Sovereignty is a word with a specific meaning. It describes the right to self-determination — to make decisions about your own territory, your own governance, your own future. It is a word that has been used in Canada in particular in the context of First Nations land rights, where it carries enormous weight and real legal consequence.
When the federal government uses the word "sovereign" to describe AI data centres owned by Telus and powered by Nvidia GPUs, it is borrowing the prestige of that word to describe something much more limited: data that doesn't cross the American border.
This is not nothing. Data that lives in Canada is subject to Canadian law, which provides more protection than American law in some respects. But it is also not sovereignty in any meaningful political sense. The compute is controlled by a private corporation. The models trained on it are proprietary. The decisions made by the systems that run on it are not subject to democratic oversight.
Who benefits from "sovereign" framing?
The announcement is structured around economic benefits — $9 billion in projected economic activity, 1,000 construction jobs. These are real numbers, and they matter to real people.
But the framing elides the question of who governs the output. When Canadian universities use this compute to train AI models, who owns those models? When Canadian healthcare systems run AI diagnostics on this infrastructure, who audits the algorithms? When Canadian government agencies use AI-assisted decision-making, who can challenge those decisions?
These are not technical questions. They are political questions, and they are not being asked.
What actual sovereign compute would look like
Genuinely sovereign AI infrastructure would be governed by a public institution with democratic accountability. It would be subject to access-to-information requests. Its board would include representation from the communities it serves — not just from industry and academia.
It would prioritise access for public good applications: health research, environmental monitoring, educational tools, small business services. Commercial AI development would be allowed, but it would not be the primary purpose.
The data processed on it would be subject to clear governance rules about retention, access, and deletion. There would be an independent audit function with real authority.
This is not a fantasy. Other countries have built public computing infrastructure on roughly these lines. Canada has the resources and the institutions to do it. What it lacks, right now, is the political will — and the public pressure — to demand it.
What you can do
Read the open letter on this platform. Sign it. Share it with people who aren't already in the tech conversation — educators, healthcare workers, community organisers. The decisions that are being made right now will shape Canadian AI infrastructure for decades. They should be made with the public, not announced to it.