Demand is no longer the central question. The winners will be the projects and jurisdictions that can align power, approvals, procurement, cooling, and community expectations early enough to deliver.
The tone at Data Centre West 2026 in Calgary this September was noticeably different from a year ago. In 2025, much of the discussion focused on whether Alberta could capture the surge in data centre investment and find enough power to support it. In 2026, the demand question felt largely settled. The harder question was how to execute projects at the speed and scale now being contemplated.
That shift matters well beyond Alberta. Across Canada and the United States, AI and cloud demand are colliding with the practical limits of electrical grids, equipment supply chains, approval processes and community tolerance. For owners and developers, a credible data centre strategy can no longer begin with the system basis of design and site selection alone. It must include a detailed plan for energy, community engagement, infrastructure, technology, construction delivery and future phase changes
Demand is strong but deliverable capacity is scarce
North American market data confirms what conference participants are seeing on the ground. CBRE reported that capacity under construction in the eight primary North American markets reached a record 7,481 MW in the first half of 2026. More than 80% was already preleased, while vacancy fell to 1.4%. In other words, substantial capital is moving into the sector, but capacity remains extremely tight.
The power implications are equally significant. The U.S. Department of Energy estimates that data centres accounted for about 4.4% of U.S. electricity consumption in 2023 and could account for 6.7% to 12% by 2028. In Ontario, the IESO’s 2026 reference outlook projects annual electricity demand from commercial data centres rising from 4.5 TWh in 2027 to 22.6 TWh by 2050. Even the low-demand scenario anticipates unprecedented growth.
These figures should not be interpreted as proof that every announced project will proceed. Connection queues include projects at very different levels of maturity. What they do show is that power availability has become a defining constraint, and that the ability to turn a power position into reliable, operating capacity will separate credible projects from speculative ones.
Power availability has become a defining constraint
Power strategy is now part of the core design
At Data Centre West, the discussion was not simply about obtaining a grid connection. It was about assembling an energy solution from several possible elements: grid supply, on-site generation, battery energy storage, renewables, microgrids and flexible or conditional connections. Alberta’s system operator has already allocated the full 1,200 MW available under the first phase of its interim large-load process and is developing the longer-term framework for additional projects.
Energy strategy, utility engagement and facility design must advance together.
This changes project sequencing. Energy strategy, utility engagement and facility design must advance together from the earliest stages. A project team needs to understand not only how many megawatts are required at full build-out, but also the ramp profile, redundancy expectations, power quality, load behaviour and the consequences of curtailment or delayed grid energization.
Behind-the-meter generation may accelerate delivery in some markets, but it does not remove the need for a full energy development plan. Fuel supply, emissions, permitting, operating responsibility, maintenance, black-start capability, and the role of the grid all have to be resolved. The best answer will vary by region and by workload. A training campus with enormous, flexible loads may warrant a different solution from a latency-sensitive inference facility or an enterprise colocation site.
Regulatory certainty is a competitive advantage
Developers are not only asking for fewer rules. They are asking for clear rules, predictable decisions and coordinated reviews. Where the province or state, system operator, utility, regulator, and municipality send conflicting signals, schedules become difficult to finance, and capital can move elsewhere.
Alberta illustrates both the opportunity and the complexity. The province’s data centre levy came into effect January 1, 2026, at up to 2% on computing equipment, with lower rates where projects rely less on the public grid. At the same time, AESO has introduced new connection requirements for transmission-connected data centres and is continuing to develop its longer-term framework for large loads and BYOG (Bring Your Own Generation) projects. For developers, the key issue is having enough clarity to understand the impact on cost and schedule.
Ontario faces a different system and market structure, while provinces with hydro-dominant grids offer another value proposition. This is why Canada should not be discussed as a single data centre market. Power availability, carbon intensity, connection processes, water constraints, tax structures and approval pathways differ materially by province and often by municipality.
Equipment availability is shaping electrical architecture
Transformers, switchgear, generators, and other major electrical components remain under pressure, with utilities and developers often pursuing the same equipment. Procurement can no longer wait for design completion. It is consistently the first topic of conversation in a confirmed project.
Conference participants described teams reconsidering traditional electrical architectures because an alternative voltage class or equipment configuration could be procured materially sooner. That approach can be sensible, but only if the project evaluates system protection, maintainability, spare-parts strategy, commissioning, operator familiarity and lifecycle risk - not just delivery time.
Owners can reduce exposure by identifying long-lead equipment early, prequalifying manufacturers, standardizing repeatable blocks where appropriate and aligning commercial commitments with realistic design maturity. The objective is not to freeze every decision prematurely. It is to preserve flexibility without leaving the critical path dependent on equipment that cannot be delivered when needed.
Start with the IT roadmap, not the cooling plant
Rapid changes in chip technology and rack density make mechanical planning challenging. A facility designed around yesterday’s air-cooled loads is poorly suited to today’s liquid-cooled GPU clusters. Conversely, oversizing a first phase for an uncertain future load can strand capacity and capital.
