BlackBerry

IThe business

BlackBerry today is a software company with three parts. QNX sells the operating system, hypervisor and safety software that runs inside vehicles and a smaller set of medical, robotic and industrial machines. Secure Communications sells encrypted voice and messaging to governments (SecuSUITE), a crisis-alert system (AtHoc) and a device-management product (UEM). Licensing holds about 6,100 patents and collects what it can from them.³ The company was founded in 1984, has its head office in Waterloo, Ontario, employs 1,749 people, 57% of them in Canada, and its shares trade in New York and Toronto under the symbol BB.³

That description would not have fitted it twelve years ago. In the fiscal year to 3 March 2012 BlackBerry had revenue of $18.4B and shipped 49.0 million handhelds. A year later revenue was $11.1B and shipments 28.1 million.⁶ By fiscal 2017 revenue was $1.3B, hardware had shrunk to $374M, the net loss was $1.2B, and the company had handed the design and manufacture of its phones to partners.¹² The shrinking did not stop there. Revenue in fiscal 2026 was $549M, which is 3% of the 2012 figure. Anyone using the word “comeback” has to start from that number: the company that is coming back is about a thirty-third the size of the one that fell.

Look at the table the way an owner would. Three of the sales were hard-won cash and one was a loss. The patents brought in $170M at closing and $30M later, with royalties that depend on a third party’s success and that the company itself will not forecast.¹⁴³ The Cylance purchase cost about $1.4B in 2019 and was sold six years later for about $118M in cash plus shares in Arctic Wolf that sit on the balance sheet at a value I treat as half of book.¹³³ Cylance’s losses were reported as discontinued operations: $136M in fiscal 2024 and $81M in fiscal 2025.³ I count that as the most expensive decision in the ten years, and the people who made it are mostly gone. John Chen, the chief executive for most of that period, retired in November 2023, and John Giamatteo, who had led the cybersecurity division since October 2021, took over on 11 December.¹⁶¹⁷¹⁰

All figures on this page are in US dollars, which is how the company reports, and the price is the New York close. The year ends in February. “Fiscal 2026” is the year to 28 February 2026. The latest quarter is the second of fiscal 2027, to 31 August 2026, reported on 24 September.⁷⁴

QNX, the part that matters

QNX makes the software that sits closest to the hardware. When a car’s instrument cluster, camera system or driver-assistance computer has to respond in a fixed time, and cannot be allowed to fail, it often runs on a real-time operating system. The company says QNX has 45 years of embedded software expertise. Today the company says QNX is in more than 275 million vehicles, up 20 million in a year and 100 million since 2020, that it works with more than 45 automakers including the top 10, and that it is used by 24 of the top 25 electric-vehicle makers and every major Tier 1 supplier.³ A “Tier 1” is a supplier that builds the computer or system a carmaker puts in the car. Those counts are cumulative: 275 million is every vehicle that has shipped with QNX, not the number on the road and not a year’s output.

QNX makes money in three ways. Development licences are sold to engineers building a product. Royalties are paid for every unit that ships with the software. Services are engineering help to get a design certified. The mix matters, because royalties repeat and grow with volume while development licences come and go with project cycles. In fiscal 2026 the company says the QNX revenue rise of $32.0M came from $18.0M of extra royalties, $9.1M of development seats and $5.9M from Radar.³ In the second quarter of fiscal 2027 it came from $10.4M of royalties, $3.9M of development licences and $2.6M of services.⁷ The company does not publish the royalty total. If I assume that automotive is 80% of QNX revenue and that it ships into about 20 million new vehicles a year, QNX earns something like $11 a vehicle in all forms of revenue, at fiscal 2026’s $268M E (estimate).³ That is small beside the price of a car, which is both the case for it, since a carmaker will not miss the money, and the case against it, since a carmaker will notice if the money rises a lot.

The financial shape is clean. QNX revenue was $215M in fiscal 2024, $236M in fiscal 2025 and $268M in fiscal 2026, up 14%. The division’s adjusted EBITDA went from $33.5M to $59.1M to $71.0M, a margin of 16%, 25% and 26%.³ Its segment adjusted gross margin was 87% in the second quarter of fiscal 2027, 4 points higher than a year before.⁴ For the year ending in February 2027 the company now expects $315M to $325M of QNX revenue and $95M to $105M of segment EBITDA, which is a margin of about 31%.⁴ The first half is on pace: revenue of $152.6M, up 27%, and segment EBITDA of $48.3M.⁴ The company says QNX customers place more of their orders in the second half, so the back half is expected to be larger.³ The profit measure is before stock pay and before about $40M a year of corporate cost that the company does not charge to the divisions; I take both off later.

