Telus and Canada Announce B.C. Sovereign AI Cluster

Canada's Minister of Artificial Intelligence and Digital Innovation, Evan Solomon, and TELUS publicly announced plans for a three-site AI data centre cluster in British Columbia on May 11, 2026, per The Canadian Press and CTV News. The network comprises an expansion of TELUS' existing Kamloops data centre plus two new Vancouver facilities in Mount Pleasant and downtown Vancouver, with the Kamloops expansion and Mount Pleasant facility scheduled to open later in 2026 and the downtown site expected in 2029, according to BNN Bloomberg and BC Business.

According to TELUS's May 11 media release, the proposed cluster is planned to scale to over 60,000 GPUs and 150 MW of power by 2032. TELUS's release and BC Business state the facilities will deploy AI hardware from NVIDIA, use liquid cooling and heat recovery systems, and are designed to run on approximately 98% clean energy while recycling waste heat to district systems. TELUS's release also includes an economic estimate of about $9 billion in value for British Columbia and workforce figures tied to construction and operations. The Canadian Press reports the federal government has earmarked $2 billion over five years, starting in 2024_25, under a program to support large-scale sovereign data centres.

Industry-pattern observations: Large-scale AI clusters centered on high GPU counts and high-density cooling are now standard for supporting training and inference of modern foundation models. Facilities that combine liquid cooling with heat recovery are increasingly common because they lower PUE (power usage effectiveness) and create opportunities to reuse thermal energy in urban heating networks. Public reporting that links onshore compute capacity to data sovereignty mirrors similar policy debates in other jurisdictions where governments subsidize local infrastructure to keep sensitive workloads and IP on domestic soil.

Editorial analysis: For practitioners, an onshore cluster of the scale TELUS describes would expand domestic options for training and hosting large models without relying solely on foreign cloud providers. Public reporting frames the federal funding and TELUS capacity targets as part of a broader push to anchor AI compute in Canada. That said, announced capacity targets and timelines frequently change between press release and full commissioning; observers should treat 2032-scale figures as planned rather than operational facts until facility-level commissioning is complete.

Editorial analysis: Observers should follow three measurable indicators:

• •site-level commissioning notices and power contracts for the downtown Vancouver facility
• •vendor disclosures about specific GPU models and cluster networking (for example, whether NVIDIA H100 or later generations are deployed)
• •regulatory or municipal approvals related to district heat integration and water use

Additionally, public procurement or research partnerships that commit model training or government workloads to these sites will clarify how much of the announced capacity is reserved for domestic sovereign workloads versus commercial customers.

The Canadian Press reports Evan Solomon acknowledged financial risks at the announcement, saying, "Of course, there are financial risks when people are investing billions of dollars" and that Canada "cannot be risk-averse in this moment." TELUS's May 11 media release frames the expansion as a response to demand following its Rimouski Sovereign AI Factory, which TELUS states has sold out.

No independent cost breakdown or power purchase agreements were provided in the public releases cited. Several high-stakes metrics (GPU counts, economic impact, energy mix) are presented by TELUS and reported by outlets; those figures are attributable to TELUS's media release or The Canadian Press coverage and have not been independently verified in reporting cited here.