Last week I wrote about Canada's Sovereign Cloud Initiative in the context of Bill C-22 — the surveillance legislation that makes the case for owning your infrastructure from the other direction. The argument there was defensive: the government is building a framework that could compel data collection and retention, and communities that own their stack are in a structurally different position than communities that rent.

Today the other half. The affirmative half. The federal government is not only building surveillance powers — it is spending $2.36 billion in 2026-27 on sovereign digital infrastructure. Shared Services Canada's 2026-27 Departmental Plan, published April 13, names "Digital Sovereignty" as one of three key priorities alongside Transformation and Modernization.

This is not a white paper. It is a budget with deliverables.

What the Government Is Building

The plan is specific. Here is what SSC is actually funding this fiscal year:

Sovereign AI Platform. Foundational infrastructure for AI and high-performance compute that meets federal data residency and security requirements. Protected B workloads, hosted on Canadian cloud providers, within Canadian jurisdiction. SSC is supporting the National Research Council's AI GPU cluster upgrade, building edge AI compute capacity at Canadian research complexes, and working with the Department of National Defence on sovereign AI hosting. CANChat — the government's in-house generative AI alternative to commercial tools — is transitioning from pilot to enterprise service on a Canadian cloud provider. The goal: AI compute where the data, the model, and the intellectual property never leave Canada.

GC Cloud One. A standardized, secure cloud environment that lets departments build and deploy applications on pre-approved platforms. SSC is explicitly "enhancing cloud-based services while supporting data residency and sovereignty through sovereign cloud hosting" and developing backup capabilities aligned with Canadian data residency requirements. They are collaborating with Canadian-based cloud service providers to "bolster the nation's digital infrastructure."

Private Cloud. Dedicated cloud infrastructure for the federal government — a sovereign private cloud within Canadian jurisdiction. Compute, storage, networking, with a roadmap to scale. Strategic procurements for hardware. Early adopters validating the service.

Protected B Secure Messaging. Expansion of a Protected B secure messaging and collaboration solution to approximately 50,000 government-issued devices. The plan states this explicitly: "This initiative will strengthen Canada's digital sovereignty, increase diversification of services, enhance the security of government communications, and ensure continuity of operations in the event of service disruptions. In addition, it will reduce reliance on foreign-hosted cloud services for sensitive communications."

Buying Canadian. Beginning in 2026-27, SSC will apply a "buying Canadian" approach to contracts valued at $25 million or more. This follows the August 2025 RFI for sovereign public cloud capability that restricted applicants to entities fully owned and controlled by Canadian persons — excluding US firms. OpenText and TELUS launched a joint Canadian Sovereign Cloud platform in July 2025. ThinkOn is marketing itself as "true sovereign" — staffed by Canadian citizens, insulated from foreign law.

The total planned spending for SSC in 2026-27: $2,361,524,776. That is not a policy aspiration. That is a line item.

The Government's Definition of Sovereign

Read the plan carefully and you notice what "sovereign" means in this context. It means Canadian-owned providers. OpenText. TELUS. ThinkOn. Canadian companies operating Canadian data centers under Canadian law.

This is a real improvement over the previous model, where the shortlist for a 25-year federal cloud contract was composed solely of US-based tech giants. The CLOUD Act compels disclosure based on who controls the data, not where it is stored. If your cloud provider is a US company, the data is reachable by US law enforcement regardless of where the servers sit. Moving to Canadian-owned providers closes that vector. That matters.

But it is procurement sovereignty, not ownership sovereignty. The government is buying sovereign infrastructure from Canadian corporations. It is not building it, operating it, or controlling it directly. The OpenText/TELUS platform is a commercial product. ThinkOn is a commercial provider. The government is a customer — a large, powerful customer with stringent requirements, but a customer.

Why That Distinction Matters for Communities

Here is where the work we do diverges from what the federal plan is doing.

When SSC procures a sovereign cloud from OpenText, the architecture is: Canadian corporation owns the infrastructure, Canadian law applies, US law cannot compel disclosure. Good. When a fire hall deploys a Proxmox cluster in its own building running Nextcloud and Matrix, the architecture is: the fire hall owns the infrastructure, Canadian law applies, US law cannot compel disclosure, and no corporation sits between the community and its data.

The government's model removes the foreign jurisdictional vector. The community-owned model removes the corporate vector too.