The order book

BlackBerry publishes one forward number for QNX: the royalty backlog. It is the company’s estimate of the royalties it will earn over the life of designs it has already won, using its own forecast of how many vehicles will ship. It was $640M at the end of fiscal 2023, $815M in 2024, $865M in 2025 and $950M in 2026.¹⁹¹⁸³ On 22 September 2026 it announced that Coretura, the software company owned by Daimler Truck and Volvo Group, had chosen Alloy Kore for its next commercial-vehicle platform, and that the win added more than $100M, the largest in QNX history.⁵ Read the backlog with care. It is a projection, not a contract; it depends on carmakers selling the vehicles they planned to sell; and it is struck once a year, so the next figure arrives with the annual report in April 2027. The company does not say over how many years the backlog will arrive, and I do not know what share of it is already in the royalties that QNX earns today. One more point: the backlog rose $85M in fiscal 2026, yet the proxy circular says new design wins added $257M, against an internal target of $280M.³¹⁰ About $172M therefore came out of the backlog in the year, as royalties were earned or forecasts changed C (calculated). A backlog is a bathtub, and the tap and the drain matter more than the level.

Beyond cars

About 20% of QNX revenue came from outside cars in fiscal 2026. The company lists medical devices, robots, industrial automation, rail, and aerospace and defence as markets where it is a preferred supplier, calls the group the general embedded market, and has packaged a stack for it, the General Embedded Development Platform.³ The newest label is physical AI, meaning machines that act in the real world. QNX added support for Hailo’s AI accelerator and says its own benchmark shows up to 14 times greater performance consistency than the real-time Linux it tested.⁷ The filings I read name no large non-automotive customer and give no growth rate for these markets, so I fold them into the QNX growth rate and do not value them on their own. I also have no figures to say whether a safety certificate matters as much to a robot maker as it does to a carmaker.

Secure Communications and Licensing

Secure Communications had revenue of $284M in fiscal 2024, $273M in fiscal 2025 and $259M in fiscal 2026, and the company expects $260M to $270M this year.³⁴ Its segment EBITDA was $56.1M in fiscal 2026, a margin of 22%.³ SecuSUITE is certified for classified use, has NIAP and NATO accreditation, and sells to governments who want encrypted calls on ordinary phones.³ AtHoc is the only critical-events platform with FedRAMP High authorization, and the company says it helps protect more than 75% of US government personnel.³ UEM, the device manager, is the legacy of the phone era and is shrinking: in the second quarter it fell $3.3M while SecuSUITE rose $5.5M.⁷ The division’s recurring revenue was $221M at 31 August 2026, and its dollar-based net retention was 91%, which means customers already on the books spend 9% less each year, and new sales have to cover that before the division grows.⁴ In September the company lowered its full-year guidance for the division by $10M at the midpoint, citing a more cautious view of some government opportunities in the second half.⁷ I value it as a stable, profitable business that will probably not grow.

Licensing is what is left of the patent business. It had revenue of $260M in fiscal 2024, when the Malikie sale was booked, and $22M in fiscal 2026. This year is $41M, because in the second quarter a new licensing arrangement brought in $22.1M against $6.6M a year before.³⁴ The company expects only about $6M in each of the next two quarters.⁴ I do not build a business on it. It matters for one reason: the total revenue growth the company reports includes it. Second-quarter revenue rose 26%; without Licensing it rose 14.8%, and for the half year 19.6% C (calculated).⁴

The comeback, checked

The framing I was asked to test is that a company left for dead is coming back to life. Some of it is true and some of it is a story, and it helps to separate them.

My reading is this. The phone business died and the company that carries its name was not dead: it had QNX, a government security business, and a few hundred million dollars of cash. What came back is not the company that fell. It is a small, profitable software company that spent six years cleaning up its mistakes and has now had a few good quarters. That is real, and it is worth noticing. But the shares did not wait for the proof. They rose 4.1 times in three months, and today’s price pays for a QNX that is far larger than the one in the filings.¹ The story “left for dead and back to life” is fully priced. The question for this page is whether the next ten years are priced too.

IIThe moat

The honest test of an edge is what it did when conditions turned. BlackBerry has one clear answer to that, and it is not a happy one. In fiscal 2012 it held a position in business phones that looked permanent: 49.0 million handhelds shipped and $18.4B of revenue, with enterprise customers who trusted its security.⁶ Within five years the revenue was $1.3B.¹² Security, a loyal customer base and a recognised name did not protect it from a competitor that changed what a phone was. I keep this in front of me when I read the claims about QNX, because the same words are used now.

What does protect QNX

• Certification takes years and has to be redone. A carmaker that puts software into a braking or steering function has to show it meets the ISO 26262 safety standard. QNX sells operating systems and a hypervisor that come pre-certified, and a new version has to be certified again.³ A rival with better code but no certificate cannot win a safety-critical design this year. This is the strongest of the barriers, because it costs time that no amount of money buys back.
• Designs last, and the royalties follow the design. A carmaker picks its software for a vehicle platform, builds a program around it, and sells the result for years. The company’s $950M royalty backlog is a forecast of exactly that, and it grew every year from $640M in fiscal 2023.¹⁹³ A win today is a contract with a cost to leave.
• The installed base is large. QNX is on more than 275 million vehicles, with more than 45 automakers including the top 10.³ That gives it reference customers, a pool of engineers who know the tools, and chip makers that publish drivers for it. NVIDIA released a development kit for its DRIVE AGX Thor computer with QNX OS for Safety 8, and Hailo added support for its AI chip.³⁷
• Engineers learn it early. The company gives the development platform free to students and hobbyists. It has issued more than 12,000 non-commercial licences and signed more than 80 agreements with universities.³ Habits formed in school come with the engineers to the carmaker.

