Kevin O'Leary dismisses AI data centre water concerns, says modern facilities are 'like a golf course'

As governments and technology companies race to expand artificial intelligence infrastructure, one of the industry's most prominent investors is challenging a key criticism levelled against the sector. Kevin O'Leary, the Canadian businessman best known for Shark Tank, argues that modern AI data centres have moved well beyond the water-intensive designs that fuelled environmental concerns over the past decade.

Speaking during a recent appearance on Glenn Beck's radio programme, O'Leary said technological improvements in cooling systems have fundamentally changed how these facilities operate, insisting that the industry's reputation for excessive water consumption no longer reflects reality.

The comments come at a time when AI companies are investing billions of dollars in new computing infrastructure while facing growing questions over the environmental impact of increasingly power-hungry data centres. Water use has become a particularly contentious issue in regions already grappling with drought and pressure on public resources.

More from Tech

Modern cooling has changed the equation, says O'Leary

O'Leary, whose investment firm O'Leary Ventures is developing AI data centre projects in Utah and Canada, said criticism of the industry's water use is largely based on older facilities rather than today's designs.

Discussing the first phase of his planned Utah project, which is expected to provide 1.4 gigawatts of computing capacity, O'Leary argued that its water requirements would be comparable to those of a golf course.

"Because data centres, not just my project," O'Leary said, "any new data centre now is like a golf course."

He attributed that shift to advances in cooling technology, arguing that facilities built more than a decade ago created a perception that no longer applies to newly constructed sites.

"The new modern era data centre does not require anywhere near the water that was used 15 years ago in Virginia, where data centres got a bad name because of how much water they consume," he said.

Data centres require cooling systems to prevent servers from overheating and also consume water indirectly through the electricity needed to power their operations. In recent years, several technology companies have promoted newer cooling methods, including closed-loop systems, which are designed to reduce reliance on evaporative cooling and lower overall water consumption.

The debate has also featured other leading AI executives. Earlier this year, OpenAI chief executive Sam Altman dismissed claims circulating online that AI queries consume large amounts of water.

"Water is totally fake," Altman said while responding to concerns about AI infrastructure. "It used to be true, we used to do evaporative cooling in data centres, but now that we don't do that, you know, you see these like things on the internet where, 'Don't use ChatGPT, it's 17 gallons of water for each query' or whatever."

However, concerns over the industry's environmental footprint have not disappeared. A previous investigation by Business Insider found that several of the largest US data centre campuses had permits allowing daily water consumption exceeding what nearly 49,000 Americans would typically use in the same period.

AI projects should build power, not just consume it

Beyond defending the industry's environmental record, O'Leary argued that developers should shoulder greater responsibility for the energy demands created by AI infrastructure.

Rather than placing additional pressure on existing electricity grids, he said future data centre projects should include investment in their own power generation.

"The new rule of engagement should be this — as part of the national solution is — if I'm going to bring a data centre anywhere, I've got to bring my own power. I got to spend billions to build power, and then give some of it back to the grid, so we can stay competitive," he said.

His remarks come as many countries struggle to expand electricity generation quickly enough to keep pace with soaring AI-driven demand, prompting governments and utilities to rethink long-term energy planning.

Meanwhile, O'Leary is also dealing with legal challenges linked to his Utah project. He is facing a defamation lawsuit over comments made in several media appearances in which he alleged that some opponents of the development were backed by the Chinese Communist Party. Before legal proceedings began, O'Leary acknowledged on social media that he had no evidence to support claims that critics of the project were funded by China.

Jeffrey Neiman, an attorney representing O'Leary, previously told Business Insider that his client intends to "defend this lawsuit vigorously" and pursue counterclaims, describing the legal action as a "cash grab".

As investment in AI infrastructure accelerates worldwide, debates over water, electricity and environmental sustainability are expected to remain central to discussions about how the next generation of data centres should be built.