Rural Canada is not asking to be rescued by this moment. It is asking to be included in it: partners, not hosts.
In June 2026, Canada launched a national artificial intelligence strategy, “AI for All,” with a headline target of sixty percent of Canadian businesses adopting AI by 2034. The strategy names rural communities and commits real money. What it does not yet do is make sure that money, and the means to use it, reach rural ground. This paper closes that gap using designs, rules, and a training program that already exist. Nothing in it asks anyone to invent.
The situation, briefly
The AI buildout is landing on rural ground because that is where the power is. An analysis done for this paper of Canada’s fifty largest generating stations puts roughly three-quarters of their capacity outside metropolitan Canada, so the new data centres are siting in rural and small communities. In Alberta alone, requests to connect data centres to the grid have passed 20,000 megawatts against a peak provincial demand of about 12,000.
But rural Canada starts from behind on the supports its peers take for granted. The European Union has reserved a share of its rural development fund for community-led local delivery for four decades; the United States runs a permanent rural development agency with an office in every state. Canada built the same kind of tool — a federal Rural Lens to check policy for rural impact — then stopped using it in 2013. The OECD’s 2024 review put the result plainly: “the current federal innovation policy has an inherent urban bias.”
That is the backdrop when a community is asked to host a data centre without the means to weigh what it is offered. In four months, two Prairie projects showed what follows: a $12-billion data centre at Sherwood, Saskatchewan drew protests and a First Nation’s duty-to-consult objection, and Manitoba rejected a hyperscale project at Ritchot after a petition of more than thirteen thousand signatures. Rural communities are not refusing the technology. They are refusing deals they were never equipped to evaluate.
The turn: domain experts can now build
For one class of technology, the digital divide has quietly inverted. Building software — until recently the most urban-concentrated work in the economy — can now be done by the person who understands the problem, working with an AI assistant, from wherever they live. The scarce input is no longer the engineering. It is knowing, in operational detail, which problem is worth solving. That input rural Canada has in abundance: the field worker who knows what data the site generates, the farmer who knows why the last software pilot failed at harvest, the operator who runs water treatment for eight hundred people.
The paper rests on a first-hand build: over roughly five weeks of part-time work, on a fresh consumer computer, a non-developer customized and extended a large open-source system for a heavily regulated field — with AI as a working partner under a disciplined method. The discipline is the point, and it is teachable.
What works: the solutions
Every fix below already exists somewhere in Canadian or peer practice. None requires a new institution or new money beyond what is committed.
- Route the money so it reaches rural. Send the regional AI money through the community-governed delivery design Canada has run since 1985 — the Community Futures model — so rural delivery is built in rather than left to chance. Federal
- Fill the empty seat. No rural delivery practitioner sits on either federal AI advisory body today. Add one, before the buildout’s terms are set rather than after. Federal & provincial
- Fund training in rural communities, by rural educators. Carry an AI-capability curriculum through the strategy’s own named channel of libraries and community hubs. A worked curriculum already exists. Federal & provincial
- Restore the measurement. Publish rural breakdowns of AI adoption so we can see where rural communities stand — the OECD’s own first recommendation to Canada. Federal
- Protect what the buildout leaves behind. Extend the decommissioning and asset-residue rules already applied to wind and solar so a data centre’s durable infrastructure passes to community use if the operator leaves. Provincial
- Make community-benefit agreements the default. More than five hundred binding agreements are already in force in resource development. Apply the instrument to data centres, with local hiring, training, and supply commitments written in as binding terms. Industry
The training already exists
The Program: AI-Assisted Building for Rural Communities is a ready-to-run companion lesson plan, not a proposal to write one. It is ten weekly sessions of about two and a half hours, delivered by a local educator through a library, adult-learning centre, or community hub, to a cohort of six to ten adults who have a real problem to solve and no technical background. It teaches the discipline that makes AI-assisted building safe, built around six plain-English safety rules. A learner finishes with one working, verified piece of their own project, the safety rules as working habits, and a written plan for what comes next.
The ask, in one line
Rural Canada is not asking to be rescued by this moment. It is asking to be included in it. The precedents exist, the capability can be taught, and the window is open now.
A featured paper by Gordon More, editor of Rural Innovations and Executive Director of Southeast Techhub in Estevan, Saskatchewan. The views and editorial choices are his own. Questions, corrections, or to request the underlying sources: gord@ruralinnovations.org.