Where it is thin

I can find three weak points. First, the competition is free. The 10-K says the division competes mainly with other providers of embedded software, including open-source Linux, that some carmakers and suppliers have developed embedded systems in-house and are looking at open source because they see it as costing nothing, and that rivals may sell at zero or negative margins.³ A carmaker with enough engineers can build on Linux and add its own safety work. Second, the edge is per program, not per customer. A win on one platform does not bind the next one. Third, I cannot measure share. The company says QNX wins an outsized share of advanced software-defined vehicle designs, but it gives no count of the programs it lost, and I could not find an independent figure for it. An edge I cannot measure, I do not count at full value.

The numbers that bear on pricing are mixed. QNX’s segment gross margin is 87%, which says the product costs little to deliver, not that customers pay a premium for it.⁴ Revenue per vehicle is small, roughly $11 on my arithmetic, so a carmaker can pay it without a fight.³ The proof that pricing holds would be revenue per vehicle rising while volume rises, and the company has not published the split. The sign I look for is royalty revenue growing faster than the 20 million vehicles a year, and in fiscal 2026 royalties rose $18.0M, more than half of the division’s $32.0M of growth.³ That could be more vehicles, more content per vehicle or both.

IIIThe competitive advantage

The question here is what QNX can do that a rival cannot, and for how long. The rivals above are not in the same position, and I know less about some than others.

Start with what is true of the market. A modern car has dozens of small computers. Carmakers are moving toward a few powerful ones, and that shift is what the industry calls the software-defined vehicle. The 10-K describes it as a move from microcontrollers to high-performance processors, from many separate control units to centralised computing, and with far more data in the car.³ A centralised computer runs mixed jobs at once: a screen the driver sees, a camera system that must not freeze, a steering function that must not fail. The software that keeps them apart is the hypervisor. This is where QNX says it is strongest, and Alloy Kore, built with Vector, a German maker of automotive software tools, is the company’s attempt to sell the whole stack and not only the kernel.³

The Alloy Kore bet. The company says Alloy Kore “has the potential to expand average selling prices by multiples on a per-vehicle basis”.⁷ That is the whole case for a much larger QNX, and it is a belief, not a result. The evidence so far is one commercial-vehicle win with Coretura, early access trials at Mercedes-Benz, and a place in the architecture BMW is planning for its Neue Klasse cars.⁵³ Coretura is a joint venture of Daimler Truck and Volvo Group, so it is a commercial truck customer, not a passenger car programme with millions of units. More than $100M of backlog from a win is the largest in QNX’s history, and it is also about 10% of the backlog the company already had.⁵³ I would want to see two more wins at a passenger-car maker before I treated the multiple as proven.

The tools and the people. QNX has spent on research and development: $61.7M in fiscal 2026 for the division, up from $59.3M, and about $19.6M in the second quarter, 51% higher than a year earlier.³⁴ Sales and marketing in the division rose 21% in fiscal 2026.³ That is growth being bought, not just harvested, and it shows in the margin guidance of 31% for this year, short of the 36% of the second quarter alone. The company is spending ahead of the revenue, which is what an owner wants if the revenue comes.

Where rivals can catch up

If Linux plus a safety layer from a small team becomes good enough, the price per vehicle falls. If Android Automotive takes over the screens and the next generation of cars joins safety and screen on one chip, QNX has to win on the hypervisor and the safety layer alone. If carmakers buy chips that come with their own software, the supplier wins the design before QNX is asked. I cannot say which is most likely. I can say that the market has not yet delivered the multiple the company hopes for, and Aptiv’s write-off is a warning on timing.⁹ I treat the advantage as real and narrow: real because certified safety software is slow to copy, narrow because a carmaker can build its own.

IVManagement

The people are new. John Giamatteo became chief executive on 11 December 2023, and before that he had been president of the cybersecurity division since October 2021 and had been at McAfee for seven years.¹⁷ Tim Foote is the chief financial officer. John Wall was named President of QNX in fiscal 2026, and Mattias Eriksson, who held the job before, is shown in the proxy as a former head of the division.³¹⁰ The chair, Richard Lynch, has been a director since 2013 and was the interim chief executive in the weeks between John Chen’s retirement and Mr. Giamatteo’s appointment.¹⁶¹⁷ On 27 October 2026 the board added David Chaikof as a director.²⁰ It is a team that has been in place for under three years, and its record is therefore short and mostly good: cost cuts, the sale of Cylance, and a profit.

