Canada is about to spend more on defence than any living Canadian has planned for. We have $180 billion in procurement over ten years, a new agency to run the buying, ten named sovereign capabilities, and a target of seventy percent of purchases aimed at Canadian firms. The emerging debate is about numbers: is it enough? Can we afford all of it? Where should it go?
But there’s another question that needs asking, and I want to explore it here because I don't have an answer yet: if the most valuable thing available to us is dual-use technology—and I will argue it is—are we in Canada structurally capable of taking delivery? Answering that question concerns the plumbing (i.e., the systems and structures through which investment produces tangible returns) rather than the politics (the funding announcements, rhetoric, and optics); it is worth concentrating on the state and functionality of that plumbing since it is what will actually determine what our money buys and unlocks in terms of long-term value.
I will skip two elements of the broader discussion, since both are well covered and neither is the terrain where the problem exists. The first is the ingredients, i.e., what’s in our national inventory—the talent, universities, critical minerals, capital markets, geography, and alliance networks. These assets are acknowledged at every conference in Ottawa, and beside the point. Nobody here argues that a country that is as large, wealthy, and educated as Canada lacks the raw materials for a successful defence buildup; that's almost beyond doubt at this point. The second is the etymology: dual-use simply means a technology with a military and civilian application. The term is accurate as far as it goes, and it explains nothing new (by that standard, dual-use is as old as the wheel, and the analysis goes nowhere).
What I want to explain is why dual-use is different now, in 2026, in Canada, in a way it wasn't in 1960 or 1990. The concept is old, but the situation is new, and seeing whether the former can meet the challenges posed by the latter is the whole story.
I. Spin-off and Spin-In
There are two technical terms worth knowing since they provide orientation for the rest of the discussion. The first is “spin-off,” which refers to technology developed for military purposes that later finds civilian use. For fifty years, this was the normal direction of travel for dual-use, and many of its examples now serve as the furniture of modern life.
ARPANET, built by the American defence research agency to keep computers talking to one another through disruption, provided the foundations of the internet. The Global Positioning System, or GPS, began life as a U.S. Air Force navigation constellation before it turned into the mechanism we rely on to find a hotel or restaurant. Radar came out of air defence and now sits in every airliner and most cars. Jet propulsion, nuclear power, microwave ovens, the integrated circuit: all of it emerged as something a government wanted or needed for military reasons.
Canada has its own lineage in this field, and it is more impressive than most Canadians realize, if less well known. In 1949 the Royal Canadian Navy and Ferranti Canada built DATAR, a digital system that fused radar and sonar from several ships into one shared picture of the battle. It worked on Lake Ontario in 1953, a decade before anyone else managed the same trick. Canada could not afford to fit the fleet, and the Americans built their own instead. But the computing went on to power one of the world’s first computerized airline reservation systems, for Trans-Canada Air Lines, and became the model for a mainframe sold to Britain that became the basis of a major British computer line. Twenty years later, Spar Aerospace built the Canadarm, and the robotics expertise flowed outward into MDA, surgical robotics, and industrial automation.
That is the pattern for how spin-offs become game-changing applications in everyday life: the state pays for the hard part, and the civilian economy inherits the benefit. In short, a country’s technological standing was a function of what its government was willing to fund and support.
Then there is “spin-in,” which is the reverse: technology developed for the commercial market, adopted for military use. And this is now the dominant direction, replacing rather than merely supplementing the spin-off model.
Consider where the leading edge of computing sits: machine learning, sensing, autonomy, batteries, space launch and manufacturing are effectively the domains of commercial companies, funded by commercial investors, improving on commercial schedules. No defence department on earth is out in front of it. The most consequential battlefield innovations of the past three years—now proving their worth in engagements across Ukraine and the Middle East—largely came from small workshops assembling uncrewed aircraft out of parts anyone can order, from commercial satellite imagery, from off-the-shelf communications terminals, from software shipped in weeks.
