CAE Inc.: The Invisible Infrastructure of the Skies

I. Cold Open & Episode Roadmap

The hydraulic legs shudder first. Then the flight deck pitches, the nose rises, and somewhere over a runway that does not exist, in a city that is only a photograph draped over a projection dome, an engine dies.

The captain in the left seat has flown this aircraft for eleven years. She has never once lost an engine at V1 — the speed past which a crew is committed to fly, no matter what. Nobody has. That is precisely the problem. The events that kill airline passengers are, by construction, the events pilots never get to practise. So the industry built a machine to manufacture them: a six-legged steel box on hydraulic actuators, wrapped around a bit-for-bit replica of a real cockpit, fed by aerodynamic models licensed from the aircraft manufacturer itself, and certified by a regulator to be legally indistinguishable from the aeroplane.

That box is the product. And the largest maker and operator of those boxes is a company most passengers have never heard of — headquartered in a Montreal suburb, founded in a borrowed hangar repairing radios for the Royal Canadian Air Force.

This is the story of CAE Inc. It is a story about how a postwar electronics shop stumbled into one of the most structurally protected demand pools in global industry: training that is not optional but legally mandated, consumed repeatedly and indefinitely, by an industry that cannot operate without it. It is also a story about how a company sitting on that kind of asset still managed to destroy most of a decade of shareholder value through its own capital allocation decisions.

The numbers frame the paradox. In fiscal 2026, ended March 31, 2026, CAE generated revenue of C$4,914.0 million, up 4% year over year, with adjusted earnings per share of C$1.20 — essentially flat against the prior year's C$1.21.² The company operates in two segments: Civil Aviation, the airline and business-jet training franchise that contributed C$2,741.6 million of that revenue, and Defence & Security, which contributed C$2,172.4 million.² As of the first quarter of fiscal 2027, reported on August 12, 2026, the adjusted backlog stood at C$19.19 billion — C$8.5 billion in Civil and C$10.7 billion in Defence — a multi-year book of contracted work that is uncommon for an industrial company of CAE's size.¹ CAE operates training sites and locations in more than 40 countries.¹

And yet. Shares that traded at C$47.40 on January 16, 2026 changed hands at C$34.61 on August 25, 2026. On May 22, 2026 — the session after fiscal 2026 results and a newly disclosed multi-year transformation plan — the stock fell roughly 13.5% in a single day.²⁶ The market's verdict, for now, is skepticism.

The arc from here: a Cold War electronics shop that backed into simulation almost by accident; the slow, decade-long pivot from selling machines to operating training networks, which is the single decision that created most of what is valuable about CAE today; sixteen years under a chief executive who built a formidable core business and then repeatedly overpaid to expand alongside it; a billion-dollar defence acquisition that unravelled; an activist investor from Los Angeles who arrived at an inflection point; a governance reset engineered with the quiet backing of Quebec's largest pension fund; and a new American chief executive who has been in the chair for less than a year and has signalled to investors that the path to recovery runs through further near-term pain.

Running beneath all of it is a harder question for anyone holding the stock: if the competitive moat is as durable as the company's advocates argue, why has return on capital persistently fallen short of the cost of it?

Start where it started — in a hangar, with eighteen people, on St. Patrick's Day.

↑ Back to Top

II. Founding & Cold War Origins (1947 – mid-1970s)

Kenneth Patrick was a former Royal Canadian Air Force officer whose founding ambition was national-industrial before it was commercial.[^4] Canada had spent the war training aircrew for the entire Commonwealth and manufacturing aircraft at scale. When peace came, the apparatus dissolved — engineers scattered, contracts evaporated, talent drifted south.

Patrick's stated goal was to "create something Canadian and take advantage of a war-trained team that was extremely innovative and very technology-intensive."³ On March 17, 1947 — St. Patrick's Day — he incorporated Canadian Aviation Electronics Ltd. and took over a vacant hangar at Saint-Hubert Airport, across the St. Lawrence from Montreal. Headcount: eighteen.³

The early work was unglamorous. CAE repaired and overhauled ground communications gear for the RCAF and installed antenna infrastructure in the Arctic — the electronic plumbing of a country that had just discovered it sat directly between the United States and the Soviet Union.³ There was no simulation strategy. There was payroll.

The pivot came in 1952, and it arrived as a customer problem rather than a founder's vision. The RCAF was introducing the CF-100 Canuck, an all-weather interceptor, and needed a way to train crews without burning fuel and airframes. CAE won the contract to build a flight simulator for it.³ From the outside, this looked like one more defence contract. In hindsight, it reoriented the entire company.

What Patrick's engineers had stepped into was a business whose demand was created by regulation and risk aversion rather than discretionary preference. As aircraft became faster, more complex, and more expensive, the marginal cost of training in the real machine rose — in fuel, in airframe hours, in lives. Simulation was the arbitrage. And the more sophisticated aviation became, the wider that arbitrage grew.

The 1960s deepened the position. CAE won a Canadian government contract for six F-104 Starfighter simulators; within five years, five NATO countries had ordered twenty-six more.³ Along the way the company developed radar land-mass simulation, visual systems, and the motion platforms that gave the illusion its physical credibility.³ Three elements — motion, visuals, and control-loading, the resistance a pilot feels through the yoke — are what separate a certified simulator from a sophisticated video game, and they are also the three elements that are genuinely difficult to engineer. CAE spent two decades developing competency across all three while most competitors mastered only one or two.

By the 1980s roughly 85% of production was exported.³ The company diversified into power-plant control systems, naval machinery control, and Canadian space robotics.⁴ Much of that diversification was later unwound. The through-line remained constant: build the machine that lets a human rehearse a rare, dangerous event without consequence.

Almost nothing about the founding era bears directly on CAE's investment case today — except one structural insight that was never articulated as strategy in 1952 but now underpins the business: aviation authorities were moving toward making simulator time a legal condition of flight certification. Not a cost-saving option. A requirement.

Which meant that whoever owned the machines — and, eventually, whoever owned the buildings they sat in — would be selling something airlines could not decline to buy.

It took CAE another thirty years to fully grasp that. When it did, it changed everything.

↑ Back to Top

III. The Services Pivot: From Hardware Maker to Training Ecosystem (1980s – 2000s)

In 1982, a United Airlines Boeing 727 simulator built by CAE became the world's first commercial simulator to receive the US Federal Aviation Administration's new Phase III approval — the certification that permitted an airline to conduct full crew training entirely through simulation.⁴

The implication is easy to skim past. Before Phase III, a simulator was a rehearsal device: useful, cost-saving, but a supplement to real flying hours. After Phase III, the simulator became the venue. A pilot could be qualified on a new aircraft type without ever touching the actual aeroplane. The industry came to call it Zero Flight Time training.

For an airline, this collapsed the economics of introducing a new fleet. Instead of ferrying jets around for training sorties, burning fuel and consuming airframe life, carriers could send crews to a windowless room. For CAE, the change did something more consequential: it made simulator fidelity a regulatory variable rather than a commercial one. A box that qualified for Phase III — and later, Level D, its modern equivalent — could do the job. One that did not was a very expensive prop. There was no middle market.

