Sask. contractors begin work on Bell Canada's Regina data centre

Bell Canada has started early construction on its $1.7 billion AI data centre south of Regina, with almost all contractors sourced from Saskatchewan and work already underway on drilling tests, earthworks, and piling. The project is expected to create about 800 local construction jobs and the first phase is slated to come online in the first part of 2027. The announcement is constructive for Bell’s AI infrastructure buildout but is unlikely to have near-term market impact.

Analysis

This is less a single-project headline than a signal that Canada’s AI buildout is shifting from announcement phase to local execution, which tends to benefit regional contractors first and equipment suppliers second. The immediate winners are civil works, piling, concrete, and energy-monitoring specialists; the less obvious beneficiary is the local labor market, where a multi-quarter construction ramp can tighten skilled trades availability and lift wage pressure across adjacent infrastructure projects. That matters because margin expansion for the prime contractors may be capped by labor scarcity even as revenue visibility improves.

The second-order effect is on power and permitting ecosystems. A large data-center project typically pulls forward grid interconnect work, transformer lead times, and compliance spend, which can create bottlenecks for other industrial projects in the same province. If Bell’s timeline holds into 2026–27, the broader trade is not just “AI capex” but utilities, electrical equipment, and environmental monitoring providers that can monetize follow-on work without taking full project risk.

The contrarian read is that local-content headlines often overstate near-term economic impact while underestimating execution risk. The real gate is not initial dirt work; it is power availability, cooling design, and geological/groundwater constraints, any of which can slow the project by months and shift spending out the curve. In that sense, the market may be too focused on construction employment and not enough on the embedded demand for grid hardware and specialty engineering, which is where the durable upside likely sits.

For public-market positioning, this favors a basket approach over a single-name bet: the upside is real, but project timing risk is high and front-loaded headlines are usually not enough to sustain alpha. A better expression is to buy companies exposed to North American data-center power and infrastructure cycles rather than contractors with one-off regional exposure.

Key Decisions for Investors

• Go long EME on a 6-12 month horizon: data-center power distribution and electrical scope should scale with AI buildouts; use a 10-15% trailing stop because project timing can slip, but upside could be 20-30% if AI infrastructure spend broadens.
• Initiate a basket long of HUBB / ETN / VRT over the next 1-2 quarters: these names capture the higher-probability spend on transformers, electrical gear, and data-center power systems; risk/reward is better than betting on any single regional contractor.
• Avoid chasing local civil-construction proxies; if using optionality, buy small upside calls on infrastructure beneficiaries with 12-18 month expiries rather than common equity, since early-phase construction headlines often fade before revenue is recognized.
• Pair trade: long VRT vs. short a broad industrial ETF over 3-6 months; if AI capex continues to re-rate power equipment, the spread can expand meaningfully while limiting macro beta.