Telus plans AI data centre expansion in B.C., including two new centres in Vancouver

Telus will develop two new Vancouver data centres and expand its Kamloops facility for AI workloads, bringing total power demand to more than 150 MW by 2032. The largest site, a 400,000-square-foot downtown Vancouver facility, is slated to reach 100 MW and the company says the three sites will house more than 60,000 Nvidia GPUs once fully operational. The move supports Telus’s push into AI infrastructure and aligns with Ottawa’s new AI infrastructure program, though the company has not disclosed capex for the projects.

Analysis

This is less a telecom capex story than a state-backed compute-supply story. If Telus can secure federal support and off-take visibility, it effectively de-risks a long-duration asset build and turns idle balance-sheet capacity into a quasi-utility platform with embedded AI demand. The second-order winner is Nvidia’s ecosystem and power/cooling/infrastructure vendors, but the more important signal is that Canadian carriers are trying to re-rate from low-growth operators into scarce digital infrastructure owners.

The competitive wrinkle is that Bell’s tenant-led model likely scales faster and with less technology execution risk, while Telus is taking more operating and technology-stack exposure. That creates a spread trade opportunity: Bell’s approach should have higher near-term capital efficiency, but Telus has more upside if it captures margin on GPU services rather than just rent. The market may be underestimating how much this shifts telecom valuation multiples from EBITDA compression optics to asset-duration and policy-option value.

Key risks are not demand but execution and power. 150 MW by 2032 is meaningful only if interconnects, grid approvals, and procurement are secured on time; any slippage pushes cash conversion out by years, which matters in a higher-rate environment. There is also policy risk: if federal support arrives with local-content or Indigenous participation conditions that add cost or delay, the headline growth narrative can quickly become a low-return regulated utility project.

The contrarian view is that the AI infrastructure premium may already be partially priced into telecoms, while the real upside is in the beneficiaries of capacity leasing and GPU deployment, not the carrier owners. If government support is real, the winners could include power infrastructure, cooling, and select semiconductor exposure more than BCE/Telus themselves. The move is therefore bullish but likely over-owned at the equity level unless the next catalyst is a signed financing/off-take package rather than another conceptual expansion plan.

Key Decisions for Investors

• Go long BCE vs. short T in a 3-6 month pair trade: Bell’s tenant-leasing model should convert to cash flow faster and with lower execution risk; target 8-12% relative outperformance if permits/funding remain slow.
• Buy 6-12 month calls on regulated power/grid beneficiaries exposed to Canadian data-center buildouts; the bottleneck is increasingly MW delivery, not GPU demand, so power-linked names have cleaner upside than telecom equity.
• Sell near-dated covered calls on T after strength: the stock can re-rate on headline AI news, but absent a signed federal support package, follow-through is likely capped over the next 1-2 quarters.
• For higher-conviction AI infrastructure exposure, express the theme via CRWV on pullbacks rather than BCE/T: if capacity scarcity persists, operators that can monetize external GPU demand should capture more economic rent.
• Set a catalyst watch for Ottawa funding/off-take announcements over the next 1-3 months; if nothing materializes, fade the AI-infrastructure premium in Canadian telecoms and rotate to balance-sheet repair trades.