CAE continued pressing the technical advantage. In 1983 it delivered the first FAA-approved simulator for an aircraft — the Boeing 757 — before the aircraft itself was certified, and built the first commercial simulator with digital control-loading and digital motion.⁴ In 1988, CAE Industries acquired Link's domestic simulation and training systems division, then the storied American name in the field, making it the largest supplier of commercial and military flight simulators in the world.⁴

That scale set up the decision that actually shaped the company.

Selling machines versus renting hours

A simulator manufacturer has a structurally mediocre business. Orders are lumpy and tied to aircraft delivery cycles. Each unit is bespoke. Customers negotiate hard because they buy infrequently. Working capital is brutal. The result is a specialised capital-goods maker with a two-year sales cycle and a handful of buyers who all talk to each other.

A training-centre operator has a structurally better one. It owns the box, staffs it with instructors, and collects fees on a schedule dictated not by an airline's capital budget but by regulation. The International Civil Aviation Organization framework requires recurrent proficiency checks at intervals measured in months, not years. Once the asset is built and the centre is staffed, incremental training hours carry high margin. The revenue recurs whether the airline is profitable or not, because a carrier that allows its crews to go out of currency cannot legally fly.

CAE did not flip a switch. Over roughly two decades, from the late 1980s through the 2000s, it gradually shifted its centre of gravity from manufacturing toward operating — announcing in the 2000s a plan to build a global training network and reposition as a training services provider.³ It opened centres with airline partners under joint ventures and capacity-lease arrangements, often placing simulators on its own balance sheet that the airline would otherwise have been required to buy.

That balance-sheet transfer was the counter-positioning move. An airline's return on capital is chronically poor, and every dollar tied up in a simulator is a dollar not in an aircraft. Airlines were, on the whole, willing to hand the capital burden to a third party — provided that party had a centre near their crew base, configured for their aircraft type, and available at the hours they needed. Meeting all three conditions simultaneously is a network problem. And in network businesses, the first mover accumulates advantages that compound.

Two acquisitions bracketed the transition. In 2001 CAE bought Reflectone and the SimuFlite business-aviation training operation, adding scale in the United States. Then, on May 16, 2012, it acquired Oxford Aviation Academy for C$314 million — approximately nine times the target's EBITDA — adding seven civil training centres to reach forty-two worldwide, forty full-flight simulators to reach 211, and four flight academies that brought its cadet training capacity to roughly 1,500 students a year.⁵

Oxford mattered strategically beyond its purchase price. It pushed CAE into ab initio training — taking a person with zero flying hours and delivering a licensed commercial pilot. No competitor covered the value chain that far upstream. Training the cadet means owning the credential record, the type rating, and eventually the recurrent training relationship for a career that may span thirty years.

The switching-cost mechanism, explained plainly

Airline retention is not a function of loyalty. It is regulatory plumbing. A training provider is written into an airline's approved training programme, which is filed with its national aviation authority. Changing providers means re-filing that programme, re-qualifying instructors and courseware, re-scheduling crews around a new location, and absorbing the risk that a crew member goes out of currency during the transition — at which point that pilot legally cannot operate a revenue flight. The switching cost is not primarily a price; it is operational disruption in a business where a grounded crew is a cancelled flight.

By the end of the 2000s, CAE had converted a cyclical machine shop into something closer to infrastructure. The hardware remained the entry ticket; the recurring, regulator-mandated service had become the business. Every subsequent chapter of the CAE story — including the troubled ones — is a story about management searching for something as durable as the franchise it had already built.

In 2009, a mechanical engineer from Bombardier took the chief executive chair and set out to do exactly that.

↑ Back to Top

IV. The Marc Parent Era: Ambition and Overreach (2009 – 2025)

Marc Parent inherited CAE in the wreckage of the global financial crisis, when airline order books were collapsing and simulator sales had dried up. He would run the company for sixteen years — long enough to be judged not on a single strategic bet but on a full capital-allocation cycle.

The verdict is genuinely split, and honest analysis requires holding both halves at once.

The half he got right

Under Parent, the Civil franchise kept compounding. The training network widened, the ab initio academies expanded, and the segment finished fiscal 2025 with revenue of C$2,709.3 million at a 21.5% adjusted segment operating margin.¹⁶ Nick Leontidis, who ran Civil for a decade, nearly tripled the segment's adjusted operating income during that stretch.⁹ Those are not the numbers of a team coasting on structural protection. They are the numbers of an operation that kept adding aircraft types, kept adding locations, and kept converting simulator sales into decades-long training relationships.

Had Parent confined himself to that, his legacy would have been straightforward.

The half he got wrong

He also deployed CAE's balance sheet on three adjacencies, and the record on all three is poor.

The first was healthcare simulation — patient mannequins, surgical trainers, clinical education software. The logic was tidy in presentation: simulation expertise transfers across domains; medicine has the same "you cannot practise the emergency on a real patient" problem as aviation. The logic collapsed in execution. Healthcare simulation has a different buyer — university procurement offices and hospital systems rather than airlines — a different regulatory pathway, a different sales cycle, and, crucially, no legal mandate driving recurrent purchase. CAE built the business for a decade and then sold it, announcing a deal on October 24, 2023 and closing on February 16, 2024, to Madison Industries at an enterprise value of C$311 million, with proceeds directed principally toward deleveraging.¹² The company never disclosed cumulative investment in the division. Selling a ten-year-old business for roughly a third of a billion dollars — into a balance sheet that needed the cash — is not the exit of a compounding asset.

The second was software. On March 1, 2022, CAE completed the purchase of Sabre Corporation's AirCentre airline operations portfolio — crew management, flight scheduling, disruption recovery — for an enterprise value of US$392.5 million.[^10] The thesis was internally coherent: CAE already sat inside the airline's crew-qualification workflow; owning the crew-scheduling workflow too would reposition it as the operating system for airline operations rather than merely its training vendor. Execution did not match the thesis. Rebranded Flightscape, the business today runs a cloud platform with more than 600 professionals across the Americas, Europe, and Asia — and on May 11, 2026, CAE announced it was pursuing "strategic alternatives" for it, including partnership, minority or majority investment, or an outright sale.²⁴ Reporting at the time noted the unit's margins had remained below the Civil segment average, and that CAE was unlikely to recover its original outlay through a sale.²⁵ Four years and roughly half a billion Canadian dollars to arrive at "this may be better owned by someone else" is a capital-allocation result, not a rounding error.

The third was the largest. On March 1, 2021, CAE agreed to acquire L3Harris Technologies' Military Training business for US$1.05 billion — the largest deal in its history. The target carried roughly US$500 million of annual revenue; CAE paid approximately 13.5 times estimated 2020 adjusted EBITDA, or roughly 10 times including C$35–45 million of anticipated annual cost synergies, and told investors to expect low-teens percentage EPS accretion in the first full year.⁶ The business — Link Simulation & Training, Doss Aviation, and AMI — closed into CAE USA in Tampa, Florida on July 2, 2021.³²

To fund the acquisition, CAE went to the equity market during the pandemic. It placed 22.4 million subscription receipts at C$31.25 each for gross proceeds of C$700 million, with Quebec's Caisse de dépôt et placement du Québec taking C$475 million and Singapore's GIC C$225 million.⁷⁸ That transaction accomplished two things simultaneously: it gave CAE the equity to close a transaction at a moment when few industrials could raise capital, and it installed CDPQ as a large, strategically motivated, locally powerful shareholder — a fact that would matter considerably four years later, when the board required rebuilding.

