Strangers and friends alike, Canadians who read the news all seem to have an opinion about the fighter jets, and almost all of those opinions seem to be about a kill switch. Can the Americans turn our jets off? Probably not with a switch, no. They can withhold the software updates, the spare parts, the threat data and the export permission, which gets you to the same place with more paperwork, so I understand the instinct. But I think the argument people should be having is a duller one, and it’s the one I keep running into in my actual job:

Can we fix them ourselves?

Fifty years is a long time to own anything

The CF-18 entered Canadian service in 1983 with a design life of about 20 years. We planned to retire it around 2003, then 2020, then 2025, and the current plan gets the last of them into a hangar for good in 2032, when the oldest ones will be pushing 50. There is no further extension available, because this is a metal fatigue deadline rather than a political one, and aluminum doesn’t care about the fiscal cycle.

I’m actually a bit proud of that as a Canadian. We buy expensive things rarely and keep them long past the point where the manufacturer has stopped caring, because we’re a mid-sized economy with a continent-sized set of responsibilities. Our frigates are old. Our submarines were famously, catastrophically old. Our passenger trains are old (70-80 years old, retiring next year). Whatever we buy in 2026 will be flying in 2070, maintained by people who haven’t been born yet, running software written by people who will be long retired.

So the important question isn’t how well a jet flies in 2026. It’s whether we can still fix it on our own in 2055, when the vendor has moved on, the political relationship has turned over three or four times, and the original engineers have enjoyed what we can only hope were long lives.

Age before beauty

If you’ve ever run infrastructure, then you already know what this looks like. There’s a version of every system where you own the hardware and the config and the ability to patch it, and there’s a version where you rent access to somebody else’s version and you’re one licensing decision away from a very bad quarter. The second is always cheaper on day one and always more expensive by year 10, and the difference isn’t really money, it’s who gets to decide.

Here’s the part that initially concerned me when I started digging into this F-35 deal: we have proven the benefits of being able to fix and maintain our own CF-18 Hornets, but we’d be giving that up with the F-35A.

Support for the Hornet has run out of Mirabel, QC for over three decades and includes the operational software. That’s why we could buy a batch of used Australian jets, load our own software onto them, and put them into service as though they’d always been ours. We have that authority over a 43-year-old aircraft, whereas under the F-35 we’d have next to no authority. The threat libraries that tell the jet what it’s looking at when a radar pings it are built in a lab on American soil under American process control. The system that tracks parts and aircraft health phones home by design. And deep repair of the stealth structure itself happens at three sites on this planet, in Texas, Italy and Japan, none of them here.

Those terms aren’t laws of physics. Israel negotiated its own depot and its own electronic warfare suite, and remains the only country that did. So the arrangement Canada is being offered isn’t the only arrangement available, it’s the one available to us.

Saab’s counter is source code access, independent upgrade rights, and a secure data centre in Montreal that keeps Canadian mission data in Canada. It isn’t open source in the sense an IT pro might mean it… But only one of these deals lets us compile a change, test it, and put it on an aircraft without asking a foreign capital first.

We’ve collectively decided that right to repair matters. We got angry at Apple for screens that refused to work when a third party installed them. Farmers spent years fighting John Deere over tractors that would rather drive themselves to a dealership than let their owner swap a sensor. The argument was simple enough that it shouldn’t have needed a policy paper: if you paid for it, and it breaks, and you can fix it, waiting for the manufacturer’s permission is a transfer of power dressed up as a warranty.

Somehow that logic evaporates the moment the purchase order has nine digits on it.

The two jets, briefly

The F-35 is a single-engine American stealth fighter designed to go first into airspace somebody has spent a lot of money defending. It’s hard to see, it fuses everything its sensors find into one picture, and it shares that picture with everyone around it. It’s less an aircraft than a node on an American intelligence network that happens to have a pilot in it, and the network is the actual product.

The Gripen is a single-engine Swedish fighter designed by a country that spent the Cold War assuming it would be invaded. It flies off short strips and highways, gets turned around by a small crew in about the time it takes to order a coffee, and costs as little as possible every hour it’s up. It also has ~35% longer flight range even fully loaded, which strikes me as a strange thing to shrug at in a country with several thousand kilometres of Arctic to look at.

How we picked

The evaluation leaked to CBC and Radio-Canada last year and it wasn’t even close. The F-35 scored about 95 percent on military capability. The Gripen got 33 percent… National Defence hasn’t confirmed the numbers and has opened a leak investigation, so hold them loosely, but nobody has disputed the direction either if that gives you a hint.

The score isn’t the interesting part. The ruler is. Those criteria were written around a specific idea of what a Canadian fighter is for, which is flying alongside the Americans into somebody else’s contested airspace, and measured that way the F-35 laps the field. That’s what happens when the people writing the requirements have spent entire careers inside an alliance, and it would honestly be strange if they’d produced anything else.

