The artificial intelligence-fuelled gold rush to build new data centres — and the risks the projects could pose for local water and power — are becoming a heated municipal election issue, with some Toronto-area candidates calling for a pause or even an outright ban.
Mississauga councillor Martin Reid, who represents the ward where the city’s first “hyperscale” data centre is being proposed, said he was initially supportive of the idea because of the promises of investment in the community and jobs.
But that was before he understood the potential strain on municipal resources that could come with what he calls one of the most “significant planning decisions the city has ever faced.”
After a recent raucous town hall with the developers and community members, Reid became convinced that not only do municipalities need safeguards for them, but they should probably not be allowed at all.
“This is not a good fit for our community,” he said.
What’s powering the data centre push — and pushback?
While small data centres are relatively common, used for everything from backing up your phone to running business software, AI requires far more power, which means physically larger centres that use significantly more energy and water cooling to operate.
Canada has been a couple of years behind the U.S. both in terms of the push to build AI data centres, and in the public backlash to them, but the “gold rush” is here, said Anne Pasek, the Canada Research Chair in Media, Culture and the Environment at Trent University.
While there are only a handful of so-called hyperscale data centres in Canada — a type of data centre that uses more than 50 megawatts or the power needed for about 37,500 homes — dozens more are being proposed.
Proponents have touted the projects as economic engines, driving investment and jobs into local communities that can fear losing an opportunity to another more welcoming city. Microsoft is spending $16 billion on cloud and AI data centre infrastructure in Ontario, including building three large AI data centres in Vaughan, Markham and Etobicoke.
But there is an increasing groundswell of opposition from many different corners — from environmental to health worries to concerns about the existential threat posed to humanity by AI, and how it can be used for harm.
In between are municipal councillors — and given the fall election, would-be councillors — who are faced with a steep learning curve, overwhelming information from all sides and zoning decisions to make.
AI data centres are “definitely going to be an election issue,” said Anna Hardy, a member of environmental advocacy group ClimateFast, who says council candidates in Toronto are already being asked where they stand.
Part of the problem, Hardy said, is that many elected officials are still uninformed about the issue, which has broad implications for municipalities.
One common concern is whether these data centres will drive up power bills. Another is that they will burden the water supply.
Could a one-year pause help?
In Mississauga and most municipalities, data centres are treated like a standard warehouse by the city’s zoning rules. Reid is now proposing a one-year moratorium on data centre approvals to allow staff to come up with suitable regulations, which is expected to be debated at a planning committee meeting at the end of the month.
Two Mississauga councillors running for mayor, Alvin Tedjo and Dipika Damerla, have pushed for the same idea. Current Mayor Carolyn Parrish said she is “looking forward” to a staff presentation on the data centre proposal at the planning committee meeting and has suggested opposition to it is election-driven.
The pause would be similar to one in the works in Hamilton, spearheaded by Hamilton councillor Nrinder Nann.
Nann said she has since received dozens of request for more information from council colleagues across the province and country, including Pickering, London, Windsor, Ottawa, Stratford and Kingston.
“Their residents are asking really important questions that they feel very ill-prepared, as a municipal representative to field,” she said.
“It’s a fast and furious pace at which all this happening,” she said, adding that AI technology is also rapidly evolving, which can change the needs of a data centre as well.
Nann said she wants to make sure that the community benefits that are promised, including jobs, are real and can stand the test of time.
Toronto staff are currently studying whether specific zoning requirements are needed for AI data centres, with a report due back early next year.
How water plays a key role in the debate
Faiz Ahmed, a member of CUPE-funded advocacy group Keep Water Public, is worried that a data centre’s thirst for water could open the door to the province allowing municipal water infrastructure to be used for profit — with unknown consequences for local residents.
While the province has denied it is privatizing municipal water infrastructure, Ahmed’s group is pressuring municipal candidates to agree that there must be public ownership of water services.
“We know that municipal councillors and mayors don’t have the legislative power to do any of this — to actually reverse it on their own,” said Ahmed, but they can increase pressure on the province to install legislative guardrails.
Research chair Pasek notes that some of the AI data centre proposals are being led by speculative developers rather than the end-user company because they think “being a landlord to an AI chip is going to be extremely lucrative.”
Introducing stricter municipal regulations could screen out bad proposals, she said, and a delay to develop those regulations could be a “productive stress test” on whether a project is actually going to benefit the community in the long term.
The timing of the municipal election in October means that interested councillors have little time left to push forward ideas like Nann’s, leaving the future of AI data centres to a newly elected council.
“This is just phase one of these requests coming into municipal governments,” said Nann. “This is just the beginning.”