Reading the tape on management credibility

The pattern across all three deals is consistent enough to be diagnostic. Each was an adjacency justified by a capability narrative — "we understand simulation," "we understand the airline workflow," "we understand training" — rather than by a demonstrated ability to earn returns in that adjacency. Each was acquired at a full price near a cyclical or narrative peak. And in each case the balance sheet, not the operating business, absorbed the cost of the error.

Parent's public narrative through fiscal 2022 and 2023 maintained the two-segment framing while the defence business quietly deteriorated. The language available to investors in that window described execution challenges and supply-chain headwinds — categories that read as external and temporary. What was actually accumulating was a bounded, nameable, quantifiable liability: a specific set of fixed-price contracts, signed before the pandemic, whose costs were running significantly ahead of estimates.

Investors did not learn the scale of that liability until May 2024. When they did, it arrived all at once.

↑ Back to Top

V. The Detonation: Defence Implosion & the Reckoning (FY2024 – FY2025)

On May 21, 2024, six days before it was scheduled to report full-year results, CAE issued a press release that no company issues voluntarily. It disclosed a re-baselining of the Defence business, a set of impairment charges, accelerated risk recognition on legacy contracts, and — appended like an afterthought, though it was nothing of the sort — the appointment of a chief operating officer.⁹

Companies pre-announce when the number is so bad that letting it land inside a quarterly earnings release would look like burying it.

What actually went wrong

The mechanism was almost boringly simple, which is what makes it instructive. CAE had eight firm fixed-price contracts in its Defence segment, signed before the pandemic, without cost-escalation provisions.⁹ A firm fixed-price contract is exactly what it sounds like: the customer agrees to pay X, the contractor agrees to deliver Y, and every dollar of cost overrun comes out of the contractor's margin. In a stable-cost environment with a well-understood scope, that is a reasonable bargain — it is how you win competitive government work.

Then 2021 through 2023 happened. Skilled engineering labour repriced sharply. Electronic components went on multi-year lead times. Programmes that assumed a 2019 cost base were being executed on a 2023 one, with revenue frozen at the 2019 number. And critically, these were not simple hardware builds — the worst of them were programme-management and systems-integration jobs, where scope is negotiated continuously with a government customer and where CAE's genuine advantage in simulation physics conferred essentially no protection.

One important correction to the popular telling: only one of the eight contracts came with the L3Harris acquisition.⁹ The rest were CAE's own. The L3Harris deal did not cause the fixed-price problem so much as it enlarged the Defence business, raised the goodwill balance, and removed any margin for error.

The charge

CAE recorded a C$568.0 million non-cash goodwill impairment against Defence, a C$90.3 million unfavourable contract profit adjustment from accelerated risk recognition on the legacy contracts, and a C$35.7 million write-down of related intangibles — approximately C$693.7 million in total.⁹

Run through the income statement, fiscal 2024 became one of the worst years in CAE's public history. Full-year revenue rose to C$4,282.8 million, but the company posted an operating loss of C$185.4 million against C$466.0 million of operating income the year before, and a net loss of C$325.3 million.¹⁰ The fourth quarter alone carried an operating loss of C$533.0 million.¹⁰ Most damning of all, the Defence segment generated adjusted segment operating income of C$0.8 million for the entire fiscal year — a rounding error on nearly C$1.85 billion of revenue, which is to say a zero-margin business.¹⁰

The market's immediate reaction was oddly muted. Shares fell about 6% on May 22, 2024 to C$25.32.¹¹ That relative restraint is itself the tell: by mid-2024, expectations for CAE's Defence business had already fallen so far that a C$694 million charge was not a surprise in direction, only in magnitude.

The credibility damage, specifically

The goodwill impairment was, in accounting terms, management's own statement that the L3Harris military training business was worth materially less than what had been paid three years earlier. CAE had raised equity from long-term institutions at C$31.25 a share to fund that purchase while pointing toward meaningful Defence margin expansion. The write-down retired that promise.

Two features of the episode warrant particular scrutiny. First, the risk was bounded and knowable — eight named contracts — but was not quantified for investors until it was recognised. For roughly two years, investors could read about "execution challenges" without being able to size the exposure. Second, the external framing leaned heavily on inflation, supply chains, and labour availability. Those factors were real. They were also, by 2022, entirely visible to management. The decision to price fixed-price integration work without escalation clauses was a CAE decision, not a macro event.

To Parent's credit, the May 2024 disclosure did include specifics. The company said it had undertaken substantial renegotiations across the contracts to define remaining work, and that the original six-to-eight quarter retirement timeline for the legacy programmes remained unchanged even as risk recognition was pulled forward.⁹ That was a falsifiable claim with a date attached — roughly, completion by fiscal 2026 — which is exactly what investors should demand when a company is asking for patience.

The same release promoted Nick Leontidis, the architect of the Civil franchise, into a newly created chief operating officer role with oversight of both segments.⁹ The signal was unmistakable: Defence needed Civil's operating discipline.

The activist arrives

Eight months later, the pressure came from outside. In December 2024, Los Angeles-based Browning West disclosed a 4.3% stake and wrote to the board. Its argument was not complicated: CAE occupied an enviable market position and had badly underperformed anyway, with the shares down roughly 4% over a five-year period in which the S&P/TSX Composite rose about 43%.¹³ With Parent's departure already announced, Browning West's specific demand was that the board not move hastily on a successor and instead engage with the fund to recruit the best available leader.¹³

This was a subtle and, in retrospect, effective form of activism. Browning West did not demand a breakup or a special dividend. It demanded a seat at the single decision that would determine the next decade.

It got one. On February 13, 2025, CAE announced a board reconstruction effective the following day: four directors retired, including chair Alan MacGibbon, and four joined.¹⁴ Calin Rovinescu — who ran Air Canada from 2009 to 2021, and who described himself as "a longstanding airline customer and partner of CAE" — became chair.¹⁴ Peter Lee joined as Browning West's nominee, and Louis Têtu as CDPQ's.¹⁴ Lee was named co-chair of the CEO search committee.¹⁴ Rovinescu subsequently took on the executive chairman role, working directly with the incoming chief executive on strategy.

The reconstruction happened fast and without a protracted proxy fight — and it happened because CDPQ, sitting on a stake acquired in the 2021 financing, had every incentive to fix governance rather than contest it. The 2021 capital raise ended up buying CAE more than cash.

Was the fiscal 2025 recovery real?

The following year looked like vindication. Fiscal 2025 revenue rose 10% to C$4,707.9 million, operating income swung from a loss to C$729.2 million, and adjusted EPS climbed 39% to C$1.21.¹⁶ Adjusted backlog jumped 65% to C$20.14 billion on C$7.70 billion of order intake.¹⁶ Free cash flow reached C$813.9 million at a 211% cash conversion rate, and net debt to adjusted EBITDA fell to 2.77 times.¹⁶

Some of that was genuine and some was arithmetic. The operating-income swing is largely the absence of the prior year's charges. The cash flow, however, was real: a 211% conversion rate means CAE collected substantially more cash than it booked as adjusted earnings, which is what happens when long-duration programmes reach milestone payments and when working capital absorbed on troubled contracts finally releases. Backlog growth was real too, driven by a defence order intake that more than doubled.¹⁶

What fiscal 2025 did not prove was that the underlying operating problem had been solved. Defence's adjusted margin recovered only to 7.5% — better than zero, but nowhere near the mid-teens the segment had once been promised.¹⁶ And in Parent's final full-year commentary he described "an exceptional fourth quarter, capping a strong year across all key financial and operational metrics."¹⁶ Twelve months later, the Civil segment he was describing would miss its own guidance badly.

