Alberta's preference for operators that supply their own power [13] does more work than its levy, because it converts a real-estate decision into a generation decision: a developer arriving with its own gas turbines is buying a permitting posture along with the parcel, and the province takes the load without routing the cost through a regulated tariff. That is the objective Ottawa wrote as the second of its five pillars, that electricity costs not fall to Canadians [4], and then declined to enforce [5].
The arithmetic on Alberta's ambition is worth doing slowly. Mordor Intelligence puts the entire Canadian data centre market at nearly $14 billion USD, about $19 billion CAD [1], while Alberta alone is chasing $100 billion of data centre investment by 2030 [10], roughly 5.3 times the national figure [17] if the provincial target is denominated in Canadian dollars, which the source does not specify. The two numbers also measure different things, one a market size and the other a cumulative capital goal, which is why an investment target makes a better announcement than a forecast. Take it at face value anyway and a two-percent levy on data centre value [12] implies something on the order of $2 billion accruing to the province [18], before property taxes and natural gas royalties. Mordor's more-than-doubling by 2031 [2] is about 12 percent a year compounded off a 2025 base [19], which is a growth market a utility can plan around rather than a stampede.
Ottawa is the latest order of government to publish a formal framework, not the first [22], and Alberta began drafting its strategy in 2024 [9]. The federal document's only leverage is its signatory list, 23 names including OpenAI, Anthropic, Microsoft, Meta, Amazon, Cohere, Bell and Telus [6], which matters mainly because grid reliability, energy costs and environmental impact have already produced strong public opposition in many jurisdictions [16]. Two readings compete here, or rather, two and a half. Provinces could copy the five pillars into binding legislation, at which point the patchwork narrows and the soft law was the real law all along. Or a project inside a targeted-policy province stalls at the municipal, utility, Indigenous consultation or court stage [15], and the concierge service [11] turns out to sell a speed it does not control.
What this material does not contain is a single electricity price, interconnection date or water term by jurisdiction, so the location bet can be ranked by posture but not priced. The thesis that provincial policy is the binding constraint fails if interconnection queues, which sit with utility operators rather than legislatures [15], are what actually decides where the load lands, in which case a two-percent levy is a rounding error against a multi-year wait for capacity. On the evidence available, the enforceable numbers in Canadian data centre policy are provincial ones, and the federal pillars are a drafting aid for the governments that write them [5]. Alberta has already collected the first proof point, having landed Canada's largest data centre development deal this year with Anthropic reportedly looking at the province and dozens of projects in the pipeline [14].
