Is AI sustainable? Why the debate goes beyond data centres and energy use

As artificial intelligence reshapes industries and everyday life, questions are growing about the environmental cost of this technological revolution. Can AI be made sustainable, or is "Green AI" simply the newest corporate buzzword?

At a panel discussion during BT India's Most Sustainable Companies Summit and Awards on AI sustainability, experts like Juhi Joshi, Partner Climate Collective also; Namrata Rana, National Head ESG KPMG in India; Rajesh Kumar Jha, Country Sustainability Manager ABB India; and Suruchi Bhadwal, Director Earth Science and Climate Change at TERI agreed on one thing - AI is here to stay. The real challenge is ensuring that the infrastructure powering it does not come at an unsustainable cost to energy, water and natural resources.

"Digital services are the requirement of the day. Our data centres are required to run it and AI is part of it very much," said Bhadwal while acknowledging AI's growing environmental footprint. She stressed that the focus should now be on technologies that reduce energy and water consumption. "We need to put our heads together to understand what we can do about it."

While concerns around AI's resource footprint are mounting, industry leaders argue that the conversation must also focus on how efficiently the underlying infrastructure is designed and operated. For them, the question is not whether AI should grow, but how that growth can be powered sustainably.

For Jha, the discussion should move beyond headline numbers around model sizes and data centre capacity. "People are talking about how big the model is, how big the data centre is, but the data centre intelligence per megawatt or per kilowatt is what we should be talking about," he said. Jha argued that smarter grids, improved energy efficiency, storage systems and digitalisation would be critical to making AI infrastructure more sustainable.

Others argued that the sustainability debate should not obscure AI's potential benefits. As per Rana, AI should be viewed as part of the solution rather than the problem. "I don't think it's AI versus sustainability. I think we need to view these together," she said adding that AI applications can help reduce energy use across sectors such as manufacturing, transport and agriculture, potentially delivering a larger climate benefit than the resources consumed by the technology itself.

Joshi pushed back against framing sustainability as a moral choice. "We have to look at it from a pure economics point of view. That is when sustainability is scalable," she said. Joshi argued that innovation, startups and market incentives would ultimately drive the development of cleaner AI infrastructure.

Yet despite optimism around technology and innovation, panellists acknowledged significant gaps. India currently lacks a dedicated framework governing the sustainability footprint of AI and data centres. "We do not have the policies and the regulation clearly defined around it," Bhadwal noted.

The consensus emerging from the discussion was that Green AI is not yet a reality. But neither is it merely a buzzword. It remains a work in progress, one that will require coordinated action from policymakers, industry, researchers and technology developers if India's AI ambitions are to be achieved sustainably.