Canada is putting more weight behind open-source AI, with Mila, Mozilla, and Hypertec teaming up on an initiative designed to make advanced AI systems easier for Canadian organizations to run locally and keep under their own control.

Announced September 17 at ALL IN, Canada’s largest AI and technology event, the initiative will develop an open-source AI foundation layer for businesses, governments, nonprofits, and other institutions. The project is backed by the Canadian government, with Mila leading technical delivery and coordination while Mozilla contributes engineering and open-source expertise.

The financial commitment includes an initial $5 million investment from Mozilla and another $1 million from Hypertec for the first year. Hypertec will use its hardware to help accelerate deployments of the technology among Canadian organizations.

The broader idea is straightforward: make running your own AI system nearly as easy as using a commercial cloud AI service, without requiring organizations to hand their data and AI workloads over to an external provider.

Building an open foundation for local AI

Open-source AI models are increasingly available, but getting one from a downloadable model into a secure, reliable production system can be considerably more complicated.

Mila and Mozilla want to address that gap by creating a foundation layer that organizations can install on their own infrastructure. The system is intended to support locally operated models, data, governance, and access controls while remaining flexible enough for individual components to be replaced.

The project will have two major components: an open standard built around interface contracts and a reference implementation that organizations can install on their own hardware or infrastructure.

That approach is intended to avoid locking organizations into a particular AI model, vendor, or hardware configuration. Instead, the foundation should provide a common structure that different open-source components can plug into.

For a small business, for example, the system could operate as a private internal AI assistant capable of searching company documents, answering technical questions, drafting reports, or helping with software development. Because the system can run locally, proprietary company information can remain within the organization’s own environment.

The same infrastructure could potentially be adapted for hospitals, nonprofits, manufacturers, and government organizations with different privacy, security, and governance requirements.

Hypertec brings the hardware

Mila and Mozilla have already spent the past six months working on the architecture, selecting open-source components and testing the system end to end.

Hypertec’s involvement is intended to take that work beyond research and reference implementations and toward actual deployments. The Canadian technology company will help organizations deploy the open-source AI foundation layer on Hypertec hardware, giving businesses and institutions a more practical route to running AI privately.

The Canadian government is also supporting the initiative as part of its broader push toward domestic AI capabilities. Government officials said open-source AI can give Canadian businesses and institutions greater control over their technology and data while making AI systems easier to adapt and potentially less expensive to operate.

The partnership also builds on Mila and Mozilla’s strategic relationship announced in March 2026, which focused on advancing open-source and sovereign AI capabilities. That earlier partnership included work around trustworthy AI and private memory architectures for AI agents.

Open-source AI still has a deployment problem

The organizations are targeting a very specific weakness in today’s open AI ecosystem.

Mozilla’s State of Open Source AI report found that 79% of developers adding AI functionality use open models, but only 53% of teams reach production. The organization attributes the gap to challenges including cost, security, integration, and ongoing maintenance.

In other words, getting access to an open model isn’t necessarily the difficult part. Building everything around that model so an organization can safely depend on it every day is where much of the work begins.

Mila and Mozilla’s project is therefore less about creating another AI model and more about building infrastructure around the models that already exist.

The organizations expect to publish working reference implementations for enterprise, government, and public-interest use cases within six months. Their longer-term goal is to make open-source AI substantially easier to deploy so that organizations can operate AI systems themselves without having to assemble the entire stack from scratch.

For Canada, that also ties into a broader effort to develop AI infrastructure and expertise domestically rather than relying exclusively on technology controlled by companies or institutions elsewhere.

The project remains under development, but its direction is clear: rather than treating AI as something organizations simply rent from a cloud provider, Mila, Mozilla, and Hypertec want to make locally owned and operated AI a practical option.