Software in the Age of AI - Canada's Constellation Software Inc.
Our perspective on software in the age of AI. Canada's Constellation Software Inc. – an unparalleled acquisition machine for mission-critical software applications. Discover why we find Constellation Software Inc. so compelling in our latest Monthly newsletter.
MONTHLY
Basis: Market capitalisation of approximately USD 47 billion (as of August 2026); Equity FCF 2025 of USD 2.29 billion; five-year horizon; OHP's own estimates. The dividend is symbolic and negligible, with a yield of around 0.2% – an increase would be interpreted as an admission of a lack of reinvestment opportunities.
The range of 7.1% to 16.5% per year is deliberately broad. It reflects the fact that the main driver of value lies not in the operating business, but in the ability to continue reinvesting the cash flow generated at attractive returns.
Conclusion
Constellation Software is one of the few companies where entrepreneurial thinking does not depend on a single individual. More than a thousand managing directors are responsible for their own business units. Executives invest their bonuses in shares using their own money.
The underlying business model is exceptionally robust: mission-critical software across more than 75 niches, recurring revenues, approximately 125,000 customers in over 100 countries, and low capital requirements. Over nine years, revenue has more than quintupled, while Equity Cash Flow per share has increased from around USD 22 to approximately USD 108 – without a single new share being issued.
The fact that the share price has lost around half its value since mid-2025 is due to two unresolved questions: the founder's departure and the impact of artificial intelligence on the value of niche software. The first has been addressed exactly as one would hope – with a successor from the company's inner circle, thirty years of service and a nine-figure investment in his own company. For the second, there is so far no empirical evidence, but there is a management team that speaks openly about it.
It is precisely this turning point that has brought the price to a level we consider attractive based on our assumptions. It is rare to find a business model of this quality at a price that already reflects so much uncertainty.
Over 1,000 software companies. Around 125,000 customer organisations in more than 100 countries. And a founder of whom hardly any public photographs exist.
Constellation Software Inc. is not a software company in the traditional sense, but a capital allocation machine that has been acquiring and permanently holding mission-critical niche software (Vertical Market Software) since 1995. Entrepreneurial in its culture, leading in its discipline – and, following the founder's withdrawal, a company worth reconsidering for the first time in years.
“Vertical Market Software is the distillation of a conversation between the vendor and the customer that has gone on frequently for a couple of decades” – Mark Leonard (2025) on the true value of Constellation's software.
Note: Constellation reports its financial results in US dollars, while its shares are listed in Canadian dollars. All amounts in this article are stated in US dollars.
Why am I writing about Constellation Software? For over thirty years, Constellation Software has been acquiring mission-critical niche software, holding it permanently and reinvesting almost every dollar earned into the next acquisition – without having to raise new capital. Concerns about AI and new corporate leadership are weighing on the share price – for long-term investors, I see first-class underlying value with intact upside potential.
The Product: The Software That Keeps Operations Running
Constellation does not develop its own software. The company acquires it – and a very specific type: Vertical Market Software – programs developed for the specific needs of individual industries. The timetable management system of a municipal transport operator. The billing software of an energy utility. The patient management system of a dental practice, the accounting system of a municipality, the lending management system of a library.
These programs have three characteristics that make them economically exceptionally attractive. They are mission-critical: If the software fails, operations come to a standstill. They are deeply embedded in workflows – switching providers means data migration, retraining and operational risk for minimal price benefits.
Five examples from the portfolio illustrate what this means in practice.
Trapeze Software - Public Transport. Calculates timetables, vehicle schedules and staff rosters for transport operators: which bus operates on which route, with which driver, while complying with driving hours, collective bargaining agreements and connections. What makes it special is the depth of its rules – the operational exceptions incorporated over decades cannot simply be replicated in a single project.
Advanced Utility Systems - Billing for Energy and Water Utilities. Covers the entire process from meter to payment: reading consumption data, calculating tariffs, issuing invoices and managing payment reminders. Around 6.6 million service connections are managed through the system. What makes it special is the complexity of the tariff models – tiered consumption levels, multiple services on a single invoice and state-specific requirements. The system is effectively the utility provider's cash register.
Jonas Club Software - Golf and Country Clubs. More than 2,300 clubs in over 20 countries manage their operations using around 60 modules: tee-time bookings, member accounts, point-of-sale systems in restaurants and pro shops, and accounting. What makes it special is the integration: A drink ordered by the pool is automatically added to the member's monthly statement.
SirsiDynix - Libraries. Manages catalogues, lending, user accounts and interlibrary loans for public and academic libraries worldwide. What makes it special is the data: catalogue records maintained over decades in library metadata standards, often comprising millions of entries. Migration is not an IT project, but a multi-year undertaking.
MEDHOST - Community and Rural Hospitals in the US. Electronic health records, including an emergency department system: patient tracking on a graphical floor plan of the ward, nursing and physician documentation, and electronic medication orders. What makes it special is its design for smaller hospitals without their own IT departments, combined with US healthcare certifications – a barrier to entry that pure technology providers regularly fail to overcome.