Canada's clearest current example runs both ways at once. RADARSAT watches ice, floods, crops, and shorelines for civilian agencies, and it watches ships and the waters off our coasts for military ones. The technology operates with the same satellites, the same engineers, and the same firm. There is a seamless, frictionless interoperability between defence and non-defence—a feature that sums up what dual-use means in 2026. It no longer refers to a military product with a civilian spin-off, nor to a civilian product with a military bolt-on; rather, it encompasses one capability with two sets of customers, and ideally, no conversion step in between. And the range of what qualifies under this category is wider than most people expect, which is part of why citizens and policymakers keep missing it.
More examples abound if you know where to look. Snapchat is a photosharing app for teenagers; it is also a company with deep expertise in real-time camera processing and face detection on cheap hardware in bad lighting, which is a fair description of a military targeting problem. A fitness tracker can count the steps, miles, and burnt calories each time you run; in 2018 one of those apps published a global heat map that revealed the internal layout and patrol routes of overseas military bases, because soldiers go running too. An ocean sensor deployed to monitor water temperature is a hydrophone, and its uses for commercial exploration or academic research purposes are countless. But a hydrophone can also hear submarines and thus becomes a tool for conducting military reconnaissance.
Hardly any of those technologies were built with defence in mind, and in most cases, nobody at the company ever considered it. That is the aperture we are dealing with: less a tidy sector of contractors and defence-adjacent firms, and more a wide and largely unmapped set of ordinary companies holding capabilities that can decide whether a nation can defend itself from threats.
II. Why the Direction Matters
This shift from spin-off to spin-in carries implications that need to be understood. Reversing the direction of flow does not just change where technology comes from; rather, it changes who has to be able to talk to whom and what each party has to be good at.
Under spin-off, the state was the origin point, and it set the technology’s requirements, funded the science, owned the result, and released applications downstream when it was ready. Everyone else was a recipient: the relationships were vertical and the sequence was clear. Under spin-in, the state is the destination. The technology is already out there, already funded, already improving without anyone’s permission. And the state's job is no longer to originate, but to recognize, evaluate, buy, and adopt—quickly enough that the tech is still current when it arrives.
Those are completely different skills: originating rewards patience, specification and control, while receiving rewards speed, judgment, and a tolerance of innovations designed for reasons other than national security. The last seventy years have seen the rise of institutions optimized for the first job: DRDC and the National Research Council on the research side, Public Services and Procurement Canada on the buying side, a small set of prime contractors sized for decade-long programs on the supply side, among others. But the changed realities of the present demand ones able to do the second. How we get those institutions to change accordingly is the crux of the problem and the focus of the argument that follows.
But before getting to who has to change, it is worth pointing out why the two ends of this have never been meaningfully tied together. For most Canadian technology firms, the Canadian Armed Forces (CAF) is practically out of reach; it would be one thing to say CAF has been a difficult customer, but the fact is that it isn’t a customer at all. Defence procurement here runs through a small number of established suppliers on long programs, and everyone else—the overwhelming majority of firms building sensing, autonomy, communications, energy, materials—has never had the military in their addressable market. Note that they have not been rejected; they simply have never been in it.
This means that the ordinary process by which market actors on the supply and demand sides discover each other has never actually taken place. For instance, when a company builds something, some customers buy and most do not; the company learns why and adjusts. When that process repeats a thousand times, the market eventually receives products servicing needs nobody could have specified in advance. Commercial technology has to pass through this loop in order to reach higher standards of usability and versatility, and it is exactly the loop spin-in depends on.
Between Canadian industry and the Canadian military, the loop has never turned. So, the requirements are not just unmet; they go undiscovered. And the reason the process cannot correct itself on its own is that the two halves of the market do not share a unit of measurement, or for that matter, any established trust with one another. On the supply side, a firm plans and operates in terms of revenue, margin, cost of acquiring a customer, repeat purchase, retention, growth. And those numbers decide whether it survives, whether it raises money, and what it builds next. On the demand side, an operator measures something else entirely: did the task get done, under what conditions, at what risk, with how much training burden, and does it work alongside what our allies carry? The operator must, therefore, think in terms of effectiveness, ease of execution, interoperability, and sustainment at the far end of a supply chain in the middle of a northern winter.