How they are paid, and what that pays them to do

Mr. Giamatteo’s total pay for fiscal 2026 was $8.30M, of which $0.70M was salary, $0.89M the annual bonus and $6.70M stock units. In fiscal 2025 he was paid $1.42M, with no stock grant, and in fiscal 2024 $8.42M.¹⁰ The stock units were valued when granted on 2 April 2025, at $3.39 a share. That is about 1.98 million units, and at $8.02 they would be worth $15.9M if every unit vested at target, more than twice what the table shows C (calculated).¹⁰¹ The cost the company books for stock pay is what the grant was worth then. What it costs the owners is what the units are worth when they vest.

The annual bonus has three measures: revenue from QNX and Secure Communications together (35% weight, target $509M), adjusted EBITDA (35%, target $84M) and corporate operating cash flow (30%, target $37M). The fiscal 2026 results were $527M, $107M and $50M, and the factor was 126.82%.¹⁰ Look at the third one. Operating cash flow of $50M includes the $30M received from Malikie that I read as operating cash. Without it, the figure would have been about $20M, 54% of target, and by my arithmetic the bonus factor would have been near 100%, not 127% E (estimate).¹⁰³ I cannot prove that the receipt sits inside the measure; the company defines it as operating cash flow, and the Malikie payment was an amount owed on a sale booked as revenue.

The long-term units (PBRSUs, 60% of the chief executive’s grant) pay on three one-year measures: QNX revenue growth (35%), the combined revenue growth of Secure Communications and Licensing (15%) and adjusted EBITDA margin (50%), then multiplied by a three-year relative share price measure against a software and services index, for a payout of 25% to 200%.¹⁰ What does that pay for? Growth of QNX and a high margin. It does not pay for free cash flow, for cash after stock pay, or for a lower share count. Licensing revenue is included in a growth measure, and licensing is the line that rose $15.5M in a quarter from one deal. A one-off patent licence is therefore worth money to the people who sign it, whether or not it repeats. I would rather the plan paid on QNX alone.

The ownership rules are firm. The chief executive must hold five times his salary in shares; the guideline was raised from four times in December 2025.¹⁰ Mr. Giamatteo owned 899,146 shares directly on 1 May 2026, worth about $7.2M, which is 10 times his salary.¹⁰ He is paid in stock and owns stock, and I count that in his favour. The directors and officers together held 3.0 million shares, 0.51% of the company, including stock units they can claim within 60 days.¹⁰ The circular of 1 May 2026 named three holders above 5%: Fairfax Financial at 6.04%, Legal & General at 5.66% (a filing as of 31 December 2024, with no newer one that I found) and BlackRock at 5.05%.¹⁰ Since then two have moved. Fairfax, which lent the company $150M through convertible debentures in November 2023 that have since been repaid, sold about 9.1 million shares on the open market on 20 and 21 April and 4 May 2026, at $5.50 and about $6.04 a share, and its filing of 6 May says it no longer owns more than 5% C (calculated).³²¹ BlackRock’s filing of 27 July shows 4.4% at 30 June.²² The long-time largest holder sold at 69% to 75% of today’s price, and I do not know why. No holder runs the company or controls the vote.

What they did with the money

Capital allocation is the record to judge them by, and for the ten years it is poor, even though the current team can be blamed for little of it. Between fiscal 2017 and fiscal 2026 the company spent about $1.4B on Cylance and sold it for about $118M in cash plus shares; it sold its patents for $170M plus $30M plus a royalty it cannot forecast; and its basic share count rose from 525 million to 592 million, 12.7%, as stock pay was settled in new shares.^1331411 In fiscal 2026 it bought back 15.6 million shares for $60.7M, an average of $3.89, and in the first quarter of fiscal 2027 another 2.6 million for $10.0M, an average of $3.85. Those 18.2 million shares are worth about $146M at $8.02, twice what was paid.³⁷ That was a good use of money, made when the shares were hated. None were bought in the second quarter, when they were at $8 and above, and the renewed bid allows up to 26.8 million shares to May 2027.⁷²³ The pattern is that of an owner buying low, which I like.

Against that, the convertible notes are a cost. The company borrowed $200M in January 2024 at 3%, with a conversion price of $3.88, and with the shares at $8.02 the holders convert into 51.5 million new shares worth about $413M. That is 8.8% of today’s shares, handed over in exchange for $200M.³⁷ The loan was cheap to service and expensive to repay. I count the conversion in my share count and ignore the debt.

VFinancials

The ten years of this company’s statements need a warning label. Fiscal 2017 to 2022 include Cylance from February 2019, which was sold in February 2025 and is removed from fiscal 2023 onward, so the earlier years are not the business of today. Fiscal 2018 net income includes arbitration awards and settlements that added $685M.¹³ Fiscal 2024 includes the Malikie patent sale. Revenue after the fall is not comparable with revenue before it. I use the numbers anyway, because the trend is the point, and I mark where it breaks.