The revenue base is correspondingly stable. Of the USD 11.6 billion in revenue generated in 2025, around 75% came from recurring revenues. Approximately half of the group's revenue originates from the public sector – municipalities, school districts, police departments, public transport operators and utilities – customers that change providers only very slowly. Payments are made in advance: At the end of 2025, USD 2.2 billion in advance payments received were recorded on the balance sheet, effectively an interest-free loan from customers.
What this means for us as investors: We are not betting on a single product, but buying exceptionally broad diversification: over 1,000 companies, around 125,000 customers, more than 100 countries and over 75 niches. No single customer is material to the group. The concentration risk that is otherwise typical of niche providers simply does not exist here.
The History: From USD 18 Million to One of Canada's Most Valuable Companies
In 1995, a 39-year-old Canadian venture capitalist named Mark Leonard raised around USD 18 million from the OMERS pension fund and former colleagues. His thesis at the time: Small software companies operating in unexciting industry niches offer better return potential than celebrated technology stocks – recurring revenues, minimal capital requirements and high switching costs. The term SaaS did not yet exist.
The 2006 IPO deserves a footnote: Its sole purpose was to provide liquidity to early investors — the company itself did not raise a single dollar.
Mark Leonard: The Visionary Founder
Leonard is an exceptional figure. He gives very few interviews, and hardly any photographs of him exist. He has become known solely through his President's Letters — annual shareholder letters that enjoy a reputation in the investment world comparable only to those of Warren Buffett.
In 2015, he gave up his salary and bonus entirely, a decision he maintained until his resignation ten years later:
“I gave up all compensation because I want to work less hard in the future.” – Mark Leonard, President's Letter, April 2015 (paraphrased translation)
In the same letter, he explained why he would pay for his own travel expenses going forward: Until then, he had flown economy class and stayed in inexpensive hotels because he did not want to benefit at shareholders' expense and wanted to set an example for employees. From then on, he would more likely be found at the front of the aircraft - at his own expense.
This attitude is embedded in the company's system. At Constellation, there are no stock options and no free shares. Instead, the bonus plan requires executives to invest their after-tax bonuses in Constellation shares — purchased on the open market, with their own money, and subject to a multi-year holding period. By 2015, the company had already created countless employee millionaires in this way.
When asked why the decentralisation of subsidiaries was so central to Constellation, Leonard responded with a biographical anecdote: He had been punished by his school principal, challenged teachers, left his parents' home at an early age, held an illegal radio licence and built a flamethrower. The essence: He cannot stand being told how to do something - and therefore does not want to impose that on anyone else.
Mark Miller: Continuity and Skin in the Game
On September 25, 2025, Leonard stepped down with immediate effect for health reasons. In March 2026, it became known that he would also not stand for re-election to the Board of Directors. It is the most significant leadership change in the company's history - and one of the main reasons why we are taking a closer look at the company right now.
His successor was Mark Miller, and the appointment is a deliberate signal of continuity. Miller was a co-founder of Trapeze Software, Constellation's first acquisition in 1995. He joined the company practically at its inception, led the Volaris Operating Group for many years and served as Chief Operating Officer under Leonard. Above all, however, he is regarded alongside the founder as an architect of the decentralised model.
Miller holds around 1% of the shares - worth approximately USD 540 million at the current valuation. At the Annual General Meeting in May 2026, he summarised his leadership philosophy in a single sentence:
“There is no group-wide CIO deciding which AI platform we use or whether we turn left or right. That is what our managing directors around the world do.” – Mark Miller, President of Constellation Software, Annual General Meeting May 2026 (paraphrased translation)
What this means for us as investors: We see corporate governance here that is rarely found: no dilution, no share-based compensation, a CEO with a nine-figure investment in his own company, and a management team required to invest their bonuses in shares using their own money. Those who are invested in this way think in decades rather than quarters.
The Market: Fragmentation as a Growth Driver
The market for vertical market software is a fragmented landscape consisting of small providers, mostly founder-led, mostly regional and often without a succession solution. This fragmentation is precisely what drives the business model. The target profile is clearly defined: companies typically generating USD 1 to 30 million in revenue, often facing a succession situation, with indispensable software and a loyal customer base.
The success behind this is impressive: Since 2004, revenue has grown by more than 26% per year, while operating cash flow has increased from USD 25 million (2005) to USD 2.73 billion (2025).
It is important to note: This growth has been driven by acquisitions — the target companies are stable and grow through price increases broadly in line with inflation. Constellation is purely an acquisition story!
Competition: Constellation's Moat in Scale and Culture
Constellation faces competition on two fronts. On one side, we see competition for customers - the users of its software solutions. Here, competitive pressure is low. Constellation deliberately selects niches with few competitors; corporations such as SAP, Oracle or Microsoft are too large, too generic and too expensive for these applications.
The real competitive front is on the transaction side: competition for acquisition targets. Here, pressure has increased significantly since 2020 - private equity firms such as Vista Equity and Thoma Bravo, the US corporation Roper Technologies and a growing number of imitators are entering the same market. Constellation's advantage here is not financial but structural: its reputation as a permanent owner with strictly decentralised management. Throughout the company's entire history, only one business has ever been sold. The selling founders/managing directors retain operational control. For a founder who cares about the fate of their life's work and their employees, this is the decisive difference compared with the buy-optimise-sell approach of financial investors.
