AI data centres and public benefit

Manitoba Premier Wab Kinew recently rejected a proposed AI data centre near Winnipeg, citing concerns over water use, noise, and a perceived lack of public benefit.

The instinct to protect communities from unwanted development is understandable. But when the objections collapse under scrutiny, and when the stakes include Canada’s standing in the global AI economy, the decision deserves a harder look.

Critics of data centres frequently invoke water consumption, and there is genuine nuance worth discussing.

Large facilities do use water for evaporative cooling. But comparison matters. A typical 18-hole golf course uses approximately 300,000 gallons of water per day during summer months.

A single pound of almonds, found in every Manitoban grocery store, requires roughly 1,900 gallons of water to produce, according to the Pacific Institute.

Neither the golf industry nor almonds attract the same regulatory alarm.

More to the point, Manitoba is arguably the ideal location in North America for a low-water data center.

The province’s cold climate makes free-air cooling viable for much of the year, dramatically reducing or eliminating water-intensive evaporative systems.

Rejecting this proposal on environmental grounds is not environmentalism. It is a missed opportunity dressed up.

The noise objection fares no better. A data centre perimeter fence typically mitigates noise to 45-to-65 decibels of ambient sound, which is roughly equivalent to a normal conversation or a refrigerator hum.

A city bus accelerating from a stop produces around 80 decibels. A passing motorcycle clocks in at 95. Active construction, the kind that builds every hockey arena, hospital, and transit line Manitoba’s government approves routinely, regularly exceeds 85 to 100 decibels.

No one proposes banning these things, which is why the noise objection to a data centre is not a serious policy argument.

The claim of “little public benefit” is the hardest to justify.

Data centre construction draws on skilled tradespeople such as electricians, HVAC mechanics, pipefitters, ironworkers, and project managers, whose wages sit well above the provincial average.

In the United States, where comparable data centre booms have played out in cities similar in size to Winnipeg, the Bureau of Labor Statistics reported a mean annual wage of US$65,360 ($90,000 CAD) across construction and extraction occupations in May 2025, with top trades, such as elevator installers, earning over US$109,000 annually and first-line supervisors averaging US$86,450.

These are the kind of careers that anchor middle-class households and fund union pension plans for decades.

“Little public benefit” is not a policy finding.

After construction ends, data centres generate permanent, well-paid technical and operations roles, plus substantial property tax revenues and downstream economic activity.

It appears to be a conclusion arrived at before the evidence was examined.

There is a deeper irony here that Canadians should find uncomfortable.

The modern AI revolution, the deep learning breakthroughs powering every large language model, image generator, and AI assistant in use today, was built substantially on Canadian soil.

Geoffrey Hinton did foundational work on neural networks at the University of Toronto. Yoshua Bengio founded what became Mila, now the world’s largest academic research centre for deep learning, at the Université de Montréal in 1993. Together, Hinton and Bengio shared the 2018 ACM Turing Award (computing’s Nobel Prize) for their contributions to the field.

The infrastructure that makes AI possible is now being built everywhere but here.

The United States is constructing data centre capacity at a historic pace. The European Union is racing to keep up.

And Manitoba, blessed with some of the most effective natural cooling conditions on the continent, just said no.

Canada helped build the intellectual foundations of the AI age. It would be a strange legacy to also be remembered as the country that refused to house it.

David Clement is the policy director at the Consumer Choice Center. He writes from the Toronto area.