committees are rarely equipped to judge whether a dataset scraped from regional language websites was collected with consent, whether it infringes copyright, or whether a model released openly could be misused.
There are practical risks as well. Without legal advice, departments may sign cloud contracts that leave rights over publicly funded research with the provider. Student and staff data are also increasingly being run through commercial AI tools without clear safeguards. That will matter far more once the main obligations of the Digital Personal Data Protection Act take effect in May 2027, when universities, like any organisation handling personal data, will have to show consent and security safeguards.
What others are doing
Other countries treat access to public compute as a design question, not just a purchasing one. In the United States, the National AI Research Resource, launched as a pilot in January 2024, set out to reach researchers at under-resourced institutions, including two-year and four-year colleges, and not only the best-funded universities.
The European Union’s AI Factories, run through the Euro HPC Joint Undertaking, offer free access for publicly funded research alongside a separate route for start-ups. In both cases, public computing power comes with clear rules about who gets it and on what terms.
The Way forward
There is a fair argument for backing the institutions that can use compute immediately; frontier models will not be built in under-equipped labs. But a mission paid for with public money should also widen the base of Indian research, not just reward those already at the top. To stop the IndiaAI Mission from becoming an expensive hard waresilo, New Delhi should pair its compute with institutional reform.
First, access should not depend on who can navigate an open portal fastest. The Ministry of Electronics and Information Technology and the University Grants Commission should create a voucher scheme that reserves a fixed share of compute, say 30 per cent, for state public universities and interdisciplinary research centres, with the researcher’s copayment reduced or waived for them.
Second, the UGC, in consultation with the Office of the Principal Scientific Adviser, should issue national guidelines for setting up Institutional AI Review Boards in central and state universities. Before public compute is allocated to a project, these boards should check where its data comes from, whether it complies with the data protection law, and whether the model is safe to release.
Third, rather than giving each department a small cluster it cannot afford to run, the Ministry of Education should set up regional compute hubs in selected state capitals, which affiliated colleges can use remotely without building their own cooling and power systems.
Fourth, the government should issue standard procurement and licensing templates so that research produced on public compute is not locked into the terms of global cloud providers. Computing power alone does not make a country a technology leader. Without fair access and the institutions to use it well, India risks paying for tomorrow’s infrastructure while leaving most of its universities governed by yesterday’s rules.
*Komal Karwa is a public policy researcher associated with the Policy, Politics and Governance Foundation (PPGF) and South Asia Monitor. An appointed UGC NEP SAARTHI Ambassador and former Lok Sabha legislative researcher, she is currently a postgraduate scholar in Political Science.