Neither set of metrics easily converts into the other. Soldiers operating in the field cannot tell a founder what the contract should be worth. By the same token, a founder cannot tell soldiers whether the application they must rely on degrades gracefully when half the network drops. The technology may have genuine dual-use functionality, but if it is used according to measurements with no exchange rate between them, the result is a failure of translation and a loss of coherence, which can be embarrassing in any organizational context but disastrous where national security is concerned. And sitting in the middle, of course, is the public service, attempting to translate the two systems into a third language—compliance, defensibility, documented process—and passing along whatever survives.
So, there is no signal in either direction. Industry cannot see the need. The military cannot see what already exists. Nobody is being obstructive; there is simply no shared instrument to read. The most uncomfortable implication is worth stating: there are almost certainly companies in this country right now holding most of what the Armed Forces need, and neither party has any way of finding out.
III. The Stakeholder Matrix
The two actors already named above, CAF and the public service, turn out to be two of five groups of stakeholders, not the whole story. Resolving the coherence gap between them would require a change of course and a more clear-eyed pursuit of alignment. (There is, furthermore, a sixth party whose standing is different from all of the rest, and I will come to that.) Each of the five groups may be doing something entirely sensible from where it sits, and yet each is also, from the vantage of the other groups, part of the problem.
(1) The end user—the Canadian Armed Forces: They define the requirement and carry the consequence when it fails. Military culture is properly intolerant of failure, and careers end over visible mistakes far more often than over opportunities quietly missed. So, it is rational to demand the complete, fully proven system and wait for it. But requirements now change faster than the buying cycle. Members operate aging equipment while imperfect-but-better options sit on the market today, and the little signal that does reach industry arrives gold-plated, slow and hostile to iteration. For dual-use, the implication is direct: the Forces must become capable of taking something developed in the civilian world before it is perfected for military use, and of learning what it can become by putting it into the hands of the people who will actually use it.
(2) The buyer—the public service: They hold the money, write the rules, run the competitions and issue the clearances. Everything passes through them, and the incentives are worse than they look. As things stand, visible waste is punished hard; opportunities never taken carry no penalty at all. And a program that fails to resolve the problem it was built for is often rewarded with another year of funding, because continued need is easier to demonstrate than success. So slow, defensible, and fully documented wins every time, and a program that never quite finishes does not lead to failure but renewal. Meanwhile, a dollar never spent never compounds, and the public service’s capacity to absorb anything new is the bottleneck the whole system runs through. That bottleneck is especially consequential for dual-use, whose commercial clock keeps running while government deliberates: a capability that takes years to buy, clear or classify may no longer be novel—or Canadian—by the time the state is ready to receive it.
(3) The money—investors: They decide which companies exist long enough to matter. Canadian early-stage capital is measured against software returns and American comparisons. A defence company has long sales cycles, one government buyer and regulatory complexity, so it prices as unfinanceable. It is rational to wait for a safer round, or to send the capital—and the company—south. But a winner cannot be picked out of a population that has been starved, or out of one whose best companies get exported the moment their value is established. This is the financial paradox of dual-use: the very possibility of selling into both civilian and defence markets should make these firms more resilient, yet the uncertainty of the military side can make Canadian capital abandon them before that second market ever has a chance to mature.
(4) The innovators—founders, accelerators, the ecosystem around them: This layer is supposed to translate between the soldier and the market. It is measured on visible activity: events held, cohorts run, logos on a slide. Its fluency is almost entirely commercial; very few people in it have served. So, it optimizes for markets where it already speaks the language, and the translation layer mistranslates instead. But translation is the central work of a dual-use ecosystem: recognizing military significance in technologies built for civilian purposes and rendering military problems intelligible to founders who did not begin by imagining themselves as defence suppliers.
(5) The primes—established industry: They bring what nobody else can, namely scale, or the ability to take something to volume and keep it in service for thirty years. Long programs and offset obligations reward a moat around the customer, so it is rational to protect margin, control access, and treat a startup as either a threat or cheap subcontract capacity. But the moat starves the ecosystem underneath it, and the country pays more for less. Dual-use technology therefore poses a particular test for the primes: whether they can serve as the bridge by which capabilities developed elsewhere enter the defence system at scale, rather than as the gate through which those capabilities must surrender their independence before they are allowed to pass.
