Do we need data centres?
Yes. Banking, health records, public services, cloud software, telecommunications, streaming and AI all depend on secure computing infrastructure.
Infrastructure, Intelligence & Impact
AI in Canada
A public explainer for moving the data-centre debate past backlash, boosterism and viral half-truths.
Some criticism is justified. Some comes from treating every data centre, every kind of computing and every proposed development as though they are identical.
The stronger position
The public is right to demand answers about power, water, cost and community benefit. But we are using the term data centre as though it describes one standardized building. It does not.
The useful middle ground is not defending every server rack or rejecting every project. It is separating essential digital infrastructure from speculative or poorly planned development.
The missing distinction
The debate is really four questions.
Coverage often compresses different infrastructure, energy, sovereignty and community-benefit issues into one overloaded term.
Do we need every proposed data centre?
No. A facility does not become strategically valuable simply because the proposal carries an AI label.
Are all data centres equally damaging?
No. Impact changes with size, purpose, energy source, cooling system, location and operating model.
Have communities sometimes been given bad deals?
Absolutely. Weak disclosure, padded job claims, public subsidies and loose oversight have helped create the backlash.
What they are
Industrial infrastructure for digital work.
A data centre is a secure industrial facility where digital information is stored, processed and moved.
- Secure industrial facilities where digital information is stored, processed and moved.
- Part warehouse, part factory and part utility infrastructure.
- Places where different workloads have different public value and different resource demands.
What they are not
One standardized building with one standardized impact.
Not all AI facilities
Not all giant campuses
Not automatically clean
Not automatically dirty
Not necessarily employment engines
Not inherently Canadian infrastructure
Where concern is informed
These are governance questions, not anti-technology reflexes.
Residents are not necessarily rejecting the digital future. They may be rejecting a process that asks them to absorb costs without clearly explaining benefits.
- Potable water used for cooling
- Electricity capacity committed without public discussion
- Grid upgrade costs shifted to ratepayers
- Diesel backup or new gas generation
- Agricultural or sensitive land converted
- Few permanent jobs relative to resources consumed
- Tax incentives negotiated behind closed doors
- Approvals completed without meaningful consultation
Where it gets blurry
Replace viral claims with specific questions.
"AI is stealing our water"
How much potable water will this specific facility consume during normal and peak conditions?
"We can simply say no to data centres"
Where will Canada's computing happen instead, under which laws, and powered by what grid?
"They consume power but produce nothing"
Does the public value of this facility justify the resources and infrastructure it requires?
"Every AI query consumes a bottle of water"
Which model, hardware, cooling system, electricity source, location and measurement boundary are being counted?
"A moratorium solves the problem"
What standards would turn a pause into a real infrastructure strategy?
The four-part test
The question is not whether a building contains servers. It is what those servers are doing, what resources they require and what Canada receives in return.
Purpose
What computing will it perform, and for whom?
Resources
How much electricity and water will it use at full capacity?
Responsibility
Who pays for the grid, water and supporting infrastructure?
Return
What lasting value does the community and country receive?
Inside the Data Centre
A series with room to educate, challenge and clarify.
The format gives the public a way to understand infrastructure choices before each local proposal becomes a yes-or-no shouting match.
- Episode 1What a data centre actually does
- Episode 2Why AI needs so much computing
- Episode 3Does every AI request use water?
- Episode 4Where do the promised jobs go?
- Episode 5Who pays for the additional electricity?
- Episode 6Could Canada build better data centres?
- Episode 7Data sovereignty or corporate subsidy?
- Episode 8How should communities judge a proposal?