Which brings the story to the engine room — the business that has to work for any of the rest to matter.

↑ Back to Top

VI. The Core Engine: Civil Aviation Training

Picture the least glamorous piece of infrastructure in commercial aviation: a windowless building near an airport perimeter road, containing eight to twelve boxes on hydraulic legs, running sixteen to twenty hours a day, seven days a week, with crews arriving at 4 a.m. because that is the slot the scheduler could find.

That building is the asset. Everything else — the pilot shortage headlines, the eVTOL optionality, the software ambitions — is commentary on whether the building is full.

What the business actually is

CAE's Civil segment does four things. It designs and builds full-flight simulators — the highest-fidelity category being Level D, the modern descendant of that 1982 Phase III approval and the only class qualified for full type-rating without the actual aircraft. It operates a global network of training centres where airline and business-jet crews arrive for initial type ratings, recurrent proficiency checks, and command upgrades. It runs ab initio academies producing licensed pilots from scratch. And, until a strategic review concludes, it sells airline operations software under the Flightscape brand.

The manufacturing side explains the barrier to entry more plainly than any market-share statistic. To build a Level D simulator for a given aircraft type, a manufacturer needs the aircraft maker's proprietary aerodynamic and systems data — the mathematical description of how that specific airframe behaves in every flight regime, including regimes no test pilot would deliberately enter. That data is licensed, not observable. The cockpit must then be physically replicated and driven by a motion platform and visual system convincing enough to fool a trained pilot's body, after which the result must pass regulatory scrutiny — separately, in each jurisdiction — before it qualifies as a legal substitute for the aeroplane. That process takes many months per type per authority, and it must be repeated across every market an operator serves.

A new entrant cannot simply decide to compete. It would need OEM data licences it cannot obtain, regulatory qualifications it cannot shortcut, and — to matter to a global airline — a network of buildings on four continents. That combination constitutes the closest thing to a genuine structural barrier in this corner of industrials.

The economics, and where they broke

Civil revenue in fiscal 2026 was essentially flat, rising 1% to C$2,741.6 million.² But adjusted segment operating income fell to C$510.5 million from C$581.5 million, and the margin compressed to 18.6% from 21.5%.²¹⁶ Training-centre utilisation — the share of available simulator hours actually sold — averaged 70% for the year, against 74% in fiscal 2025.²¹⁶ CAE delivered 52 full-flight simulators in fiscal 2026, down from 61.²¹⁶

That cluster of numbers tells a coherent story, and it is not a demand story. It is an operating-leverage story. A training centre is a fixed-cost asset: the building, the simulators, instructor payroll, and maintenance are roughly constant whether the boxes run at 70% or 78%. Utilisation is therefore the single most powerful lever on Civil margin. Four percentage points of utilisation is not a rounding error; it accounts for most of a 290-basis-point margin decline.

Two causes were identified. The first was geopolitical: conflict in the Middle East reduced training activity from carriers in a region that had been one of CAE's growth engines. On the first-quarter fiscal 2027 call, chief financial officer Ryan McLeod attributed roughly two-thirds of the Civil margin decline to that regional disruption and described rerouting training to other facilities at temporary incremental cost.²¹ The second was self-inflicted and more instructive: CAE had built too much capacity.

The admission: a network that was overbuilt

Alongside its fiscal 2026 results on May 21, 2026, CAE announced a transformation plan whose Civil component amounts to a public acknowledgement that the post-pandemic growth assumptions were wrong. The company will remove approximately 10% of its commercial full-flight simulator fleet, relocate and optimise more than a dozen additional simulators, and consolidate real estate significantly.² On the first-quarter call, management added specifics: roughly 25 commercial simulators retired in total, four to six training centres closed, and global square footage reduced by 1.7 million square feet, or about 17%.²¹

Chief executive Matthew Bromberg stated that customer attrition from the closures would be "less than 1% of our Civil revenue."²¹ On the fiscal 2026 call he had been slightly more qualified: "We hope to retain every single one of them. There will be some attrition, which is why we factored it into our 2027 long-term guidance."²²

Investors should treat the sub-1% attrition figure as a forecast, not a fact. It is precisely the number the switching-cost thesis predicts, which makes it both plausible and unverifiable until it materialises. If it holds, the rationalisation converts directly into margin: the same training hours delivered from fewer, fuller buildings. If it does not — if a closed centre gives a regional competitor the opening an airline needed to re-tender — the switching-cost argument is weaker than the company's framing suggests.

Myth versus reality

Three consensus claims about CAE's Civil business warrant scrutiny.

Myth: CAE holds roughly 56% of the global market and trains most of the world's pilots. The widely cited 56% figure refers to CAE's share of the installed base of flight simulators and traces to company disclosures from more than a decade ago. It is not a metric CAE reports in current results. What the company does disclose — deliveries, utilisation, backlog — is more useful and less flattering to the monopoly framing. Independent market surveys name CAE alongside FlightSafety International, Thales, Airbus, and TRU Simulation as the leading players.³¹ CAE is unambiguously the largest; "monopoly" is not the right characterisation, and a meaningful share of the world's pilots are still trained on simulators their own airlines own.

Myth: mandated demand means margins cannot fall. They just did — by nearly three percentage points over two years, with the mandate fully in force.²¹⁶ Regulation guarantees that training happens. It guarantees nothing about where, at what price, or at what utilisation rate. Regulatory demand protects the volume of the industry; it does not protect the profitability of any particular supplier.

Myth: the pilot shortage is an automatic growth engine. CAE's own 2025 Aviation Talent Forecast projected 1.465 million new civil aviation professionals needed between 2025 and 2034, including 300,000 pilots — 267,000 commercial and 33,000 in business aviation — alongside 416,000 maintenance technicians and 678,000 cabin crew, with Asia-Pacific as the largest source of demand.¹⁷ The demographics are real: retirements are contractual, not discretionary, and every delivered aircraft needs type-rated crew. But fiscal 2026 Civil revenue grew just 1% against that backdrop, and Civil's twelve-month book-to-sales ratio was 0.96 — below one, meaning the segment consumed backlog faster than it replenished it.² A structural tailwind and a weak year are not contradictory; they operate on different clocks. Aircraft delivery delays at Boeing and Airbus push crew-training demand out, and a pilot shortage does not benefit CAE until an airline actually schedules the session.

The optionality nobody should size too large

On January 6, 2026, Joby Aviation announced it had taken delivery of the first of two CAE-built simulators for its electric air taxi — a Level 7 flight training device with a 300-by-130-degree field of view, to be followed by a Level C full-flight simulator, together supporting up to 250 pilots a year.²⁷ Joby had begun working with CAE in 2022 to have training devices qualified ahead of entry into service.²⁷

This is the eVTOL thesis in miniature. Every new aircraft category creates a regulatory training requirement from scratch, and the incumbent with the physics models, the visual systems, and the regulator relationships tends to win the first contract almost by default. If electric air taxis become a meaningful industry in the 2030s, CAE is structurally positioned. That is a call option with a long expiry and an unknown strike. It is not a fiscal 2027 earnings driver, and any valuation that leans on it is leaning on a story.

