Canada’s new AI for All strategy is a serious step forward. Its foreign-policy dimension recognizes an important reality: Canada does not have to choose between pursuing digital sovereignty alone and depending entirely on the American technology ecosystem. It can work with democratic partners, help shape trusted digital standards and give Canadian firms a stronger role in the markets and institutions that will govern AI.

One crucial question remains underdeveloped: how Canada will deploy AI. Competition is not decided only by who builds foundation models or controls semiconductor supply chains. It is also decided in the “deployment layer”: the software, services, technical support and institutional relationships that turn a general-purpose model into something a hospital, manufacturer, public department or small business can actually use.

That layer is where AI becomes embedded in everyday work. A model may be built in one country and hosted in another, but the firm that integrates it into records systems, trains staff, manages compliance and becomes the long-term service provider can still acquire enormous influence. Canada needs a strategy for that part of the AI economy.

The Open-Source Deployment Phase

In this emerging phase, “open” options are increasingly important for firms and states without an established foothold in AI to compete with proprietary, “closed” alternatives from companies such as OpenAI. This strategy has given their models and infrastructure a global head start at the deployment layer. 

Today, open-source solutions such as Alibaba’s Qwen from China challenge the dominance of proprietary US models. Their strategy is similar to Google’s approach with Android against Apple’s smartphone dominance: secure a foothold in the market with an open foundation, then capture value through the productivity, services, and integration layers built on top of it.

Many digital economies, caught in a state of underdevelopment relative to the US, increasingly turn to open-source solutions to circumvent the costs, regulatory constraints, and infrastructure bottlenecks associated with US solutions. For instance, Singapore has localised an open version of Qwen for domestic deployment across the education, healthcare, and finance sectors. Whether open or closed, deployment-phase strategies rely on three common “capture” mechanisms: (1) institution capture, in which deployment gives the solution provider influence over the adopter’s organisational standards, tools, and procurement pathways; (2) enterprise capture, in which institutions become costly to move once their workflows and compliance systems are built around a particular solution; and (3) value capture, in which open systems create paid dependencies through hosting, support, and integration at scale. Open solutions are particularly adept at incentivising initial adoption while capturing value over the long term.

At........

© OpenCanada