As Ontario moves to guide data centre development, debate about the environmental and operational impacts of these facilities has intensified. A University of Waterloo computer science professor says the key difference between traditional data centres and the new generation built to support artificial intelligence is not the building itself but the computers inside it.
Computers and workloads define a data centre
Martin Karsten, a professor at the University of Waterloo, told CBC News that conventional data centres are optimised for a class of tasks focused on storing, processing and retrieving information in predictable, deterministic ways. In contrast, the infrastructure that supports large-scale machine learning and generative AI relies on a different type of processor and a different pattern of computation.
"For these computers, you kind of want to distinguish between more traditional computing work, which has a lot to do with storing information, processing information, retrieving information in a deterministic type [of way],"
Karsten said AI workloads tend to have simpler control flow but vastly heavier computational demands. That shift — from complex control flows to massively parallel numerical computation — drives choices about the processors used and therefore affects how a data centre is designed and operated.
What that means for operations, energy and planning
Because AI workloads rely on specialised processors and sustained, heavy computation, the resulting facilities can have different energy and cooling demands than traditional installations. That technical reality underlies much of the public concern and the policy response seen across Canada, including Ontario’s recent data centre playbook.
Those concerns are not purely hypothetical. A recent study cited by the CBC found that while Canada currently hosts only five hyperscale data centres, there are 96 facilities either proposed or under construction nationwide. These figures underscore the pace of expansion and the need for municipal and provincial planning.
| Metric | Number |
|---|---|
| Hyperscale data centres in Canada | 5 |
| Facilities proposed or under construction | 96 |
Local context: Waterloo’s tech ecosystem and public debate
Waterloo sits at the intersection of academic computing research and a dense cluster of technology companies. That combination means local residents and governments must weigh the benefits of investment and jobs against concerns about energy use, water consumption and land use. The technical distinctions Karsten describes help explain why AI-oriented installations may prompt different regulatory responses than traditional data halls.
Municipal decision-makers and community groups are watching provincial guidance closely. Ontario’s playbook seeks to provide a framework for siting and operating data centres while addressing environmental issues raised by constituents and municipalities. The practical differences in processor types and workloads, as explained by Karsten, are central to setting the right rules and expectations.
Questions for residents and planners
Understanding the technical differences can help residents, planners and councillors ask sharper questions of developers: What kind of processors will the facility house? What are the projected energy and cooling profiles? How will peak and baseline power requirements be managed? Answers to those questions affect everything from local grid planning to environmental assessments.
- Processors matter: AI data centres use different hardware than traditional centres, leading to different operational profiles.
- Growth is rapid: National figures show a small number of hyperscale sites today and many more proposed or under construction.
- Policy follows tech: Provincial playbooks and municipal approvals must reflect technical realities to protect community interests.
As Ontario refines its approach, the distinctions emphasised by Waterloo researchers will likely inform how local governments evaluate proposals and balance economic opportunities against environmental and infrastructure impacts. For residents, a more technical conversation means more targeted, evidence-based public engagement rather than broad opposition or support.
