Canada’s next AI battle may not be fought over chatbots, apps, or even talent. It may come down to where the country’s data lives, who controls the servers, and whether Canadian companies can build future-defining technology without depending almost entirely on foreign-owned infrastructure.
A new report says Amazon, Microsoft, and Google now control 85% of Canada’s public cloud market, a striking figure arriving just as Ottawa prepares a national AI strategy built partly around sovereignty. The issue is not simply whether U.S. cloud giants offer powerful tools. They clearly do. The deeper question is whether Canada can remain competitive in AI while relying on a small group of foreign hyperscalers for the computing backbone behind government systems, business software, research labs, and fast-growing startups.
Canada’s Cloud Market Is More Concentrated Than the Global Average
The 85% figure is the number that turns a technical debate into a national economic story. According to the report cited by The Canadian Press, Amazon holds 42% of Canada’s public cloud market, Microsoft holds 31%, and Google holds 12%. Together, the three U.S. companies dominate the infrastructure that stores data, runs applications, and supports the computing workloads behind everything from online banking tools to artificial intelligence systems.
That concentration is higher than the global average for the same three companies, which the report puts at roughly two-thirds of the cloud market. The difference matters because Canada is not just buying storage space. It is buying access to a digital operating layer that increasingly determines how fast governments modernize, how safely companies handle sensitive information, and how easily startups can scale. Cloud infrastructure has become less like office software and more like national plumbing: mostly invisible until control, cost, or access becomes a problem.
Why AI Makes Cloud Dependency More Urgent
Artificial intelligence has turned cloud infrastructure from a back-office IT issue into a front-line competitiveness issue. Training, testing, and deploying advanced AI systems requires enormous computing power, especially specialized hardware and high-performance data centres. For many Canadian companies, particularly small and mid-sized firms, buying that infrastructure directly is unrealistic. The cloud becomes the place where ambition meets affordability.
Ottawa already recognizes this pressure. The federal government’s Canadian Sovereign AI Compute Strategy is backed by $2 billion over five years and is designed to give Canadian researchers, businesses, and innovators better access to compute capacity. Its three main parts include mobilizing private-sector investment, building public supercomputing infrastructure, and creating an AI Compute Access Fund. The timing is critical: if Canada wants homegrown AI companies to stay and scale, compute cannot remain a luxury only the best-funded firms can access.
Sovereignty Does Not Mean Cutting Off U.S. Technology
The sovereignty debate can easily be misunderstood. It does not mean Canada suddenly stops using Amazon Web Services, Microsoft Azure, or Google Cloud. These firms offer global scale, strong security tools, advanced AI services, and reliability that few smaller providers can match. For many organizations, abandoning them would be expensive, risky, and impractical. A serious sovereignty plan has to start from that reality.
Instead, the issue is control and choice. The Government of Canada’s own digital sovereignty framework says sovereignty means the ability to exercise autonomy over digital infrastructure, data, and intellectual property, while acknowledging that complete digital autonomy is impossible in a connected world. In practical terms, Canada needs enough domestic capacity, clear contract rules, strong encryption, and vendor-neutral systems so it is not locked into one narrow path. Sovereignty is less about isolation and more about having credible options when geopolitical, legal, or commercial risks change.
Ottawa Is Already Deeply Entangled With U.S. Cloud Providers
The federal government is not watching this issue from the sidelines. Newly released documents reported by The Canadian Press showed Ottawa had spent almost $1.3 billion since 2021 on cloud services from U.S. companies, with more than $1 billion going to Microsoft. The same reporting said Amazon Web Services, Microsoft, and Google services were being used across government, including for applications described as mission-critical by National Defence.
That spending reflects a broader federal “cloud-first” posture. Government guidance has long directed departments to consider cloud services as a principal delivery option for new IT investments, with public cloud often prioritized before hybrid, private, or non-cloud options. This approach can make sense when government systems need faster upgrades and better scalability. But it also means the same institutions discussing sovereignty are already dependent on foreign-controlled platforms for important digital operations. The policy challenge is not theoretical. It is already sitting inside federal procurement.
Data Stored in Canada Is Not Always Fully Canadian-Controlled
One of the most common assumptions in the cloud debate is that data stored in Canada is automatically under Canadian control. Federal guidance is more cautious. The Government of Canada’s digital sovereignty framework says using a Canadian supplier or storing data in Canada does not guarantee that data will be outside the jurisdiction of foreign courts. The reason is simple: companies can be subject to laws in countries where they operate or are headquartered.