Once again, the five groups hold defensible positions when taken on their own, but taken together, they amount to collective failure. This realization was not born in the abstract. Rather, it came to me from the experience of sitting in three of those five chairs and spending a lot of time around the other two. Nine years in the army taught me why a soldier wants a hundred-percent-effective solution he can depend on. Time inside a federal department taught me why a public servant documents the decision rather than makes it. And doing investment diligence taught me why a defence company looks unfinanceable on a spreadsheet.
What surprised me was the symmetry. In every room, the people were competent and behaving sensibly given what they were measured on—and the other four groups were named as the obstacle. Nobody here is being irrational. That is the finding, and it is worse than the alternative. Consider that a system failing because of a few bad apples could be fixed by removing those actors, but a system failing because everyone is behaving sensibly calls for a deeper remedy, one that involves changing what “sensible” means. As to what exactly that will look like—that is a greater discussion than what can perhaps be captured here; suffice it to say, for now, that the five groups are in urgent need of a new alignment and a new equilibrium, and what’s required to get there will be outlined in the next section.
A sixth party is central to all of it: calling them a stakeholder would understate the case, and equilibrium would not be possible without them either.
(6) Indigenous nations and communities: they are rights-holders, with jurisdiction, treaty obligations and land, and much of what this country calls Arctic security physically sits in Inuit Nunangat. And that premise ought not be regarded as a mere abstraction. For instance, the operations and maintenance contract for the North Warning System, worth $592 million over seven years, is held by Nasittuq, a majority Inuit-owned company. Northern defence infrastructure means northern employment, northern logistics, and northern consent.
Thus, adoption cannot be understood simply as a transaction between a civilian innovator and a military customer: where a dual-use capability depends on Indigenous lands, infrastructure, knowledge or data, the rights-holders themselves are part of the institutional architecture through which that technology must be developed, governed and put to use.
On paper, everyone agrees about this. The Defence Industrial Strategy names Indigenous rights-holders explicitly, and I have not sat in any of the five chairs and heard anyone argue against the principle, which has come as close to a consensus as anything. But whether it is being carried out in practice is a different question, and I am not the right person to answer it. I am working through OCAP training—the First Nations principles of Ownership, Control, Access and Possession—and I have more to learn here than I have to say.
What I can offer is one observation from the part I do know. Unanimous agreement on paper is exactly the condition under which nothing gets tested. Friction is how a system discovers whether it is actually doing the thing; where no one argues, there is no signal either way. That should make all of us uneasy, and it deserves someone better positioned than me to examine it properly.
IV. Getting to Alignment
What is missing on the road to alignment, in fact, are two different things, each corresponding to a different problem. The first is a place to meet. Right now, there are only a few doors into a large public institution: the job competition, the security clearance, and above all the request for proposal—the document a government publishes listing what it wants and how bids will be scored, with the scoring fixed before the first bid arrives.
Every one of those doors filters for the same quality: whether a technological application or capability can be described in advance and matched to a template that already exists. This means that an RFP can purchase only a solution that the institution already knows how to specify. If the strongest available response instead reframes the problem (“you asked for a better sensor, but what you need is to stop needing the sensor”), the procurement process has no obvious way to accommodate it. There is no category into which such a proposal readily fits. It is therefore not rejected on its merits; more often, it never reaches the point at which it can be formally submitted.
Ursula Franklin named this pattern in her 1989 Massey Lectures. Technology, she argued, is not a pile of objects but a practice: “the way things are done around here.” She distinguished systems where the person doing the work controls the process from systems that break work into fixed steps set by somebody else. The other kind reproduces what is already known with tremendous efficiency and builds a culture of compliance while doing it. Canadian defence procurement is in the latter category. It is not broken. It is working exactly as designed. Predictable inputs guarantee predictable outputs, and a predictable output is never a breakthrough.
Our usual response to this is itself part of the pattern: we add a program. In 2018 the federal government counted ninety-two business innovation program streams across twenty departments and announced it would cut them by up to two-thirds. Whatever became of that, the Defence Industrial Strategy has since added more. These efforts were made by people fixing a real gap. But each represents another door, built to the same specification as the doors already there. Instead of more access, they built a maze with excellent intentions.