But look at what it apparently didn’t measure. The most common thing a Canadian fighter will ever do is fly a very long way to look at something and fly a very long way home, and the aircraft with meaningfully more range couldn’t turn that into a score. Neither, as far as anyone can tell, could the question of whether we’d be able to maintain the thing ourselves for the half century we know from experience we’ll be keeping it.

Oh it gets better, though. In March, the American general who actually commands NORAD told a Senate committee that stealth isn’t needed to defend North America and is better saved for overseas, and that he’d rather modernize older aircraft for the intercept job. So the officer responsible for the mission we buy fighters for went in front of his own government and made, more or less, the Gripen argument, while our evaluation was busy scoring the Gripen at 33 percent for not being stealthy enough to do it.

Both can be true. Stealth matters enormously for the worst day we might ever have, and barely at all for the job we do every week. That gap is the entire debate, and our scoring model discretely settled it without anyone asking us.

What changed, and where the money goes

We’re legally on the hook for 16 jets, not 88, whatever the news implies. The first goes to Arizona for training this year. Long-lead payments hold slots for another 14. The remaining 72 are what everyone is actually arguing about.

Meanwhile Saab offered to assemble 72 Gripens in Canada with full technology transfer and the source code, and Ottawa has already bought the surveillance half of that package, six radar aircraft built on a Bombardier airframe, with NATO now negotiating for up to 10 more. We’ve signed a defence partnership with the European Union, become the first non-European country let into its procurement fund, and taken an observer seat on the next-generation fighter that Britain, Japan and Italy are building. Sweden joined NATO in 2024, so no one can argue we’re leaving the alliance at this point either, rather we’d be leaving a supplier: The USA.

On money, the F-35 purchase has gone from 19 billion to 27.7 billion, with the auditor general saying another 5.5 billion is needed for infrastructure and weapons nobody counted. Flying the thing is expensive too. An hour in an F-35 runs somewhere between 33,000 and 46,000 US dollars, against about 22,000 for a Gripen, so we’d be paying roughly double for every hour spent looking at the same empty stretch of Arctic. That’s assuming the aircraft is available to fly at all, which the Americans measured last year at 44 percent across the fleet.

Then there’s a cost absent from every comparison I’ve read. The RCAF is standing up an entirely new trade whose job is guarding sensitive technology, the F-35 very much included, and internal documents suggest it may eventually need more than a thousand people. That’s an air force of about 15,000 that is already short 2,000.

This aircraft is the most concentrated collection of Western military technology ever parked on a runway, and China has spent close to 20 years trying to take it. Chinese operations exfiltrated F-35 design material out of contractor networks, radar systems, engine schematics, methods for cooling the exhaust, as part of a wider campaign against American defence programs that American intelligence assessed at something like 50 terabytes in total. Their next two fighters came out looking oddly familiar… So we’re hiring a thousand people to stand guard over a design that was already stolen before we ordered it, and none of that appears in cost per flying hour, because cost per flying hour measures the aircraft. The sticker price is for the jet. The monthly bill is for the ecosystem it arrives in.

But don’t stop at the dollar figure we’re paying, look at how little we’re getting back: Normally a foreign supplier winning a big Canadian contract has to generate economic activity here equal to the full value of it. That’s the standard rule, it applies to essentially everything, and yet we waived it for the F-35. So the ledger is roughly 30 billion out, a few hundred million a year back in supply chain work our firms won on merit, with no assembly line, no source code, and the mission data living somewhere else. Lockheed reminded us in 2016 that it could always move that work to countries that were buying, which emphasizes how little the Americans care. Not to mention the gall some of them have to ever say Canada benefits more than the USA on just about every deal we have with them from forest fire fighting to a shiny new Gordie Howe International Bridge.

Too good to be true

Switching to Sweden doesn’t get us all the way out, and anyone selling you that is selling you something. Roughly a third of the Gripen is still American content, and the decisive piece is the GE engine, which means Washington keeps a permission slip over where those aircraft can go. Saab can’t hand over intellectual property it doesn’t own.

The reason I’m confident that lever is real is that it got pulled 18 months ago. When Colombia selected the Gripen, the Americans moved to withhold the engine re-export licence, apparently to steer Bogotá toward an F-16 instead. That is my entire argument, demonstrated by the Americans, on a third country, specifically to stop a Gripen sale. And they haven’t been subtle with us either, since their ambassador said in January that the NORAD arrangements would have to be altered if we trimmed our order. So the real summary is that the Gripen gets us the software, the data, and the factory while leaving the propulsion dependency in place, which is a long way from where we are and rather short of full independence.

The other fair objection is delivery. Brazil’s Gripen line took considerably longer than planned, and while that’s worth knowing, it almost certainly says more about standing up aircraft production in a country that hasn’t done it lately, than it does about Saab. Anyone who has watched a Canadian infrastructure project will surely find this unsurprising. Look no further than our newer metro rail or especially the daunting outlook on nuclear energy. We thought Small Modular Nuclear reactors would make quick and cheap work until the proof of concepts took just as long and cost nearly as much as past full sized reactors.

