Soil Is Not Sovereignty

Commentary on “Another day, another data centre announcement. What do they mean for Canada and its digital sovereignty?”  ·  Financial Post

This series has spent most of its life arguing that sovereignty is decided at the infrastructure layer, in the data centre, the grid connection, the physical apparatus of compute. A Financial Post explainer, surveying the Canadian landscape, contains the sharpest available challenge to that view, and it is worth taking seriously rather than around.

The scene it describes is by now familiar. Independent counts put roughly 200 to 300 data-centre projects across Canada; a York University study identified 194 active facilities, 14 under construction, and 199 announced as of June. The active ones cluster in Ontario, Quebec, and British Columbia, while Alberta, with a tenth of the facilities, holds more than 90 per cent of planned capacity. Bell's Regina expansion, quadrupled to 1.2 gigawatts and $52 billion, anchors the headlines. And the backlash is arriving on schedule: a one-year moratorium in Oakville, a pending pause in Mississauga, a rejected project in Manitoba, and, even in Alberta, a utilities commission refusing a gas plant meant to power a data centre.

Then the explainer makes its real point, through the people who study this. The ownership of these facilities is predominantly foreign, and it becomes more so the larger the project gets: identified US-headquartered firms account for around half of small active facilities and about 85 per cent of large announced ones. And ownership of the building, the experts argue, is not even the deepest problem.

The strongest argument against this series' own framing

Vass Bednar of the Canadian Shield Institute puts it bluntly: having "the physical foundation in your backyard does nothing" for the ability to govern the digital economy. A data centre on Canadian soil can still process data controlled by someone else, under someone else's law. The US CLOUD Act lets American authorities compel data from US-based providers, Google, Amazon, Microsoft, Apple, regardless of where the servers physically sit. So a Canadian-soil facility, running a US hyperscaler's services, is subject to a foreign legal reach that the concrete does nothing to block.

That is correct, and this series should say so plainly. It is, in fact, the point made here before under a different headline: ownership is not a security control. Physical presence is not data control. A flag on the building governs nothing.

But "focus upstream, the building is a distraction" is the opposite error

The reading that tempts people from Bednar's true premise is that infrastructure is therefore beside the point, and that sovereignty is really an "upstream" matter of data, IP, and governance frameworks. Take that too far and it fails for the mirror-image reason.

Governance without infrastructure is a rulebook with nowhere to run. You cannot exercise jurisdiction over data that lives, at the moment it matters, on compute you do not operate, in a facility you do not control, connected to a grid you did not build. Data-residency rules, procurement standards, and IP frameworks are all necessary, and all of them ultimately have to execute somewhere. If that somewhere is a hyperscaler's cloud, the governance is a request, not a control. The building is not sufficient for sovereignty. It is still necessary for it.

Canada is currently getting the worst split of the two

Put the two halves together and the Canadian position comes into focus, and it is not comfortable. The country is building the physical layer, taking on the energy demand, the water questions, the community fights, the moratoriums, while conceding the control layer, with 85 per cent of the largest announced projects foreign-owned and the data inside them reachable under foreign law.

That is the worst of both worlds: the local costs of the buildout without the sovereignty benefit that was supposed to justify it. It is exactly the outcome you get when "sovereignty" is satisfied by pointing at a building on Canadian soil and asking no further questions. Bednar is right that the backlash is fed by unanswered ones, who these are for, what they are for, how they make the country stronger.

And the control layer is now on a trade table

There is a timing risk the explainer names that the infrastructure debate usually ignores. Sovereignty is not only something you build or fail to build. It is something you can negotiate away. Canada stopped collecting its Digital Services Tax last year to advance trade talks with the United States, and analysts quoted in the piece warn that ceding further data, tech, and AI sovereignty in those negotiations would leave the country badly exposed. You can pour every slab domestically and still surrender the governance layer at a negotiating table. The concrete does not protect what the treaty gives away.

The Canadian read

So the correction this series owes its own argument is a both-and, not a retreat. Sovereignty requires the physical layer and the control layer, aligned: infrastructure located in Canada, owned or genuinely controlled by parties answerable to Canadian law, running data governed by that law, built by firms the country can actually keep rather than watch get absorbed. Any one of those alone is a partial claim that the others can hollow out. A domestically owned building running foreign-controlled services is not sovereign. Perfect governance rules with no domestic compute to run on are not sovereign either. And all of it is negotiable if no one is guarding it in the room where trade deals are made.

The federal Responsible Data Centre Development Principles gesture at parts of this, but as one expert notes, they arrive without regulatory teeth, and decisions remain "fundamentally local." Principles without instruments are the governance version of soil without control: the right words, no grip.

What this means going forward

The useful thing the explainer does is refuse the easy version of the sovereignty story in either direction. It is not enough to build the data centres, and it is not enough to write the governance rules. The devil, as one researcher puts it, is in the details, and the details are that sovereignty is a stack: soil, ownership, law, and the will to keep all three.

Canada is currently strong on the first, weak on the second, unsettled on the third, and negotiating on the fourth. A country can fix that, but not by celebrating another announcement. Soil is where sovereignty starts. It is not where it is decided.

VOLTEDGE

Reference: “Another day, another data centre announcement. What do they mean for Canada and its digital sovereignty?” · Financial Post · read the article