Canada has decided to build uncrewed systems at home. The Defence Industrial Strategy names them a sovereign capability. A new Defence Investment Agency is being stood up to direct work toward Canadian firms, and the newly launched Defence Drone Initiative is standing up a marketplace pathway from concept through operational testing toward production. The DND/CAF Drone Surge contest has moved 14 Canadian attritable-drone designs into prototype development, and Canada is already co-producing a Ukrainian design domestically through the Airlogix-Sentinel arrangement.
Every one of those moves is good. And after two decades of carrying complex hardware from concept into production, I can tell you what happens next if nothing changes: a winning design emerges in 2027, the applause fades, and the company faces the question every Canadian hardware firm eventually faces. Who can actually build this, in quantity, to spec, and how do we know? Today the honest answer is one factory at a time, learned painfully, with the knowledge locked in whoever did it first.
Why Canada funds prototypes but not production
Look at the funding landscape side by side and a pattern appears, and it is worth naming with some affection, because it is recognizably Canadian. We fund research superbly. We run good prototype contests. We help individual firms build capability through grants and capacity programs. Each instrument is sound on its own terms.
Taken together, they carry a structural blind spot: every dollar strengthens a firm or a design, and no dollar strengthens the connections between firms that turn a collection of capable companies into a production system. Scale is being pursued firm by firm. But a hundred somewhat-more-capable firms still cannot build one thing at rate together, because that ability lives in the seams: shared standards, transferable production packages, common quality gates, connected data, a qualification method. The seams are precisely what no individual firm can fund for itself and no current program owns. The fragmentation compounds across stages too: each program hands its output to no defined next step, and the handoffs between research, prototype, and production are where industrialization quietly dies.
This matters more than it used to, because the strategic environment changed. Modern conflict consumes attritable systems at rates that devour peacetime supply arrangements. A country that cannot surge production of a system does not, in any operationally meaningful sense, have that system. Industrial capacity is operational capacity. Canada’s own strategy documents now say versions of this. What is missing is the mechanism.
What Ukraine’s distributed manufacturing actually proved
Ukraine has run the largest live experiment in distributed defence manufacturing in modern history, and the structural findings are independently corroborated by CSIS supply-chain analysis and a NATO-commissioned Ifri battlefield study. The lessons are not the ones you’d guess.
Production disperses; it does not deploy forward in romantic container factories. Quality travels with the work as controlled production packages, verified components, frozen configurations, documented instructions, trained operators, a hard test gate, with safety-critical steps kept centralized. Finished inventory decays while designs iterate in weeks, so the smart position is to hold current designs, component kits, and qualified capacity rather than warehouses of units going stale. The localization cost premium turns out to be small, and it gets closed on the engineering bench through design for production, never through cheap labour. And no single design wins; the NATO-commissioned study puts it plainly: the innovation is in adaptation capacity, not in any one design.
That last finding dissolves a political trap, incidentally. Canada does not need to pick a winning drone, which is fortunate, because no government survives picking one. It needs the system that can produce whichever qualified designs the moment requires.
Industrialization starts in design
Here is the principle underneath all of it, the one I most want Canadian program designers to sit with, because we have lived it across hundreds of programs.
A product’s ability to be manufactured at scale, and above all its ability to be manufactured by someone other than its designer, across multiple independent sites, is decided early. It is decided in architecture and module boundaries, in tolerance philosophy, in process selection, in test access, in how much of the build depends on knowledge living in one technician’s hands. None of it can be bolted on later at reasonable cost.
Which means a design matured through three contest rounds without design-for-industrialization requirements will arrive at the end of the pipeline unable to distribute or scale, regardless of who funds it next. The money spent maturing it will have quietly locked in that outcome. Nobody decided this. It was decided by omission, at the CAD stage, years before anyone asked about production.
The practical implication is immediate and cheap: design-for-distributed-production requirements belong in the pipeline now, at the prototype stage, where meeting them costs almost nothing, rather than after selection, when meeting them means redesign. The designs now entering prototype development are the natural place to start.
What the missing layer looks like
The full architecture is more than a blog post can carry, but its shape is simple. In one sentence: a manufacturing-readiness network is shared infrastructure, standards, qualification, data, and demand, that lets many independent producers build any approved design at qualified quality. Approved designs travel as controlled production packages. A plural qualification gate admits every design that meets a published standard, so government standardizes interfaces and codifies many designs rather than anointing one. A federated digital backbone connects independent producers for configuration control and traceability, purpose-built for the mission, because no off-the-shelf enterprise system spans independent companies. Critical inputs get managed deliberately, because the chokepoints are materials and components, not assembly. And the whole thing is paired with standing demand, because a production network with no consuming customer is a demonstration of process, not a defence capability.
The first step should be small on purpose. Prove the method on a bounded product, a battery pack and its repair ecosystem is the right kind of hard, before extending it to complete systems drawn from the prototype pipeline, and eventually to Northern sustainment. The vision is national. The first step is not.
Why Inertia is saying this
Inertia builds products for a living, from strategy through design into manufacturing, under one quality system spanning ISO 13485 and ISO 9001, across facilities in Toronto and Guangzhou. We were founded to close a specific failure mode at program scale: every vendor performing well inside their slice while the program fails, because nobody owns the whole. Canada’s defence-industrial push is currently running the same failure mode at national scale, and we recognize it because closing it is our trade.
We’re sharing this thinking openly because the network only works as a genuinely shared national capability. A version captured by any single firm, ours included, would defeat its own purpose. If you’re working on any piece of this, in government, at a prime, at a ten-person drone shop, the conversation is worth having now, while the design decisions that determine everything are still cheap to make.