What matters far more, near term, is whether utilisation climbs. It moved to 72.2% in the first quarter of fiscal 2027 from 68.8% a year earlier — a step in the right direction — while margin still fell to 16.5% from 20.2%, because the transformation costs and network transition expenses land before the savings do.¹ Management guided Civil revenue flat to slightly down for fiscal 2027.²

That is an unusual position for the company's anchor business: structurally advantaged, temporarily underperforming its own history, and being deliberately shrunk to recover profitability. The other half of the company is moving in the opposite direction.

↑ Back to Top

VII. The Other Half: Defence & Security

In May 2024, the Government of Canada awarded a C$11.2 billion, twenty-five-year contract for the Future Aircrew Training programme to SkyAlyne, a joint venture between CAE and Kelowna-based KF Aerospace, after selecting the consortium as preferred bidder the previous July.²⁰ The programme will supply the Royal Canadian Air Force with classroom instruction, simulator and flight training, and on-site support for the aircraft fleets it is acquiring. CAE signed a separate twenty-five-year subcontract with SkyAlyne worth roughly C$1.7 billion to build and deliver the simulators and training devices.²⁰

The timing is almost too neat. In the same month CAE disclosed a C$694 million defence write-down, it locked in one of the longest-duration training contracts in Canadian history. Both facts describe the same segment — which is why understanding the Defence business requires separating two quite different operations that share a P&L line.

Two businesses wearing one name

The first is what CAE has demonstrated an ability to execute: building and operating training systems for military customers on long-duration service arrangements, where the customer is a sovereign government, funding is appropriated, and contract terms allow costs to pass through. Aircrew training programmes, mission rehearsal, sustainment. Economically, these resemble the Civil training business with a different customer class and longer contract durations.

The second is development-heavy programme management sold on firm fixed price — and it is the one that produced the write-downs described in the preceding section. CAE's simulation expertise in physics libraries, instructor tooling, and visual systems provides meaningful protection in the first type of contract. It provides essentially no protection in the second, where cost is driven by scope negotiation and bespoke engineering rather than fidelity to an aerodynamic model.

Bromberg has stated that the strategy is to exit the second category. On the first-quarter fiscal 2027 call, he described the target as low-teens margins achieved by minimising non-recurring engineering exposure and shifting the revenue mix toward services — using OEM partnerships to spread development costs across multiple NATO customers rather than absorbing bespoke costs for each sovereign buyer.²¹ The logic is straightforward: if eight countries purchase the same platform, developing the training system once and amortising it across eight programmes converts a fixed-price liability into a scale business. It requires being embedded with the aircraft or ship manufacturer before the training requirement goes out to tender.

The evidence that this approach is being pursued, rather than merely described, is accumulating. On March 4, 2026, CAE signed a teaming agreement with TKMS to support its bid for the Canadian Patrol Submarine Project, pairing the German shipbuilder's submarine design with CAE's simulation-based training and mission-system support for the Royal Canadian Navy.²⁸ On May 29, 2026, the two extended that into a broader framework for naval training, simulation, and operational readiness beyond the Canadian programme.²⁹ On the first-quarter call, Stifel's Daryl Young asked directly whether CAE had prior experience working with TKMS and what product-development risk the partnership carried — a fair challenge given the company's record on development-heavy contracts, and one management did not fully resolve.²¹

The numbers, and what they suggest

Defence revenue grew 9% in fiscal 2026 to C$2,172.4 million, and adjusted segment operating income rose to C$200.2 million from C$150.5 million, lifting the margin to 9.2% from 7.5%.²¹⁶ The fourth quarter reached 10.2% — the first time in years the segment had crossed double digits.² The first quarter of fiscal 2027 delivered 8.3% revenue growth at a 9.5% margin, up modestly year over year.¹ Defence adjusted backlog stood at C$10.7 billion at the end of that quarter, and the segment's twelve-month book-to-sales ratio was 1.10, meaning it continued to add contracted work faster than it consumed it.¹

The honest reading: the trajectory is real; the level is unremarkable. A defence training and sustainment business earning 9–10% margins is a serviceable industrial-services business — neither distressed nor distinguished. The gap between that and the low-teens management is targeting is the entire turnaround thesis. Two consecutive quarters above 12% would constitute the first hard evidence; anything short of that is directionally encouraging but analytically inconclusive.

The tailwind, and its limits

The macro environment is more favourable than it has been in a generation. Bromberg told The Globe and Mail in December 2025 that defence spending was growing at 7–8% annually rather than the historical 2–3%, that training represents "fifteen cents of every defence dollar that's spent," and that sovereign customers are reliable counterparties.²³ Canada's own rearmament — the F-35 acquisition, naval recapitalisation, NORAD modernisation — gives CAE an unusual home-market advantage in a company that has historically competed hardest in the United States and Europe.

Two cautions apply. First, appropriated defence budgets are political, and political majorities change; European rearmament cycles have been confidently projected before. Second, growth in the total addressable market does not automatically accrue to CAE. Competitors in defence training include L3Harris in simulation products, Cubic, Rheinmetall's expanding European footprint, BAE Systems Electronic Systems, Leonardo DRS, and a long tail of OEM-aligned integrators. CAE's differentiator is that it is simulation-first — most rivals are defence primes for whom simulation is one capability among many, so CAE's civil-derived physics modelling and visual systems carry into military programmes at lower incremental cost. That is a genuine structural advantage. It is also precisely the advantage that failed to protect the eight legacy fixed-price contracts, because those contracts were not won on simulation quality.

The open question

Management committed in May 2024 that the legacy fixed-price programmes would retire on the original six-to-eight quarter schedule.⁹ Two years on, the margin trajectory is consistent with that claim, and by fiscal 2026 the company described the headwind as diminishing. But no unambiguous public statement has confirmed that all eight contracts are fully closed. Investors reading fiscal 2027 transcripts should watch for that specific language in Q&A rather than in prepared remarks — and should note that a company which discovered eight problematic contracts without disclosing their scale for roughly two years has demonstrated that its bid-review and disclosure processes can fail at scale.

That is the operational inheritance. The people who now own it are almost entirely new.

↑ Back to Top

VIII. Current Management & Capital Allocation

Matthew Bromberg is the first American to run CAE in the company's history.¹⁵ That fact is less interesting than the resume behind it.

He read physics at the University of California, Berkeley, then took both an MBA and a master's in mechanical engineering at MIT. He served as a submarine officer in the US Navy — a background that shows up in how he talks about operations, which is to say procedurally. He ran commercial aftermarket operations at Pratt & Whitney from 2013 to 2017, then military engines at Raytheon Technologies from 2017 to 2022, then global operations at Northrop Grumman.¹⁵ Aftermarket services, military programmes, and supply-chain restructuring at scale: the three things CAE most needed.

