Three Power Shifts Shaping Canada’s Data Centre Market And The Always-present Threat Of Technological Change

The Canadian data centre space is entering a new phase as governments at all levels look to promote, and sometimes curtail, the new and existing projects popping up to meet the ever-increasing demand for digital infrastructure. Power is at the centre of this conversation, with access to electricity and its cost shaping how and where projects can move forward.

As we head into the final part of 2026, here are the three developments we’ll be watching unfold across the country.

1. Access to power becoming competitive and conditional 

Across several major Canadian markets, access to power grids has become a key project risk, with the provincial governments actively working through new regulation for prioritizing, allocating, and approving electricity connections for large data centre projects. 

In the remainder of 2026, three provinces are expected to make strides in this area: 

• British Columbia: Amid substantial growth in electricity demand, BC introduced a competitive allocation process aimed at directing clean electricity to projects that generate the “greatest value for people and communities in BC.” BC Hydro expects to notify applicants to the 2026 Call for Demand for Emerging Industries (AI/Data centres) of the outcome by mid-September. 

• Ontario: A consultation open until September 12 proposes a framework under which new large data centres could require government approval before connecting or reconnecting to the electricity grid. Connection decisions could consider strategic and economic value alongside reliability and cost responsibility. Expect to learn more this fall. 

• Alberta: The province enacted the Data Centre Regulation in June, establishing a new framework for connecting large data centres to the grid and prioritizing projects that bring their own generation. AESO is now developing its longer-term Phase II framework, including a BYOG connection process covering project readiness, grid capacity, and MW allocation. 

2. The price of power is changing 

As the provinces grapple with the scale of electricity demand from data centres, some are also reconsidering how these large users should pay for the power they consume and the costs they place on the grid.

A proposed electricity rate class for large consumers in Ontario 

Through the Ontario consultations mentioned earlier, the Economic and Strategic Assessment Framework for New Data Centres, the province is also explicitly asking what operational and financial impact a separate electricity rate class for new data centres would have. The example threshold for the size of a data centre project that would be impacted by this rate is currently set at greater than 1 MW.

Under this proposal, data centres located in Ontario as part of this new class would not be eligible for the Industrial Conservation Initiative, through which qualifying large consumers can currently reduce Global Adjustment costs by lowering consumption during peak periods.

An imminent decision on Québec’s proposed electricity rate 

Hydro-Québec has applied for approval of a new dedicated Rate CD for data centres with authorized demand of at least 5 MW. The proposal would introduce a specific rate structure for these facilities, including minimum billing requirements and additional charges where demand during a project’s ramp-up exceeds agreed levels.

The Régie de l’énergie has confirmed an early October hearing. The initial proposal includes a November 1 implementation date, but that is subject to the Régie de l’énergie’s decision.

3. Déjà vu all over again: The threat of SLMs 

In addition to regulatory and power supply variables, developers and investors must also weigh the risks associated with broader technological uncertainty in the marketplace.

As recently discussed in the Financial Times, the growing viability of Small Language Models (SLMs) as a decentralized alternative to Large Language Models (LLMs), is calling into question the investment thesis of many data centre developers. 

Some believe that this dynamic could parallel the shift observed in the early 1980s, when “good-enough” home computing began to displace centralized mainframe infrastructure for many user applications. While early home computers lacked the raw processing capabilities of mainframes, they were "good enough" for everyday tasks and provided tertiary benefits that larger, more expensive, and clunky mainframe systems could not offer. 

If the market shifts toward SLMs—an approach that some observers see aligning with Apple's strategic AI choices—the anticipated demand for massive, centralized data centre infrastructure may fizzle out. As Mark Cuban recently noted, this potential decentralization of computing power carries the significant risk that some of today’s large-scale data centre projects could become obsolete creating, as he framed it, wonderful new spaces for "pickleball courts": the other burning issue that keeps municipal councilors up at night.

Accordingly, when modelling long-term project economics, it is prudent to factor in the potentiality of technological dispersion and its impact on future capacity requirements: in other words, we may end up using good enough AI solutions reliant on our phones and laptops and only turning to power hungry, centralized data centre tools only for the most sophisticated and complex problems.

What to consider as these changes take shape 

Taken together, these developments mean access to power is becoming both harder to get and, potentially, more costly. Developers and project owners should consider early how evolving connection rules and pricing models could affect project timelines, regulatory approvals, project economics, and contractual commitments. 

Electricity costs, minimum demand requirements, and other utility obligations should therefore be understood and built into project modelling before key development and investment decisions are made.