The top line first. It fell from $1.3B in fiscal 2017 to $549M in fiscal 2026, and with Cylance taken out and Malikie put aside the continuing business is about flat: $526M in fiscal 2023, $535M in fiscal 2025 and $549M in fiscal 2026, with fiscal 2024 at $759M because of the patent sale.¹¹ Growth is a very recent fact. It starts with QNX in fiscal 2025 and is clear only in fiscal 2027, when the outlook has revenue 14% above fiscal 2026.⁴ The net result swung from losses of $1.2B in fiscal 2017, $1.1B in fiscal 2021 and $734M in fiscal 2023 to profits of $53M in fiscal 2026 and $42M in the first half of fiscal 2027.¹¹⁴ A comeback is visible in these bars. So is how far it has come from.

Free cash flow tells the same story with less flattery. Over the ten years it adds to $363M. But $689M of that is fiscal 2018, with the arbitration money; $170M is the Malikie closing payment in fiscal 2024; and $30M is the later Malikie payment in fiscal 2026. Take those three out and the ten years burned about $0.5B C (calculated).^1114153 The burn was concentrated: $270M in fiscal 2023, when the business still carried Cylance’s losses, and $241M in fiscal 2017, when the phone business was ending.¹¹ Capital spending is no longer an issue. In fiscal 2011 the company spent $1.0B on property and equipment, and in fiscal 2026 it spent $3.8M.¹¹

Cash tells where it went. Cash and investments were $2.35B at the end of fiscal 2018 and $1.0B a year later, after the Cylance purchase, and they fell to $298M in fiscal 2024 before recovering to $432M in fiscal 2026 and $447M at 31 August 2026.¹³¹⁹³⁴ The company paid for its mistakes from a pile of cash left over from the phone years, and that pile is gone. What remains was rebuilt from patent sales and a better business.

What an owner can really take out

Free cash flow as the company reports it leaves out two costs. The first is intangible assets, mostly capitalised software and patents: $13.8M, $7.0M and $5.7M in the last three fiscal years.³ The second is stock pay, which the company adds back because it is not cash, but which the owners pay for through the shares that are issued or have to be bought back. I take it off at the cost the company books, $23.2M in fiscal 2026. After both, owner cash was −$58M in fiscal 2024, −$19M in fiscal 2025 and +$18M in fiscal 2026 C (calculated). Take out the Malikie receipts that I read as operating cash, and the last year is −$12M.³ The first two years are also dragged down by Cylance, whose operations were reported separately but whose cash is in the consolidated statement until the sale in February 2025. By my estimate it burned about $85M in fiscal 2024 and $55M in fiscal 2025, which would put the continuing business near −$142M and +$36M without the patent cash E (estimate).³

Set that against reported earnings. In fiscal 2026 net income was $53.2M and adjusted net income $97.3M, and owner cash was $17.6M, or −$12.4M without the Malikie cash.³ Earnings were three to five times the cash an owner could take.

So the real turn is this year. For fiscal 2027 the company guides to operating cash flow of about $115M, of which $33.9M was earned in the first half. After purchases and stock pay that is about $72M of owner cash, 4 times the best year before it C (calculated).⁴⁷ A large part of that is the second half: $81M of operating cash in six months, in a business whose first-half pattern is $34M.⁴ The company says QNX orders come late in the year. I take it at its word, and I note that the price is paid today for a cash flow that arrives mostly after the end of the calendar year. If the second half comes in at the first half’s pace, the year’s owner cash is about $26M.

The first half was strong in every line except one. QNX revenue rose 27% and its segment profit 45%. Secure Communications revenue rose 13% on a strong first quarter ($73.6M) and was flat in the second ($60.9M, +2%), with a segment margin of 13% against 16% a year earlier.⁴ Licensing is lumpy. The company raised its full-year view for QNX by $20M at the midpoint and lowered it for Secure Communications by $10M.⁷

The balance sheet

There is little debt. The only borrowing is the $200M of 3% notes due in February 2029, and at $8.02 they convert into 51.5 million shares, so I treat them as equity. The company had no other debt at 31 August 2026.⁷ Cash and short-term investments of $377M are free to use. The long-term investments of $55.5M are mostly shares in Arctic Wolf, a private company, which cannot be sold on an exchange, and I count half their book value.⁷ Customers have prepaid $128M, which will turn into revenue.⁷ And the tax shelter is large. At 28 February 2026 the company had $1.9B of gross operating losses, $476M of research credits and $207M of minimum tax credits to use against future profit, with deferred tax assets of $1.5B that carry a full valuation allowance.³ A $1.5B asset on a company whose market value is $5B is not a rounding error, and it is the reason cash tax stays low. Losses of $309M expire in 2036 and $492M in 2037, and rules on changes of control or on how losses are used can cut the benefit. I assume cash tax of 10% for several years and rising to 25% by fiscal 2035 in my base.

The price in numbers

At $8.02 the enterprise value is $4.9B, which is 7.8 times the revenue and 33 times the adjusted EBITDA that the company expects this year. Owner cash of about $72M is a 1.4% yield. A government bond pays 4.91%.²⁴ An owner at this price is buying growth, not income.

