Telus Builds AI Muscle: 60,000 GPUs Signal Deep Commitment to Canadian Compute Power

The foundational vision driving Telus’s latest infrastructure play is clear: secure a robust, domestic compute backbone to power the escalating demands of artificial intelligence. By strategically expanding its data centre footprint across Vancouver and Kamloops, Telus is moving past simply being a connectivity provider; it is positioning itself as a core enabler of Canada's AI economic ambition. The sheer scale of this undertaking—housing over 60,000 Nvidia GPUs—is the defining feature. This isn't merely adding capacity; it represents the deliberate aggregation of specialized computational power. The architecture involves a phased rollout: activating an existing facility in Kamloops this year, expanding a repurposed site in Vancouver’s Mount Pleasant neighbourhood, and finally launching its largest component by 2029—a 400,000-square-feet centre with a capacity up to 100 MW. This staggered deployment strategy allows Telus to manage massive capital expenditure while meeting immediate, growing AI workloads. The necessity for this infrastructure is driven by global trends and domestic policy aims. As AI adoption moves from proof-of-concept to commercial reality across sectors like healthcare and finance, the computational horsepower needed to train and run sophisticated models becomes paramount. Furthermore, the strategic importance of localized compute capacity cannot be overstated. By investing heavily in these facilities, Telus contributes directly to lessening dependency on foreign tech giants, a concern echoed by federal initiatives from Innovation, Science and Economic Development Canada. When comparing this move to its peer, Bell Canada (BCE Inc.), the difference in strategy is instructive. While BCE is opting for a leasing model—building space but allowing tenants like CoreWeave and Cerebras to bring their own specialized GPU clusters—Telus appears focused on building out comprehensive internal capacity. This suggests Telus aims not just to host AI workloads, but potentially to run them or offer deeply integrated services built around its compute core. The strategic significance of these multi-megawatt facilities demands significant planning and resource allocation, including an anticipated requirement of over 150 megawatts by 2032. This level of power commitment is a major indicator of long-term corporate confidence in the sector's growth trajectory.