A more resilient approach starts with the owner’s best available IT roadmap: expected rack densities, deployment timing, cooling-water temperatures, redundancy requirements and the likely balance between training, inference and conventional compute. The mechanical and electrical systems can then be planned as modular infrastructure with defined pathways for expansion or conversion.
There is a real opportunity for Canadian markets. Canada’s cool climate can increase the hours available for free cooling, while more than 80% of Canadian electricity comes from non-emitting sources. Climate alone does not guarantee an efficient facility, Water strategy, heat-rejection temperatures, part-load performance, and controls integration need to be addressed together to drive efficiency. In suitable locations, waste-heat recovery can also become part of the project’s relationship with the surrounding community.
Community acceptance is now a project requirement
Loudoun County, Virginia offers a useful warning for emerging markets. As of March 2026, the county reported approximately 233 existing data centre buildings representing 56.5 million square feet. Those facilities generated $1.2 billion in real and personal property tax revenue in fiscal 2026, equal to 39% of the county’s overall budget. Yet the scale and pace of development have also intensified concerns about land use, transmission infrastructure, water, noise, air quality and community character.
Loudoun has since moved away from broad by-right development and now requires special-exception approval for future proposals. The lesson is straightforward: economic value does not eliminate local impacts, and technical compliance does not guarantee public trust.
Community considerations should therefore influence site selection and design from the beginning. Poor separation from residential areas is difficult to correct later. Low-frequency noise from mechanical and electrical equipment can create complaints even where conventional limits are technically met. New substations and transmission corridors may be more visible to neighbours than the data centre itself. Cooling choices affect water demand, plume, noise, and heat rejection. These are design issues, approval issues and communications issues at the same time.
Transparency matters. Communities need a plain-language explanation of what is being proposed, what supporting infrastructure is required, how impacts will be mitigated and what benefits will remain locally. Those benefits may include assessment revenue, construction and operations employment, supplier opportunities, workforce training and investment in enabling infrastructure. Claims should be specific and credible; overstating job creation or understating resource use will erode trust quickly.
Technical compliance does not guarantee public trust.
Canada has an opportunity, but it is not automatic
Canada brings genuine advantages: cooler environmental conditions, large clean-electricity resources, political stability, strong engineering and construction capability and proximity and connectivity to the world’s major markets. Alberta adds abundant natural gas, available land and an electricity market that can support creative power arrangements. Ontario offers scale, connectivity and access to a large technology and financial ecosystem. Quebec and British Columbia can offer low-carbon hydroelectric power, subject to availability and provincial priorities.
The opportunity is also broader than hyperscale campuses. CBRE expects growing demand for smaller 25-to-75-MW facilities serving inference, enterprise AI and latency-sensitive uses, as well as urban edge sites in the 2-to-20-MW range. That creates potential for a more distributed Canadian market, including conversions and expansions where power, fiber, and community fit are stronger than at a greenfield site.
But no jurisdiction can rely on climate, land, or energy resources alone. Projects will favour places that can offer a credible path to power, coordinated approvals, skilled labour, supply-chain access and public legitimacy. The same test applies in the United States, where transmission build-out, zoning delays and local opposition increasingly constrain otherwise attractive sites.
What successful execution looks like
The next phase of data centre growth will reward disciplined integration. For owners and developers, seven actions stand out:
- Treat energy as a project workstream from day one. Define the load ramp, resilience model and realistic grid and on-site power options before committing to a site or schedule.
- Align the IT roadmap with facility infrastructure. Use credible workload and density scenarios to guide cooling, electrical distribution, and phasing.
- Bring procurement into concept design. Identify long-lead equipment, evaluate alternate architectures, and establish decision dates before the schedule is exposed.
- Design for modular change. Create repeatable capacity blocks and clear conversion paths without paying on day one for every possible future scenario.
- Integrate mechanical and electrical decisions. Power density, liquid cooling, water use, controls and heat rejection must be optimized as one system.
- Design for community fit. Address siting, noise, water, emissions, and visible infrastructure before they become approval or reputation risks.
- Plan for operations and commissioning early. Complex power and cooling systems only create value when operators can test, maintain and run them safely under real load conditions.
The message from Data Centre West was ultimately optimistic. Demand is real, and Canada has credible reasons to compete for a meaningful share of North America’s next generation of digital infrastructure. The constraint is execution. Projects that coordinate power, design, procurement, approvals, and community outcomes early will be far more likely to move from announcement to operation - and to deliver durable value once they get there.
Have Questions? Contact:
Ash Parmar
Senior Engineering Consultant, Science + Technology
E ash.parmar@hhangus.com
Craig Sievenpiper, P.Eng., M.B.A., LEED AP, CCP
Vice President, Science + Technology | Principal
Connect with Craig