Look at the path of the price, because it affects how to read the number. The shares were $7.02 in November 2016 and $8.02 now, a gain of 14% in ten years with no dividend. In between they closed as low as $2.11 on 7 August 2024 and as high as $25.10 on 27 January 2021, a one-day spike that ended at $14.65 the next day.¹ From the 2026 low of $3.15 on 30 March the price rose 4.1 times to $12.81 on 30 June, it jumped 20% on 25 June, the day of the first-quarter results, and has since fallen 37%.¹ I do not know why it fell in July. The second-quarter report, on 24 September, saw it rise 4% on the day and fall 6% the next.¹ Anyone who bought on the way up has a large gain and anyone who bought at the top has a large loss, and the price today holds both views.

VIValuation

First, how this goes badly wrong

Before any numbers, I ask what would hurt an owner most at this price. I see three ways.

The price is a multiple of a business that has not yet shown it can earn it. At $8.02 the enterprise value is 7.8 times this year’s revenue and 33 times its adjusted EBITDA, for a company whose owner cash was negative three years ago.⁷⁴ Even if every operating number comes in, the multiple has room to fall from 33 to 20 as growth slows from 19% to the low teens, and a fall like that takes about 40% off the enterprise value without any bad news. That is the risk I put first, because it needs no mistake by management.

Carmakers decide software-defined vehicles are slower and cheaper than the pitch. Aptiv wrote off $648M of Wind River goodwill in 2025 and gave delays in software-defined vehicle launches as a reason.⁹ If cars are built on Linux and Android for the screen and on a thin safety layer for the rest, QNX still sells, but at the price of a component. My test with half the growth takes the base to $2.65.

The cash arrives late and the dilution arrives on time. The company expects most of its year’s operating cash in the second half, owner cash was negative as recently as fiscal 2025, and 51.5 million new shares come from the notes, 8.8% more, while the units for staff add about 1.7% a year.⁴¹⁰⁷ Cash can be late, and shares do not come back.

What an owner can really take out

The company’s operating cash flow for this year is about $115M. Take off capital spending of about $14M and stock pay of about $29M and the owner cash is about $72M, a 1.4% yield on the diluted market value of $5.29B C (calculated). In my base case owner cash is $85M in fiscal 2028, $125M in fiscal 2030 and $159M in fiscal 2032, when it is 3.0% of today’s market value. The gap between the adjusted EBITDA of $150M and owner cash of $72M this year is the part I most want a reader to see: it is about half, because of stock pay, small taxes and the cash that is not yet in hand.

What I can’t judge

I cannot judge four things. The first is how much of a vehicle’s software cost a carmaker is willing to pay a supplier, which is what Alloy Kore is a test of. The company believes it can raise the price per vehicle several-fold, and I have one result, from trucks.⁷⁵ The second is what share of QNX’s revenue is royalty, since the company does not say, and so I cannot split growth into more cars and more money per car. The third is how much of the tax loss pool the company can actually use in time, given rules I cannot read from here. The fourth is the Malikie royalty, which depends on a company that is not BlackBerry and which the company excludes from its own forecasts.³ Where I cannot know, I give a range, not a number, and I let the price stay above the range.

What QNX could be worth

BlackBerry has three businesses and one set of costs, so I value the parts first and add them, as a check on the cash flow model that follows. For QNX I use a multiple of segment EBITDA, since the company has no profit figure for it after corporate costs. I then take off the costs QNX does not carry on its own: about $40M a year of unallocated corporate cost and about $29M a year of stock pay, each at 8 times. I value Secure Communications at 7 to 12 times EBITDA because it is profitable and flat, and Licensing on what the stream is worth.

On those multiples the parts add to $2.68, $4.52 and $6.44 a share, which brackets the DCF base. The price is above the top of the range. Turn the question around: if I give everything but QNX my base multiples, the price leaves $4.8B for QNX.

That is 48 times QNX’s segment EBITDA for this year and 15 times its revenue. It is $17 for every vehicle that has shipped with QNX since the start, when the company earns roughly $11 a year on each new one.³ For comparison, my bull multiple of 35 times is still 27% below what the price pays.

Now the next 5 and 10 years. The table below takes QNX alone, builds revenue and margin in each case, and applies a multiple that falls as growth slows.

Read the last rows. In the base case QNX is worth about $4.8B in fiscal 2032, almost exactly what the price implies today, and $5.8B in fiscal 2037. In five years QNX has more than doubled its profit, from $100M to $240M, and an owner who paid today’s price earns nothing from it. That is the whole sentence. The value of QNX grows 9.5% a year over ten years in my bull, which is below the 10% I require. The company has to do better than my bull for the shares to be a good purchase at $8.02, and the bull is already a business of $1.2B of QNX revenue, 3.8 times this year’s.