He joined on June 16, 2025 as incoming chief executive, worked alongside Parent through the transition, and took the role on August 13, 2025.¹⁵

The board that hired him was itself new. Rovinescu, who steered Air Canada through its own long transformation, sits as executive chairman working directly on strategy. Sophie Brochu serves as lead independent director.¹⁵ The finance function turned over completely: Sonya Branco, CFO since 2016, announced on July 25, 2024 that she would step down at the end of August; chief accounting officer Constantino Malatesta served as interim; and on January 19, 2026 CAE named Ryan McLeod, a CPA who had been CFO of ATS Corporation since 2020, effective February 23, 2026.¹⁸¹⁹ At ATS, McLeod oversaw revenue growth from C$1.4 billion to roughly C$3.0 billion, led a US initial public offering, and executed eighteen acquisitions.¹⁹

Eighteen months without a permanent CFO, at a company whose central problem was contract risk recognition, is a governance fact worth stating plainly. It is not disqualifying — Malatesta had run the controller's office — but finance leadership continuity matters most precisely when accounting judgment is under scrutiny.

The plan

Bromberg's transformation programme, unveiled with fiscal 2026 results, rests on three stated principles: aligning the portfolio to areas of competitive advantage, strengthening capital discipline by prioritising higher-return investments and optimising the commercial training network, and improving operational performance through better integration of people, processes and technology.²

The mechanics: C$125–150 million of annual run-rate savings targeted by fiscal 2030, at a total transformation cost of C$200–250 million, of which roughly C$100 million is non-cash.² On the first-quarter call, management disclosed C$133 million already spent, of which C$71 million was non-cash, and broke the savings into roughly 50% labour productivity, 30% reduced square footage, and 20% operational improvements.²¹ The fiscal 2030 destination is adjusted segment operating income of C$950 million to C$1 billion, cumulative cash conversion of 100% over four years, leverage around 2.5 times, and mid-single-digit organic revenue growth — with more than half the improvement expected to come from internal initiatives rather than market growth.²

Analysts did not simply accept this. On the fiscal 2026 call, they pressed on whether the roughly C$975 million midpoint adequately reflected mid-single-digit organic growth; McLeod defended the bridge by pointing to the roughly 8% of revenue slated for divestiture and the discontinuation of certain government funding, both of which reduce the long-term baseline.²² That is a substantive answer rather than an evasion, and it also quietly reveals that the headline target is being measured against a smaller company.

The capital allocation scorecard

On the credit side: CAE survived the pandemic without permanently impairing the Civil franchise. It deleveraged hard, from post-acquisition peaks toward 2.29 times net debt to adjusted EBITDA at fiscal 2026 year-end and 2.27 times at the end of the first quarter of fiscal 2027.²¹ S&P Global Ratings revised its outlook on CAE to stable from negative on March 17, 2026, having moved it to negative in November 2024 — a small but real external validation of the balance-sheet repair.³⁰ The company has restarted buybacks, repurchasing 85,100 shares at a weighted average of C$36.35 in fiscal 2026 and stepping that up materially to 1.1 million shares at an average of C$35.26, or C$39.0 million, in the first quarter of fiscal 2027.²¹

On the debit side: roughly C$694 million of value acknowledged as destroyed on the defence goodwill and contract charges; a software platform bought for US$392.5 million now on the block at a price that reporting suggests will not recoup the investment; a healthcare division built over a decade and sold for C$311 million into a deleveraging need; and equity issued at C$31.25 in 2021 that has spent most of the intervening five years below that level. No dividend has been reinstated; on the first-quarter call Bromberg said buybacks would stay "measured, disciplined and transparent," with funding the transformation taking precedence.²¹

The metric that ties it together

Adjusted return on invested capital was 7.6% at the fourth quarter of fiscal 2026, against an adjusted return on capital employed of 7.2% in fiscal 2025.²¹⁶ For a capital-intensive industrial carrying more than C$2.6 billion of net debt, that is at or below a reasonable estimate of the cost of capital.² This is the single most important sentence in the CAE bear case, and it does not depend on any forecast: a business with a supposedly formidable moat has not been earning an economic return on the capital deployed into it.

The moat is in the Civil training network. The capital went substantially elsewhere. Whether Bromberg's transformation closes that gap is the whole question — and it will be answered in returns, not in slideware.

To judge whether it can be closed, it helps to look at the industry structure CAE actually operates within.

↑ Back to Top

IX. Industry Structure, Competitive Moat, and Investor Framework

Strip away the narrative and ask the war-game question: if a well-capitalised competitor decided tomorrow to take CAE's civil training business, what exactly would it do?

Porter's five forces, applied honestly

New entrants. This is where the moat is real. A challenger would need OEM data licences for each aircraft type, regulatory qualification in each jurisdiction it wished to serve, a physical network near airline crew bases on multiple continents, and instructors qualified on each type. The capital required to replicate CAE's civil network would rival CAE's entire enterprise value, and the qualification timeline would run years before the first billable training hour. No credible new entrant had appeared in decades. Threat: very low.

Buyer power. Airlines are sophisticated, cost-conscious, and increasingly consolidated. They negotiate hard on price. On the first-quarter fiscal 2027 call, National Bank's Cameron Doerksen pressed Bromberg specifically on pricing power at contract renewal, given the squeeze airlines themselves face — a question Bromberg answered by saying CAE was "working cautiously" on repricing while pursuing disciplined aftermarket pricing first.²¹ "Working cautiously" is not the language of a monopolist. Buyers cannot easily leave, but they can grind. Threat: moderate.

Supplier power. The critical supplier is the aircraft manufacturer, which controls the aerodynamic and systems data required to build a simulator for any new type. Boeing and Airbus each operate their own training organisations and could, in principle, tighten data terms to favour their own businesses. In practice they have not — airlines prefer independent multi-type training, and OEMs would rather sell aircraft than run training centres. But this is a structural dependency CAE does not own and cannot fully hedge. Threat: moderate to high, and underappreciated.

Substitutes. Virtual and augmented reality training is advancing quickly; Bromberg has named AR, VR, and AI integration as priorities.²³ Today, none of it can legally replace a qualified Level D full-flight simulator for type rating in major jurisdictions. The more relevant scenario is not that VR displaces the simulator outright — it is that regulators eventually allow a portion of recurrent training to migrate to cheaper devices, compressing the highest-margin hours at the margin. That is a slow-moving but genuinely material long-term risk that would show up first in regulatory consultation papers rather than in earnings. Threat: low near term, but the trajectory warrants watching.

Rivalry. FlightSafety International, owned by Berkshire Hathaway, is the closest peer in scale and dominates business-aviation training in the United States. Thales competes in simulator manufacturing, particularly in Europe. Boeing's training arm is simultaneously competitor and data counterparty. Airbus and TRU round out the named leaders in independent industry surveys.³¹ Rivalry is real on specific aircraft types and specific geographies, but no competitor matches CAE's breadth. Threat: moderate.

Seven Powers, and which ones actually hold

Of Hamilton Helmer's seven sources of durable competitive advantage, three are convincingly present at CAE; the rest are weaker than the standard telling implies.

Scale economies is the strongest. Bromberg described a portfolio of roughly 360 simulators and a library of 220 platforms — more aircraft types than any competitor globally.²³ Fixed cost per training hour falls as density rises; a broader type library wins multi-fleet airline contracts a narrower rival cannot bid. This is the durable foundation.

