Small data centres could help narrow India’s AI divide by allowing more regions and institutions to use artificial intelligence without relying entirely on large, energy-intensive facilities, IIT Madras director V Kamakoti said at an AI governance conclave held at the institute’s Chennai campus.
Kamakoti said large data centres would remain important for training artificial intelligence models, but smaller facilities, described as “pico-clouds”, could be sufficient for using or inferring from those models. The distinction is significant because large data centres require substantial electricity and cooling infrastructure, while model use can be supported through more distributed computing capacity, according to his remarks.
“We will need large language models to train AI models. But small data centres or pico-clouds are enough to use the models,” Kamakoti said. He added that India should pursue responsible AI adoption in a way that avoids adverse effects on the country’s energy requirements and other natural resources.
The comments place computing infrastructure alongside governance as a central issue in India’s AI expansion. While the development and training of large language models can be concentrated in major facilities, their practical use across sectors may require computing access closer to institutions, businesses and public services. Kamakoti’s remarks focused on the potential role of smaller data centres in providing that access, although he did not announce a specific deployment plan, capacity target or implementation timeline.
The proposal also points to the infrastructure demands behind the growing use of AI. Data centres need reliable electricity and cooling systems, and the scale of those requirements varies according to whether a facility is training large models or applying existing models to tasks. A more distributed network of smaller facilities could therefore form part of the country’s approach to expanding AI use, subject to decisions on standards, investment, connectivity and administration.
Kamakoti made the remarks while launching a conclave on AI governance at IIT Madras. The institute is also planning to release 500 videos explaining how AI is improving different areas of life in India and how technology can improve quality of life. The Centre for Responsible AI at IIT Madras is partnering with Better India on the project.
Each video will focus on a different sector, including agriculture, space, legal services and manufacturing, Kamakoti said. The planned series is intended to illustrate practical applications of AI across a range of activities rather than limit the discussion to model development or computing infrastructure.
The conclave also addressed the effect of AI on employment. Professor B Ravindran, head of the Wadhwani School of Data Science and AI and the Centre for Responsible AI at IIT Madras, said AI would not necessarily replace all jobs because individual jobs contain different kinds of skills and AI would not be equally capable of replacing each of them.
“Each job will have to be reimagined with the impact of AI,” Ravindran said. His comments linked AI adoption to changes in the composition of work, rather than treating employment as a simple question of whether a job would disappear.
IIT Madras’s remarks leave the next steps for pico-cloud adoption open. The institute’s immediate announced initiative is the 500-video series, while the wider question of how India develops distributed, energy-conscious AI infrastructure remains part of the country’s emerging AI governance debate.