How the cases are built

I build each case year by year from the segments. QNX revenue starts at the $320M midpoint of the outlook and grows at a rate that I set for each case and let fade. Its margin is the segment margin before corporate costs. Secure Communications starts at $265M. Licensing is the stream I think survives. I take off corporate costs of $40.5M a year (rising 2% to 4%), then stock pay, restructuring, cash tax, capital spending and purchases of intangibles, and what is left is owner cash. Cash tax is 10% of taxable profit (adjusted EBITDA less stock pay, restructuring and about $8M of depreciation) for the first years and rises to 25% by fiscal 2035 in the base, because the losses of $1.9B shelter profit but expire in part from 2036, and in the bear it rises sooner. The second half of fiscal 2027 uses the company’s outlook directly: operating cash flow of about $81M, less capital spending and half a year of stock pay. I discount at 10.0%. That is the 10-year US Treasury yield of 4.91% on 2 November plus 5.1 points for a small company in a cyclical market with a mixed record on capital and a business I would not call proven.²⁴ I stop in fiscal 2037 and grow the last year’s owner cash at 2.5%, 3.5% or 4.0%. I count net cash of $405M and no debt, and 659.3 million shares.

Three comparisons help. My base fiscal 2028 adjusted EBITDA is $145M against the $149.5M the company expects this year, lower because Licensing falls from $36M to $12M. QNX revenue grows 14% a year to fiscal 2032, below this year’s 19%, and my base QNX margin reaches 39% by then. Against it, the bull has QNX growing 18% a year for five years, to $737M, with a 44% margin, a high bar for which I have no precedent in the filings I read. I did not build a higher bull. An even higher case exists, in which Alloy Kore lifts the price per vehicle by a multiple of what QNX earns now, and I weighed it, but I cannot put a number on the chance without a second win.

In plain words, the price needs QNX to reach $1.0B of revenue in five years, 3.0 times this year’s, and $1.9B in ten, 5.9 times, at a 40% margin, with the rest of the company as I assume. It could come from more vehicles, from far more content per vehicle, from other markets such as robots and medical devices (about 20% of QNX revenue today) or from all three. It also corresponds to a discount rate of 6.7% on my base cash flows, only 1.8 points over the Treasury yield. None of these is absurd in isolation. Together they say that the market is paying for a company that is as good as the best software businesses and that does not yet look like one E (estimate).

The margin of safety, and the price against the range

My margin-of-safety price is $2.89. It is 70% of my base value. At that price an owner would have a 30% cushion against being wrong about QNX’s growth and the tax and stock lines, and the price would be 76% above my bear. It is the sort of price the shares traded at in March 2026: the lowest close that month was $3.15 on 30 March, 9% above it.¹ At $8.02 there is no margin of safety. The price is 94% above my base, 19% above my bull, and 4.9 times my bear. My weighted value, with 25% on the bear, 50% on the base and 25% on the bull, is $4.16.

The price reflects a real debate, and the case for the market being right deserves to be written out. It says that QNX is the best-positioned supplier of certified safety software in the world’s cars, that 275 million vehicles and a $950M backlog are a base that will compound, that Alloy Kore could multiply revenue per vehicle, that physical AI and robotics open a market bigger than cars, and that a business with an 87% gross margin and a 31% segment margin deserves a premium. If I give all of that its weight, my bull case is where the price lands, and it takes more. If I am too harsh on the discount rate and use 9.0% with 4.5% growth for ever, the base rises to $5.42, still below the price. The case against says that the revenue is $626M, not billions; that this year’s growth includes a one-off licence; that the royalty backlog is a forecast, not a contract; that the cash is mostly in the second half; that Aptiv has already written down Wind River, a direct rival; and that a stock that rose four times in three months is priced on belief. I give the second side more of the weight, and I would not be surprised if the first side were right for a while.

The next best use of the same dollar

The nearest comparison on this site is another software company with real owner cash. On this site, Intuit is priced 20% below my base value, with owner cash of $5.0B in my work. Constellation Software is priced 11% above its base. BlackBerry is 94% above. A 10-year US Treasury yields 4.91%, and my base return on BlackBerry to November 2036 is +0.5% a year E (estimate).²⁴ The bond beats my base case by 4.4 points a year with no risk, and my bull case earns 7.0%, short of the 10% I ask. QNX is a good business and has become a better one. Price is the other question. A good business at a price that needs more than my bull is a poor purchase, for the same reason a poor business at a low price can be a good one. I give neither a grade.

The bull case also needs many things at once, so I multiply them out. QNX growth near 20% a year for five years (about 20% likely, by my guess), a 44% segment margin (35%), software-defined vehicles arriving on schedule with no car downturn in ten years (60%), Secure Communications holding its revenue (50%) and a 22 times multiple still available in fiscal 2037 (40%). Together that is under 1% if they are independent. They are not, since a strong QNX helps three of them, so I call it 5% to 8%, not 25% E (estimate). My 25% weight on the bull is generous on that arithmetic.