Switching costs are genuine, for the regulatory-plumbing reasons detailed earlier in this story. The cleaner test is arriving in real time: if CAE closes four to six centres and truly loses under 1% of Civil revenue in the process, switching costs are as strong as the company's framing suggests. If attrition runs meaningfully higher, the market has been overpricing this power for years.

Counter-positioning is present in a specific, bounded form. CAE absorbs simulator capital that airlines are structurally reluctant to carry. A rival choosing to displace CAE would have to accept the same balance-sheet burden — which is precisely why airlines, not new entrants, are the most plausible competitive threat, and why they have mostly chosen not to be one.

Process power is moderate. Cornered resource is weaker than commonly assumed, because the most valuable input — OEM aerodynamic and systems data — is licensed rather than owned. Network economies in the strict demand-side sense are largely absent; what CAE has is an internal knowledge and data flywheel, which confers a cost advantage rather than a network effect. Branding carries genuine weight with airline customers, less so in defence procurement, where programme-of-record status and past-performance ratings matter more than name recognition.

The synthesis

Civil aviation training is a strong, durable franchise built on scale and switching costs, with a real and rarely discussed dependency on aircraft manufacturers. Defence is a competent, cyclically favoured services business with no comparable moat — and the fixed-price episode demonstrated that structural advantage in simulation physics does not protect against self-inflicted contracting risk.

The uncomfortable conclusion for anyone who admires the business: moat quality and capital-allocation quality are separate variables. CAE has demonstrated the first for four decades. It has not yet demonstrated the second.

↑ Back to Top

X. The Investment Spine: Why Win / Why Not

Every long thesis on CAE reduces to a single sentence: a structurally advantaged business is being fixed by new management, and the market has not yet paid for the fix. Every short thesis reduces to an equally short one: this management team is asking for four years and a smaller company in exchange for targets that a previous management team also promised.

Both sentences are defensible. The work is in the evidence.

Why CAE wins from here

The demand side is not in dispute. Recurrent proficiency training is legally required, retirements are demographic rather than discretionary, and CAE's own decade forecast points to 300,000 new pilots needed globally through 2034.¹⁷ Whatever happens to any single year's airline capacity plans, the installed base of pilots must be requalified on a rolling schedule for as long as commercial aviation exists.

The defence trajectory is improving on evidence rather than assertion. The segment moved from a zero-margin year in fiscal 2024 to 7.5%, then 9.2%, with a fourth-quarter print above 10%.¹⁰¹⁶² Backlog is being replenished faster than it is consumed, and the SkyAlyne subcontract provides a twenty-five-year revenue spine that is structurally different in kind from the fixed-price work that caused the damage.¹²⁰

The balance sheet is genuinely repaired. Leverage has come down to the low 2s, free cash flow inflected, the credit outlook was restored to stable, and buybacks have resumed at prices well below the January 2026 highs.²¹³⁰ A company that in 2024 looked like it might need to choose between deleveraging and investing no longer faces that choice.

And the governance change is not cosmetic. A board reconstructed with an activist nominee, a pension-fund nominee, and a former airline chief executive as chair produced a CEO hire with exactly the operational profile the situation called for. Incentives for fiscal 2027 onward are anchored on cash flow, margin, returns and earnings growth rather than growth-at-any-price metrics — a structure that at least points management at the right variable.

Why the case could break

Start with the guidance record, because it is the cleanest available test of credibility and it does not require trusting anyone's judgment.

In May 2024, alongside the re-baselining, CAE guided fiscal 2025 to roughly a 23% Civil adjusted segment operating margin and 6–7% in Defence.¹⁰ Civil delivered 21.5% — a miss. Defence delivered 7.5% — a beat.¹⁶ In May 2025, CAE guided fiscal 2026 to mid- to high-single-digit Civil adjusted operating income growth with modest margin expansion, low-double-digit Defence operating income growth at an 8–8.5% margin, and free cash flow conversion around 150%.¹⁶ Civil operating income instead fell roughly 12% with margin down nearly three points; Defence operating income grew about 33% at a 9.2% margin; and cash conversion came in at 123% on an updated definition, with free cash flow of C$473.8 million against the prior year's C$813.9 million.²¹⁶

The pattern is an inversion of what anyone would have predicted. The segment management understood least — Defence, freshly humiliated — beat its guidance two years running. The crown jewel, the business the company had operated for four decades, missed badly twice. Some of that is the Middle East, which is not forecastable. But guiding Civil to margin expansion in May 2025 while sitting on a network the company would describe as overbuilt twelve months later is a planning failure, not a geopolitical one.

That is the core of the activist stress test, and it goes further. A skeptical investor would ask why an "overbuilt" network was not identified before a new chief executive arrived from outside; why the finance function went eighteen months without a permanent leader during exactly the period when contract-risk judgment was under scrutiny; why a US$392.5 million software acquisition required four years to be labelled non-core; why roughly 8% of revenue is only now being classified for divestiture; and why returns on invested capital, at 7.6%, remain below any plausible hurdle rate for a business whose competitive position is described in monopolistic terms.²²⁴²²

The specific forward risks follow from that. Execution risk is the largest: a four-year transformation with C$200–250 million of upfront cost, C$133 million of it already spent, delivering savings that arrive mostly at the back end.²²¹ Multi-year cost programmes routinely slip on timing even when they eventually work, and fiscal 2027 is explicitly a year in which costs land ahead of benefits — which is precisely why the shares fell 13.5% when the plan was disclosed.²⁶

Customer-attrition risk sits underneath the network closures. Concentration risk sits underneath the Middle East exposure, which management quantified as roughly two-thirds of the Civil margin decline — a striking dependence on one region for a company that markets itself as globally diversified.²¹ OEM dependency remains an unhedged structural exposure. And the legacy contracts, while apparently retiring, have never been publicly declared complete as a set.

There is also an accounting judgment worth naming. Goodwill and intangible balances arising from acquisitions require annual impairment testing, and CAE has already demonstrated once that its estimates of defence programme profitability could be materially wrong. Should the Flightscape process conclude below carrying value, a further — though far smaller — write-down would follow. Reporting suggests the sale is unlikely to recover the original outlay.²⁵ Against that, on the first-quarter call Bromberg said the review was "well underway" and generating strong interest, while declining to commit to timing; National Bank's Doerksen has argued the asset could still be attractive to a buyer given its incremental margin potential at scale, and Scotiabank's Konark Gupta estimated a full sale could be accretive by C$0.75 to C$2.50 per share.²¹²⁵ Sell-side accretion estimates are not evidence of value; they are evidence that the market has not decided what this asset is worth.

The three numbers that matter

Everything above compresses into three metrics an investor can track without a model.

Civil training centre utilisation. It ran 70% for fiscal 2026 and 72.2% in the first quarter of fiscal 2027, against a management target above 75%.²¹ This is the most direct profit lever in the most valuable business, and it is the number that will reveal whether closing centres concentrates demand or simply loses it.

Defence adjusted segment operating margin. It stood at 9.5% in the first quarter of fiscal 2027 against a low-teens ambition.¹²¹ Sustained readings above 12% across consecutive quarters would confirm the mix shift away from development-heavy fixed-price work. Any relapse would compound the credibility damage rather than merely repeat it.

Net debt to adjusted EBITDA. At 2.27 times, with a fiscal 2030 target around 2.5 times, this gates everything else — buyback capacity, dividend reinstatement, and the ability to invest through a downturn.¹² It is also the metric on which this management team has most consistently delivered.