Generally, most objections about the Gripen only work if the alternative is on time, and it isn’t. The new F-35 tailored fighter facilities at Cold Lake and Bagotville have slipped past 2031, while the Gripen takes on any short runway track with a tiny crew. The F-35’s major software upgrade has been cut back and pushed to at least 2031. Some aircraft have been delivered with “ballast” (think rim weights to balance the tire) bolted in where the new radar was supposed to go. Every path we can take has a delivery risk running at that 2032 deadline, which means delivery risk stops being a reason to pick one over the other. It’s just the weather now and maybe having two options increases our ability to receive an aircraft on time.

The Arrow, from the other side of the table

You can’t write about Canadian aerospace without someone bringing up the Avro Arrow, usually as a grievance about a small-minded prime minister and some destroyed blueprints. The more useful reading is that the Arrow was a superb aircraft with nobody to sell it to, built for a mission only we had, priced for a customer base of one, in an era before open competitions, offsets or technology transfer existed.

In 2026 that problem has inverted. Saab has more demand than it can build, Ukraine alone has asked for more aircraft than Sweden can produce, and NATO is negotiating for more of the surveillance planes we’ve already bought. We aren’t a desperate seller hunting for a customer this time, we’re a customer holding the thing a manufacturer urgently needs, which is somewhere to build. That’s the strongest position Canadian aerospace has been in since 1958 and the only way to waste it is to not use it.

To be clear, Saab keeps design authority and we would be assembling their aircraft rather than designing our own. But treating that as failure misunderstands how any of this works. There’s a ladder here. At the bottom you have no authority at all, which is where the F-35 puts us. Above that is the ability to modify and fix your own aircraft, which is what Saab is offering. Above that is being a partner with a vote in what the thing becomes. Design authority, the Arrow rung, hasn’t been handed to a middle power in 70 years and is not on any menu we’re being shown.

You cannot jump from the bottom rung to the top. What qualifies a country to be a partner worth having is a workforce that has actually built the aircraft and touched the code, and we currently have neither. We’re sitting in an observer’s chair on a sixth-generation fighter program with Britain, Japan, and Italy right now, and observing is all we can do, because we have nothing to bring. Assembly and source code aren’t a consolation prize for missing out on design authority; they’re the entrance exam.

What I actually think

The evidence doesn’t settle this, and every piece I’ve read that claims otherwise got there by picking which evidence counted. The tactical case points at the F-35 unless you care about range. The cost, sovereignty, longevity and industrial cases point at the Gripen. Those aren’t competing readings of the same facts, they’re different facts, and no amount of analysis magically dissolves them.

So the tiebreaker isn’t in the evidence. It’s a judgment about whether the last two years of American behaviour were weather or climate. If weather, the F-35 is the efficient answer and all this hedging will look faintly hysterical in a decade. If climate, then every year we spend deepening a dependency makes the eventual correction more painful.

I’ve come around to thinking it’s climate, and the more important argument isn’t even about the Americans.

A Western fighter market with one serious supplier is bad for everyone in it, including the supplier. Monopoly suppliers don’t improve. They don’t control costs, they don’t fix readiness problems with any urgency, and they don’t have to answer questions, because where else is anybody going to go? Look no further than China, who has to steal 50TB of F-35 confidential information to improve their own next-generation fighter. The only thing that has ever fixed the limitations of a monopoly is a credible second option, and a credible second option only exists if somebody actually buys from it. Canada is one of very few countries with the money, the requirement and the industrial base to make that happen.

The obvious objection is that we’d be paying for it while Norway and Denmark and Finland enjoy a more competitive market for free. That’s true, and it’s the honest cost of going first, and somebody has to or the thing never exists at all. If we don’t, the market in 2040 is the market we have now, and we’ll be complaining about it with less leverage than we have today and a fighter we have to replace so much sooner after a much higher total cost of ownership. Which puts us in a bit of a lousy position financially as a country, one where most people would agree military spending is already too high…

Anyone who has moved a production system off a vendor knows how this goes. You spend 18 months learning interfaces you didn’t ask for, you run both systems in parallel far longer than anyone promised, half your institutional knowledge turns out to be about the old thing and evaporates, and there’s a specific awful moment in year 2 where you can’t tell whether your problems are migration problems or whether you simply chose wrong. Sometimes you did choose wrong, and you only find out after you’ve committed.

But the alternative is staying, indefinitely, with a supplier whose government has spent 18 months treating your economy as a bargaining chip, on terms where you can’t open the hood to check the oil. We’ve all worked somewhere we stayed too long because leaving looked hard. It was never really about whether leaving was hard. It was about whether we believed the situation was going to improve on its own.

Fifty years is a very long time to be wrong about that.