This is not an argument that the government should run its own infrastructure instead of procuring. The federal government operates at a scale where procurement is the only viable model — you do not build a sovereign AI platform for 300,000 public servants by hand. The procurement approach is correct for SSC.

The argument is that communities are not the federal government. A fire hall serving 2,000 residents. A town office serving 8,000. A community center with 300 members. At that scale, the calculus is different. You can own the stack. A Proxmox node, a Nextcloud instance, a Continuwuity Matrix server, a Keycloak identity provider — that is infrastructure one person can deploy and a volunteer can operate. You do not need to procure it from anyone. You do not need a vendor relationship. You do not need a contract.

The government is spending $2.36 billion to achieve through procurement what a community can achieve through ownership. The end state — data under Canadian jurisdiction, no foreign corporate parent, local accountability — is the same. The path is different, and for small organizations, the ownership path is cheaper, more transparent, and more durable.

The Trade Skepticism

The CCIANet analysis from September 2025 raised concerns that Canada's sovereign cloud initiative could function as digital protectionism — excluding foreign providers in ways that raise costs, reduce competition, and fragment the market. The report noted that Canada's WTO Government Procurement Agreement commitments cover data processing services, and that favoring Canadian-owned providers introduces local preference that may contravene those commitments.

There is a legitimate version of this concern and a self-serving one. The self-serving version comes from US industry groups whose business model depends on global cloud monopoly. The legitimate version is this: if sovereign procurement means Canadian SMEs are forced to use smaller, less competitive domestic providers, they pay more for less. That is a real cost, and it falls on the organizations least able to absorb it.

But this concern assumes the only options are "foreign hyperscaler" or "Canadian corporate provider." Community-owned infrastructure is a third option that the trade debate completely ignores. When a town office runs its own Nextcloud, it is not procuring from anyone. It is not affected by vendor discrimination. It is not paying a premium for domestic cloud. It is building the thing itself, at a hardware cost it controls.

The trade argument is between two procurement models. The ownership model is outside that argument entirely. That is worth saying clearly: the framing of "sovereign cloud" as a choice between US providers and Canadian providers misses the option where the community is the provider.

What This Validates

The federal government treating digital sovereignty as a core priority — not in a strategy document but in a departmental plan with a budget — does something useful for the work we do. It validates the argument.

When we sit in a town council meeting and argue that a community should own its infrastructure rather than rent it from Microsoft, the counter-argument is often: "If sovereignty mattered, the government would be doing it." Now the government is doing it. The same federal government that was happy to run on AWS and Azure for a decade has decided that sovereignty is worth $2.36 billion and a trade fight with its largest trading partner.

The government arrived at this conclusion through a national security lens — CLOUD Act exposure, foreign jurisdictional reach, the risk of a 25-year contract with US firms during a period of geopolitical instability. Communities arrive at the same conclusion through a different lens: autonomy, cost control, auditability, and the simple fact that a community which owns its infrastructure can shape it to its own needs.

The destination is the same. The federal plan is useful evidence that the destination is worth reaching. What the plan does not do — and what no government plan will do — is tell communities that the ownership path is available to them. That is the part we say.

The Practical Takeaway

For Canadian communities watching this unfold:

  1. The sovereignty argument is now government-endorsed. When you make the case for community-owned infrastructure, you can point to the federal government's own $2.36 billion investment in sovereign infrastructure as evidence that the principle is sound. This is useful in a room where "sovereignty" still sounds abstract.

  2. Government sovereign and community sovereign are different things. The government is procuring sovereign infrastructure from Canadian corporations. Communities can own it directly. Both reduce foreign jurisdictional reach. Only ownership gives you structural control over architecture, retention, and auditability.

  3. The trade debate is not your debate. The argument about whether Canada's sovereign cloud procurement violates WTO commitments is an argument between vendors. Community-owned infrastructure is not procurement. It is not affected by trade rules. It is the option that exists outside the frame.

  4. The tools are the same ones the government is building analogues for. SSC is building CANChat because it does not want government data flowing through commercial AI tools. A community running local LLMs on its own hardware is doing the same thing at a different scale. The principle — your data, your compute, your jurisdiction — scales down.

The federal government found $2 billion reasons that sovereignty matters. For communities, the reason is simpler and the cost is lower: you can own the stack. The government is buying sovereignty. You can build it.