The same question applies to the new agency. The Defence Investment Agency is being staffed substantially from within: many of the people building it come from the procurement organizations it is meant to improve upon. That may be unavoidable, and there is a fair defence of it: one cannot reform a machine nobody in the room has operated. But the question is worth asking out loud. Is this a new agency, or the existing one with a new name and a larger budget?
I do not know. I do know the answer will not be in the mandate letter. Rather, it will be determined by whether the first intake process the new agency designs looks any different from the last one. So, we need a different surface: call it a bench. Concretely, six things: a buying process that does not outlast the company trying to win it; clearances that move at the speed of the work rather than the speed of the queue; a path that does not force a company to choose between its civilian market and its military one; neutral ground where all six groups can meet without defending their employers; money shaped like a ten-year hardware company rather than a software startup; and enough published accountability that people will use the thing twice.
The second missing element is harder, and it is the one I would put money on being decisive. A table where things can be set down is useless if nobody around it will pick anything up. Call that willingness absorptive capacity: the ability to accept a solution you did not build, in a form you would not have chosen, from someone you do not yet fully understand.
Every one of the five groups is short of that capacity: the soldier being asked to field something imperfect; the public servant being asked to exercise discretion rather than document its absence; the investor being asked to fund a category the spreadsheet rejects; the accelerator being asked to learn a language it does not speak; the prime being asked to open the moat. And, in a form none of the five has yet had to reckon with seriously, the whole system is being asked to receive direction from rights-holders rather than merely consult them.
That is ultimately what alignment means in a dual-use system: not that the six parties come to think alike, but that a capability born in one world can move through all of them—recognized, financed, adapted, authorized, scaled and governed—without being stopped at each boundary precisely because it did not originate on the other side.
V. An Open Question
I said at the outset that I was posing a question rather than making a claim: are we structurally capable of taking delivery of dual-use technology? I meant that distinction. But we can already identify what the answer will depend on. It will not depend simply on the money, which has so far only been announced; nor on the strategy, which has been written; nor on the agency, which is now being built. These are all necessary pieces, and yet Canada has never been short of pieces.
It depends on whether anything changes about the way the six parties receive what the others bring. If nothing does, we will spend $180 billion through a filter that admits only what can be described in advance—and buy a great deal of what we already knew how to ask for, from the people who already knew how to describe it, at exactly the moment that demanded something else. And then we will wonder, again, why the engineers left.
The hopeful reading is that this is a smaller problem than it looks. For most of the last century, the technology that decided whether a middle power stayed relevant required an industrial machine we could neither build nor afford. That is no longer true. Spin-in has put the frontier inside the ordinary economy, within reach of a country with good universities, capable founders and functioning capital markets. That Canada has all three of these assets is not in question. What we do not have, however, is the habit of picking things up.
That is not a budget problem nor a technology problem. It comes down to whether a procurement officer publishes a problem instead of a specification. Whether an investor writes a first cheque into a company whose only customer is the government. Whether an officer fields something at eighty percent and improves it in service. Whether an accelerator learns enough of the military problem to recognize the defence application in something built for another market. Whether a prime gives a small company real access to the end user. Whether Indigenous rights-holders are brought in early enough to shape how a capability touching their lands, infrastructure, knowledge or data is developed and used, rather than being asked to consent to it after the important decisions have already been made. Small moves, from ordinary chairs.
And it is also worth remembering that most of the people who will build this industry over the next twenty years are not in it yet. Some are in engineering programs right now; some are in uniform; some are making software for something completely unrelated and have never thought of themselves as having anything to do with defence. Whether they end up here—or somewhere else, one at a time, the way they always have—is the actual question in front of us.
In 1953, we had the capability and no way to carry it; consequently, the technological edge that Canada produced fell into the hands of others. Today, we are building the next wave of dual-use innovations. I would like to think we will recognize the capability to absorb them when it turns up in a form we did not specify. I am not yet sure we will.
Stewart Leckie is a nine-year veteran of the Canadian Armed Forces with the Cameron Highlanders of Ottawa Reserve. He has worked in federal government and early-stage venture capital, and is Director of Defence and Dual-Use Technologies at the startup accelerator L-SPARK. He writes on defence and dual-use innovation from Ottawa.