“Because controlling our energy means controlling our destiny.”
Mark Carney, Forward Guidance: A Stronger Canada, 12:19-12:25.
Energy and compute are becoming one strategic question.
Every organisation operating at scale depends on electricity. Increasingly, it also depends on computation: for research, manufacturing, health, public services, small business, logistics, and the systems that coordinate them. AI intensifies that relationship. It makes compute more valuable, but it also makes the underlying choices about power, networks, cooling, security, ownership, and continuity more consequential.
Canada’s response must be more ambitious than importing capacity and more disciplined than treating every proposed installation as progress.
Infrastructure must explain itself
A large number expressed in megawatts is not a public case. Communities and decision-makers need to know what the number describes: a utility connection, total facility load, or IT equipment load. They need clear accounts of expected annual energy use, peak demand, grid upgrades, demand-response capability, backup generation, and who carries the cost when public infrastructure must expand.
Water and noise also belong in the public record. Cooling is not a technical footnote when it affects local systems. A credible installation explains its cooling approach, water use where relevant, seasonal constraints, acoustic controls, generator testing, fuel storage, emissions, and emergency procedures before it asks for public trust.
The same standard applies to data governance. A server located in Canada does not alone establish Canadian control. The material questions are more specific: who operates the service, which laws may apply to the provider, where data travels, who holds encryption keys, who receives privileged access, how incidents are handled, and whether an organisation can leave without losing an essential service.
Right-sized capacity is a national capability
Canada should build compute capacity that matches real work.
Some workloads need proximity because continuity, privacy, response time, or local operation is essential. Others benefit from regional pooling or trusted global services. A resilient country does not force every workload into one model. It builds the capacity to classify work, choose proportionate controls, and avoid dependence on a single provider, grid node, or corporate decision for systems people cannot do without.
That is the case for right-sized distributed infrastructure. It is not a case against cloud services, large facilities, or international exchange. It is a case for meaningful choice: local where local control matters, regional where shared operations are stronger, national where common standards and public-interest capacity are required, and international where trusted exchange adds real value.
The federal government’s September 15, 2026 announcement of a Productivity Mega Deduction signals renewed attention to capital investment, including technology assets. That announcement does not make every project suitable for every place. It raises the standard for governance. Public capital, planning approval, and community trust should support infrastructure that is operable, repairable, accountable, and capable of delivering enduring public value.
The test
Before a digital installation asks for public trust, it should publish its commitments on power, water, noise, backup, cyber security, data governance, workforce, community benefit, and performance reporting.
Before an organisation commits a critical workflow to AI, it should identify the workload, operator, service dependency, data pathway, energy context, fallback mode, and public-value test.
By 2080, Canada will have the capacity to govern critical systems in the public interest. The work starts now with infrastructure that communities can inspect and institutions can explain.