Bell Canada’s new artificial intelligence data centre near Regina, Saskatchewan, is touted as the largest private-sector capital investment in Canadian history. The facility requires up to 1.2 gigawatts of electricity once fully operational. To manage immediate environmental impact, the company employs a sealed, closed-loop water-cooling system paired with air-cooled heat rejection. Bell Canada states this infrastructure is not connected to local groundwater or municipal supplies, aiming to avoid competing for local resources.

However, this onsite strategy does not capture the full environmental cost. The data centre draws 300 megawatts from the provincial grid operated by SaskPower, while 900 megawatts come from a dedicated onsite scheme. According to Scroll, which published the underlying analysis, the water used by the electricity grid to generate that 300-megawatt share is excluded from the facility’s direct water accounting. This distinction is critical because, as the report notes, most electricity generation remains highly water-intensive.

SaskPower’s portfolio includes hydro, wind, solar, coal, and natural gas. Despite strong renewable resources, most capacity comes from fossil fuels. Approximately 25% of Saskatchewan’s power generation capacity comes from coal-fired plants. These plants use a combination of dry and wet cooling technologies. If the grid allocates power to the data centre, cooling water is still required at these distant power plants. Consequently, water demand shifts from the data centre site to the power generation facilities, potentially increasing total water withdrawal from the ecosystem far from Regina.

The analysis also highlights the complexities of using natural gas as a bridge power source. If Saskatchewan follows Alberta’s path and imports natural gas to power data centres, water is required to drill, extract, process, and deliver the fuel. Saskatchewan has relatively little natural gas production, meaning much of it will likely be imported from Alberta gas fields. This adds another layer of water consumption that is not visible in the data centre’s operational metrics.

Regional water management faces additional challenges due to seasonal swings in precipitation and temperatures. The Assiniboine River basin, where the data centre is located, experiences variable water entry from spring snowmelt and wetland filling. Authorities must consider the cumulative effects of increased water demand from industry, power generation, agriculture, and municipalities. Careful management is required to settle the discrepancy between industrial water needs and the finite water resources available during prolonged droughts.