Three numbers. One for the moat, one for the turnaround, one for the balance sheet. If all three move in the right direction across fiscal 2027 and 2028, the story management is telling will have been true. If utilisation stalls while defence margin plateaus near 10%, it will not.

↑ Back to Top

XI. Lessons and Lasting Framework

Great businesses teach general lessons only when you separate what was structural from what was chosen. CAE offers an unusually clean separation, because the structural part has been stable for forty years while the choices swung wildly.

The compounding event was a business-model change, not a technology breakthrough. CAE's simulators were excellent in 1985 and the company was still a mediocre investment. What changed everything was the decision to stop selling boxes and start renting hours in buildings it owned — converting lumpy capital-goods revenue into recurring, regulator-mandated consumption. Investors hunting for this pattern should look for the same three ingredients: a mandated consumption stream, an asset the customer would rather not own, and a switching cost created by regulation rather than by contract.

Acquisition logic built on capability, not on returns, tends to fail. All three of CAE's problematic deals were justified by what the company knew how to do. None was justified by demonstrated returns in the target market. Healthcare shared a technology and nothing else. AirCentre shared a customer and not an operating model. L3Harris shared a customer and a domain, but included a contracting model CAE had no advantage in. Capability adjacency is the most seductive and least reliable acquisition thesis in industrials.

Fixed-price integration work is a different business from product manufacturing. Building the hundredth simulator of a type is a scale exercise with known costs. Integrating bespoke mission systems for a sovereign customer is a negotiation that runs for years. The first benefits from CAE's advantages; the second does not. The company paid roughly C$694 million to learn where that boundary sits.⁹ Bromberg's stated fix — pushing development costs into multi-country OEM programmes — is a direct response to exactly this lesson, and it is the right structural answer.²¹

Moats protect against competitors, not against yourself. No rival took share from CAE's civil network during the drawdown. The value destruction was entirely internal. When a company with a genuine structural advantage underperforms for five years, the explanation is almost never competitive; it is almost always capital allocation.

Mandated, recurring demand is the best foundation an investor can have — and it is not a substitute for operating discipline. Regulation guarantees that the training happens. It does not guarantee the price, the utilisation, or the return on the capital deployed to serve it. CAE's Civil margin fell three points while every word of the regulatory mandate remained in force. Investors should treat regulatory demand as a floor under the industry, not as a floor under any one company's profitability.

The final lesson is about timing. CAE's story turned not when the business improved but when the governance changed — a board reconstruction in February 2025 that preceded any operational inflection. For long-term investors, that sequencing is the point: in businesses with durable structural advantages, the variable that most often changes the outcome is who is deciding where the money goes.

↑ Back to Top

XII. Epilogue: The Bromberg Test

A year into the job, Matthew Bromberg has done something that new chief executives of underperforming companies almost never do voluntarily: he has told investors that the year ahead will look worse.

Fiscal 2027 guidance calls for low-single-digit revenue growth, an adjusted segment operating margin of 14.6% to 15.1%, adjusted earnings per share of C$1.21 to C$1.28, and cash conversion of 85% to 95% — with Civil flat to slightly down and Defence up mid-single digits.² Those numbers were reaffirmed after the first quarter.¹ Bromberg has called fiscal 2027 a reset year and described the work ahead in unadorned terms: "This is hard work. We know how to do these, can do it, we have done it before."²²

There is a version of that sentence that is reassuring and a version that is a warning. Northrop Grumman, Raytheon and Pratt & Whitney are large, process-heavy organisations where operational transformation is an established discipline. CAE is a smaller company where the most valuable asset is a network of relationships with airlines that hate disruption. The skills transfer is not automatic.

What the next two years will settle:

The commercial network surgery is the near-term test. Closing four to six centres, retiring roughly 25 simulators and shedding 1.7 million square feet is the largest deliberate contraction in CAE's civil history.²¹ It is being executed while the segment is already margin-impaired and while a regional conflict suppresses one of its growth markets. Success looks like utilisation above 75% with revenue roughly held; failure looks like utilisation improving because the denominator shrank while customers quietly went elsewhere.

The portfolio reset is the medium-term test. Roughly 8% of revenue is earmarked for divestiture, with Flightscape the largest piece.²²²⁵ The outcome — sale price, structure, use of proceeds — will be the first hard read on whether the new team's capital discipline differs from the old team's, since this is the first significant capital decision entirely its own.

The defence mix shift is the structural test. Embedding with OEMs early, spreading development cost across NATO buyers, and refusing bespoke fixed-price integration is a coherent strategy. The TKMS agreements are the first substantial proof points, but neither has yet converted into a won programme.²⁸²⁹ Whether that model produces low-teens margins will not be knowable before fiscal 2028.

Three demand pools sit underneath all of it, and none requires an acquisition to access: a demographic pilot shortage, a rearmament cycle across NATO and Canada, and an electric-aircraft training market forming from zero. That is an unusually rich opportunity set for a company trading well below its January highs.

The tension for investors is that CAE has always had a rich opportunity set. What it has lacked, for most of the past decade, is the discipline to convert opportunity into returns on capital. Bromberg's plan addresses that directly and explicitly. It also does not deliver its first meaningful evidence until fiscal 2028, and asks shareholders to sit through a year in which reported results get worse before the savings arrive.

The Civil flywheel is real; the four decades of evidence are not seriously contestable. The open question is whether the generation of management that follows Marc Parent can stop damaging it.

↑ Back to Top

XIII. Outro

Three questions run through the preceding story. How do you train a pilot for an emergency that must never happen in service? How does a company build a near-monopoly on that answer without most of the world noticing? And how does it come close to squandering that position — not to a competitor, but through its own capital allocation?

CAE answered the first in 1952 with a contract that nobody at the time read as strategic. The second took another three decades to work out, arriving at the insight that owning the buildings was a more durable business than selling the machines inside them. The third unfolded between 2021 and 2024, as the balance sheet was stretched across adjacencies that shared a capability with the core business but not its economics.

The result is a company that sits at an unusual inflection. Almost every passenger who has boarded a commercial aircraft in the past generation has been flown by a crew that trained in one of CAE's simulators. Almost none of them know the name. That anonymity is the hallmark of infrastructure — and infrastructure businesses are, in the long run, judged not on the quality of the underlying asset but on what the owners choose to do with the cash it generates.

For the better part of a decade, those choices were poor. Whether the generation now in the chair makes different ones is the question the next two fiscal years will begin to answer.

CAE

Table of Contents

CAE Inc.: The Invisible Infrastructure of the Skies

I. Cold Open & Episode Roadmap

II. Founding & Cold War Origins (1947 – mid-1970s)

III. The Services Pivot: From Hardware Maker to Training Ecosystem (1980s – 2000s)

IV. The Marc Parent Era: Ambition and Overreach (2009 – 2025)

V. The Detonation: Defence Implosion & the Reckoning (FY2024 – FY2025)

VI. The Core Engine: Civil Aviation Training

VII. The Other Half: Defence & Security

VIII. Current Management & Capital Allocation

IX. Industry Structure, Competitive Moat, and Investor Framework

X. The Investment Spine: Why Win / Why Not

XI. Lessons and Lasting Framework

XII. Epilogue: The Bromberg Test

XIII. Outro

References