Where this could be in 5 and 10 years

For the base to be right at five years, QNX needs revenue near $616M and a 39% margin in fiscal 2032, and the value is then $5.01 a share, below today’s price, because five years of cash have gone out and the rest has grown. At ten years the base reaches $5.92. In the bear, QNX stays under $600M of revenue and the value is $1.79; in the bull it is $1.2B and the value $10.54. The inputs that drive it are QNX growth, its margin, the tax line and the 10% rate. These are estimates of value, not forecasts of the share price. The price can sit above or below value for years, and in the last ten years it ranged from $2.11 to $25.10.¹

VIIThe case for the price

What most people miss

Three things cut in BlackBerry’s favour and get little attention. First, owners’ cash is barely taxed: the company had $1.9B of operating losses and $0.7B of credits to use at 28 February 2026, a $1.5B deferred tax asset carried at nothing, and a tax charge of 10% of profit in fiscal 2026.³ Second, it bought 18.2 million of its own shares between May 2025 and May 2026 for $70.7M, about $3.88 each, and they are worth about $146M now.³⁷ Third, the pricing room is wide. A carmaker pays QNX roughly $11 for each new vehicle by my estimate, a small part of the cost of a car, and software in the safety path can raise its price a long way before a buyer rethinks.³

The bull case

In plain words: QNX is a leading supplier of safe real-time software for cars, it is on 275 million vehicles, and the next generation of cars needs much more software. John Giamatteo, the chief executive, said on 24 September 2026 that revenue, profitability and cash flow all beat the company’s expectations and that the first Alloy Kore win was the largest design win in QNX history, and on 25 June that software-defined vehicles and physical AI add significantly to QNX’s long-term potential.⁴²⁵ The company raised its fiscal 2027 QNX outlook in September.⁴ I found no named analyst or fund manager whose published view on this company I could verify, so the bull case here is management’s. Fairfax, the long-time largest holder, has been a seller this year at $5.50 to $6.04, and a holding is not a statement of view in any case.²¹

Against reality

If QNX hits its guidance, wins more designs like Coretura and the margin goes to 40%, my model gives a value near my bull, $6.75. The price of $8.02 is 19% above that. For the shares to be worth $8.02 on my model, QNX would have to reach about $973M of revenue in fiscal 2032 from $320M this year. That is 25% a year for five years, against 12% a year from fiscal 2023 to this year’s outlook C (calculated).⁴¹⁸ By fiscal 2037 it would need $1.9B at a 40% margin, 36% more than my bull. The distance is about the multiple as much as the growth: the price pays 48 times this year’s QNX EBITDA, and my base uses 25 times.

What a 10x would take. Ten times $8.02 is $80.20, a diluted market value of about $53B. At 25 times EBITDA that needs about $2.1B of adjusted EBITDA, 14 times this year’s $150M C (calculated). At a 35% margin that is revenue of about $6B, 10 times this year’s guidance, or 25% a year for ten years. QNX revenue grew 11.6% a year from fiscal 2023 to this year’s guidance, and the best run in my price data is 6.1 times, from 7 August 2024 to 30 June 2026, with 37% since given back.¹¹⁸⁴ Tenfold from here is close to impossible, and I would not build anything on it.

Where I come down

What moved me was the operating turn: QNX growing near 20% at a 31% margin, and owner cash this year of about $72M after stock pay. What did not move me was the price, since the best case I can write for QNX is worth $6.75 a share. What would raise my values is a second Alloy Kore win at a passenger-car maker, a royalty backlog above $1.2B in April, and QNX growth near 20% into fiscal 2029, which would add about $1.10 a share to my base.

VIIIRisks, and what would change my view

The three ways this goes badly wrong are at the start of Valuation. These are the rest, and a test of each.

• Cars and trade. The 10-K says a US administration has imposed or threatened tariffs on imports from Canada, Mexico, China and others, including steel, aluminium and automobiles, and that talks on the Canada–United States–Mexico agreement are due to start later in 2026. These would hit the company’s automotive customers and so the demand for its software.³ I cannot estimate the odds; my test of a two-year downturn is in the table below.
• China. The company names Leapmotor and a leading Chinese automaker among QNX customers.³ The company does not give a revenue split by customer, and trade rules could cut that business. I do not know how big it is.
• Concentration. QNX is 51% of revenue this year and, by my SOTP, most of the value. Two customers were each more than 10% of receivables at the year-end.³
• Old claims. A class action filed in Canada in 2013 over the handset-era accounts is still described in the August 2026 10-Q. The company says it cannot estimate any loss.⁷
• Grants. QNX has recognised $28.6M from a Canadian government innovation fund, and part could be repayable if terms are not met. The company does not think that is probable.⁷
• Goodwill. $478M of goodwill from past purchases sits on the balance sheet. A writedown would not change cash, but it would say something about the earlier price paid.⁷
• Control. No holder controls the company, and Fairfax, the largest at the time of the circular, has sold down to under 5%. That helps minority holders and leaves the company open to a bid or to a fight.¹⁰²¹

What a bad year looks like in the model

The surprise is how little a car downturn costs if it ends. QNX growing 2% and 3% in two years takes $0.90 off the base, because the value sits in the long run. What matters more is anything permanent: halving QNX’s growth takes $1.48, and a margin squeezed to 25% with slower growth takes $2.08, most of the way to my bear. The upside cases are smaller than the downside cases. Alloy Kore working adds $1.10 and a Malikie payout $0.28. On my numbers, what can go wrong costs more than what can go right adds.