This empty field south of Regina is the location of a Bell Canada AI data centre set for construction this spring. (Alexander Quon/CBC)
The Rural Municipality of Sherwood in Saskatchewan approved development of a 500-megawatt AI data center requiring millions of gallons of water daily for cooling despite significant local opposition over environmental impact and resource consumption, CBC reported March 16, as AI infrastructure expansion reaches smaller jurisdictions offering tax incentives and looser environmental regulations than major tech hubs.
The facility, proposed by an undisclosed data center operator, will consume enough electricity to power approximately 40,000 homes and require water withdrawals equivalent to a small city's daily usage. Local officials approved the project citing economic development benefits including construction jobs, ongoing employment, and property tax revenue, overriding concerns from residents and environmental groups about strain on municipal water supplies and electrical grid capacity.
Small Communities Compete for AI Infrastructure Investment
The Saskatchewan approval illustrates how AI's massive infrastructure requirements are pushing data center development beyond traditional tech corridors into smaller communities willing to offer incentives and expedited permitting that larger jurisdictions increasingly resist. Rural municipalities facing economic challenges view data centers as rare opportunities to attract major capital investment and create jobs in regions with limited alternative economic development prospects.
The RM of Sherwood approved significant property tax abatements, subsidized utility connections, and streamlined environmental reviews to secure the project, competing against other Canadian and US jurisdictions pursuing the same limited pool of data center operators seeking sites for AI compute infrastructure. This dynamic gives companies leverage to negotiate favorable terms by threatening to locate facilities elsewhere if demands aren't met.
The trend raises concerns about whether small communities possess technical expertise and regulatory capacity to evaluate complex AI infrastructure proposals, negotiate balanced agreements protecting local interests, or enforce environmental and operational commitments once facilities are operational. Unlike established tech hubs with experienced planning staff and established data center regulatory frameworks, rural municipalities may lack resources to conduct thorough due diligence or ongoing oversight.
Water and Energy Consumption Spark Environmental Concerns
Opposition to the Saskatchewan facility centers on water consumption for cooling the massive GPU clusters required for AI training and inference workloads. Modern AI data centers use water-intensive cooling systems to dissipate heat from thousands of high-performance processors running continuously, with facilities consuming millions of gallons daily—volumes that can strain municipal water infrastructure and compete with agricultural, residential, and industrial users during drought conditions.
Environmental groups questioned whether the RM adequately assessed water availability given Saskatchewan's periodic drought cycles and existing agricultural water demands. The province has experienced severe water shortages in recent years, raising concerns about whether adding major industrial water consumers creates unsustainable resource allocation conflicts when climate change increases drought frequency and intensity.
Energy demands also pose challenges for electrical grids not designed for sudden additions of 500-megawatt loads. The facility will require dedicated substation infrastructure and potentially new generation capacity to serve AI workloads without affecting reliability for existing customers. Critics argue that using substantial electrical capacity for AI training rather than decarbonization, manufacturing, or residential needs misallocates limited energy resources.
Economic Benefits Versus Long-Term Community Impact
Sherwood officials defended the approval by emphasizing economic benefits including approximately 50 permanent jobs, construction employment, and projected annual property tax revenue exceeding $2 million once abatements expire. For small rural municipalities with limited commercial tax bases, this revenue represents meaningful funding for infrastructure, services, and community investment that wouldn't otherwise materialize.
However, critics note that data centers provide fewer jobs per dollar of capital investment than manufacturing or other industries, employ primarily technical specialists rather than local workers, and create minimal economic multiplier effects beyond direct tax payments since facilities operate with minimal supply chain integration into local economies. The jobs versus environmental impact tradeoff may favor communities differently than officials project.
National Pattern of AI Infrastructure Decentralization
The Saskatchewan approval follows similar data center controversies in smaller US communities including rural Georgia, Iowa, and Oklahoma where AI infrastructure proposals sparked debates over resource consumption, environmental impact, and whether economic benefits justify costs. As major tech hubs like Northern Virginia, Oregon, and Ireland face growing resistance to new data center development, operators increasingly target jurisdictions desperate for investment and willing to accommodate industry demands.
This geographic dispersion of AI infrastructure also raises questions about whether decentralized development fragments regulatory oversight, creates regulatory arbitrage opportunities where companies seek locations with weakest environmental standards, and makes coordinating energy and water resource planning more difficult when critical infrastructure scatters across multiple jurisdictions with varying capacity and priorities.