This is where the U.S. CLOUD Act enters the conversation. The law can allow U.S. authorities, under legal process, to seek data held by American companies even when that data is stored abroad. That does not mean every Canadian file sitting in a U.S.-owned cloud is being accessed by foreign authorities. It does mean data residency and data sovereignty are not the same thing. For sensitive government, defence, health, research, and business information, the question becomes who can access the system, under what law, and under whose control.
The Hardest Problem May Be Switching Costs
Cloud giants are dominant partly because they are very good at what they do. They offer instant scale, global networks, mature security services, advanced databases, machine-learning platforms, and developer ecosystems that took decades and billions of dollars to build. A startup can go from a prototype to a product serving thousands of users without buying servers or signing a data-centre lease. A government department can modernize faster than it could by building everything alone.
But that convenience can create lock-in. Once an organization builds around one provider’s tools, databases, identity systems, and AI services, switching becomes expensive and technically difficult. The Canadian Anti-Monopoly Project argues Ottawa could use procurement rules to require interoperability and substitutability, making it easier for buyers to move between providers over time. That sounds dry, but it is a major lever. If public contracts require portable systems, Canada can keep using global cloud services while reducing the risk of becoming trapped inside them.
Canadian Startups Need Compute, Not Just Capital
Canada has world-class AI talent, but talent alone is no longer enough. The Dais at Toronto Metropolitan University has warned that Canada’s AI compute gap could threaten the country’s innovation advantage. In plain terms, researchers and startups can have brilliant ideas but still struggle if they cannot afford the computing power needed to train models, test products, or serve customers at scale.
The business adoption data shows why this matters. Statistics Canada found that 12.2% of Canadian firms used AI to produce goods or deliver services in 2025, double the share from the previous year, while another 14.5% planned to adopt AI within the next 12 months. That is still early-stage adoption, but the direction is clear. More companies will need compute. If most of that capacity is rented from foreign hyperscalers, Canada may build AI users without building enough AI owners.
Data Centres Are Also Energy Projects
AI sovereignty is not only a technology file. It is also an electricity, land, cooling, and infrastructure file. The International Energy Agency estimates that data centres consumed about 415 terawatt-hours of electricity globally in 2024, equal to roughly 1.5% of global electricity use. Its base case projects that global data-centre electricity consumption could double to about 945 terawatt-hours by 2030, driven heavily by AI and accelerated servers.
Canada has advantages in this race, including cool climates, available land in some regions, and relatively clean electricity in several provinces. But the grid challenge is real. The Canadian Climate Institute has noted that modern AI facilities can exceed 100 megawatts of demand, far above many traditional data centres. That kind of load can strain local grids, require new transmission, and raise hard questions about who pays for new infrastructure. Sovereign AI cannot be planned separately from power policy.
A Canadian Cloud Push Could Still Become Another Oligopoly
There is a political temptation to answer foreign dependency with a simple Canadian substitute. But that approach carries risks. The Canadian Anti-Monopoly Project has warned that simply directing public funding toward domestic telecom incumbents, without competition rules and interoperability conditions, could recreate the same structural problem under a Canadian label. In other words, replacing foreign concentration with domestic concentration would not necessarily give businesses more choice.
This is where Ottawa’s plan needs discipline. A serious sovereignty strategy should support Canadian-controlled infrastructure, but it should also encourage open standards, transparent pricing, portable data, and real competition. Otherwise, Canadian startups and public agencies could trade one form of dependency for another. The goal should not be a “maplewashed” version of the same locked-in model. It should be a market where Canadian providers can grow, global providers can still compete, and customers can leave when performance, cost, or control no longer works.
The Real Test Is Whether Ottawa Uses Its Buying Power
Governments often shape markets less through speeches than through procurement. Ottawa is a major cloud buyer, and that gives it leverage. If federal contracts reward interoperability, Canadian data control, supplier diversification, transparent subcontracting, and credible exit plans, vendors will adapt. If contracts continue to prioritize convenience and speed above long-term control, the market will keep moving toward the biggest incumbents.
The government’s Spring Economic Update framed AI sovereignty as one of the pillars of a broader “AI for All” strategy, promising sovereign compute infrastructure that is resilient, sustainable, and under Canadian governance. The words are ambitious. The implementation will be harder. Canada has to balance security with innovation, domestic control with access to world-class tools, and competition policy with urgent AI adoption. The 85% cloud-market figure is not just a statistic. It is a warning that the infrastructure behind Canada’s AI future is already concentrated, and the window to build real alternatives is narrowing.
