Toronto Metropolitan University's Independent Student Newspaper Since 1967

A collage about data centres and water consumption. A glass of water with brown colour mixing into it, as well as a data centre are spliced in front of a recycling logo.
(PHOTO COMPOSITION BY DANIEL OPASINIS/THE EYEOPENER)
All Business & Technology

Student paper explores possibility of ‘water-neutral’ AI data centres

By Yuna Cho

Gianluca Provenzano, a fourth-year industrial engineering student at Toronto Metropolitan University (TMU), presented a “closed-loop system” at the 2026 Canadian Nuclear Society Annual Conference this summer. The system would reuse heat and cooling water generated by data centres.

“Data centres produce a lot of heat. It’s actually a really big topic right now in sustainability groups,” he said in an interview with The Eyeopener.

As access to water resources becomes increasingly challenging, Provenzano said, freshwater availability could eventually become a serious concern.

According to IEEE Spectrum, up to 60 per cent of the water used in data centres can be lost due to evaporation and not returned to the water system.

“This approach could reduce water losses from consumption to nearly zero, while also having the advantage of being compatible with existing technologies” which Provenzano defined as “water-neutral.”

Erfan Shahab, an editor at Soft Computing and an industrial engineering PhD student at TMU, said “the concept is worth serious consideration.”

He said one of the strengths of the proposed system is that it is based on existing technologies and engineering principles rather than relying on hypothetical or unproven technologies.

However, Shahab said there are more steps before the concept can move toward actual implementation. He noted that whether commercial facilities can achieve complete water neutrality would need to be practiced through detailed engineering analysis and real-world operational data.

“The biggest challenge would probably be integration,” Shahab said. Connecting multiple technologies and making sure they can work reliably together will be key to whether the system can actually be put into use. He said that even if the system is technically efficient, its commercial value could be limited if it creates operational problems.

In Canada, up to $700 million was invested in new or expanded data centres in 2024, according to CBC. However, rapid development in the AI industry has raised a range of concerns. Municipalities in Ontario are studying guardrails and policies around data centres.

Dario Amodei, CEO of AI company Anthropic, has called for a slower pace of artificial intelligence development. He warned in a post on his website that if society fails to manage the potential risks of the technology, “we could all die in the immediate future.” This since has been criticized as a publicity move.

Maheer Kazi, a master of electrical engineering student at TMU, said he is largely opposed to the use of AI, raising concerns about issues such as ‘AI psychosis.’ This refers to when generative AI systems induce, amplify, or reconstruct the psychotic experiences which diminish one’s sense of self-existence. This can result in the loss of immediacy of experience that mirror the nature of psychosis and schizophrenia.

Kazi raised concerns about people using generative AI to spread disinformation through creating fake and misleading images.

Eric Alchin, a fourth-year marketing management student, said his own opposition to the technology is over its potential sociocultural impact. “People will just keep asking [AI] until they get a ‘yes’ answer,” he said.

He argued that as people increasingly rely on AI to do their thinking, they may become more vulnerable to misinformation and more likely to reinforce their existing biases.

As more researchers and experts raise concerns about the pace and scale of AI development, “the issue should be taken seriously,” he said. 

Provenzano said the focus should not be on stopping AI development, but on asking “how much environmental impact society is willing to accept” to make the technology accessible. 

Though the paper proved the system technically feasible, he said remaining hurdles have to be addressed before implementation.

WHAT'S HAPPENING ON CAMPUS?

Sign up for our newsletter

We don’t spam! Read our privacy policy for more info.

Leave a Reply