Added to this is acquisition discipline. Constellation does not publish its return requirements, but former executives have described them repeatedly in industry interviews: around 30% IRR for companies with revenue below USD 1 million, and 15% IRR for companies with revenue of USD 50 million or more. Around 90% of all transactions achieve an IRR of 25%. (Sources: InPractise / SBO Financial / Colin Keeley / Windsor Drake).
Financial Performance
(Note: Revenue, revenue growth, Equity FCF and FCFA2S, Constellation Software 2016–2025; all amounts in USD million. Equity FCF = operating cash flow less capital expenditures, lease repayments and minority interests, own calculation based on annual reports. Constellation has only reported the FCFA2S metric since 2018.)
The figures since 2016 demonstrate one thing above all: consistency. Revenue has increased from USD 2.1 billion to USD 11.6 billion – more than fivefold, approximately 21% per year, without declining a single time in nine years, not even during the pandemic year of 2020. Equity FCF grew from USD 472 million to USD 2.29 billion over the same period, approximately 19% per year. On a per-share basis, this represents an increase from around USD 22 to approximately USD 108, without a single new share being issued.
The fact that cash flow grew somewhat more slowly than revenue is primarily due to the increasing size of acquisitions: Larger transactions are more expensive, initially tie up more capital, and the Topicus spin-off introduced significant minority interest deductions for the first time from 2021 onwards. This does not change the supply of acquisition targets – as described, the pool of target companies remains large, fragmented and attractively priced.
Reported net income is only of limited use as a benchmark in this context. In 2025, it fell by 30% to USD 512 million, while operating cash flow increased by 24% to USD 2.73 billion. The reason is non-cash valuation effects of approximately USD 1.9 billion – primarily the amortisation of intangible assets arising from purchase price allocations and two revaluations relating to the subsidiary Topicus.
The current year impressively confirms this momentum. In the second quarter of 2026, revenue increased by 17% to USD 3.34 billion, and acquisitions with a total value of USD 893 million were completed in that quarter alone. The acquisition machine is running.
What this means for us as investors: The market, competitive position and financial figures form a coherent whole: The fragmented niche market continuously provides acquisition targets, the reputation as a permanent owner provides access to them, and the tiered minimum return requirements ensure that only economically attractive acquisitions are made. The result: approximately 21% revenue growth and 19% cash flow growth per year over nine years, entirely without dilution. For us, this is the core of the investment thesis – not the growth of individual niches, but the ability to repeatedly allocate capital at attractive returns.
One striking observation: The gap between our Equity FCF and the company's own FCFA2S metric is becoming increasingly wide. This is essentially attributable to a single balance sheet item – and because it is central to our valuation, it deserves a section of its own.
Excursus: The IRGA Liability – And Why We Calculate Differently from Management
When Constellation acquired the Dutch software provider TSS (Topicus) in 2014, an investor group associated with the local management – the Joday Group – acquired a one-third stake. At the time of Topicus's IPO in 2021, the agreement was replaced by the Investor Rights and Governance Agreement (IRGA). Today, Joday holds approximately 29.4% of the shares in the Topicus cooperative.
The core of the agreement is a put right: Joday can sell its shares to Constellation at any time for a contractually defined price in cash. Because this creates a cash payment obligation, the item is classified as a financial liability and must be remeasured at fair value every quarter.
The value of the liability depends on Topicus's recurring revenue, its net assets and the share prices of Topicus's Polish investments (Asseco and Sygnity). If the business performs well, the liability increases – and the resulting expense reduces the group's reported profit.
Constellation deducts this revaluation expense in full in its FCFA2S metric, even though no cash has been paid. This is remarkably conservative – but for our valuation, it is nevertheless not the appropriate benchmark for two reasons.
First, it is not a recurring payment. Even if Joday exercises its right, the amount is payable only once. Deducting it year after year from an ongoing cash flow measure that is subsequently capitalised using a multiple would effectively carry a one-off payment forward into future periods.
Second, it remains uncertain whether and when the payment will occur at all. The put right has existed since January 2021 and has not been exercised to date. In its disclosures, Constellation merely assumes, for calculation purposes, that Joday exercises its right for 100% of its holdings.
We therefore use Equity FCF: operating cash flow less capital expenditures, lease repayments and minority interests – without the IRGA deduction. For 2025, this results in USD 2.29 billion compared with USD 1.68 billion under FCFA2S.
The "AI" Risk – Why We Believe in the Niche
Between July 2025 and February 2026, Constellation lost approximately half of its share value, and the trigger was almost exclusively concerns about artificial intelligence. The reasoning behind this is plausible: If AI agents drastically reduce the cost of building customised software, Constellation's subsidiaries could lose their economic foundation.
In our view, however, the decisive point is different. AI does not make software worthless; it changes what is scarce. Code, development capacity and the capital required to obtain both become cheaper. As a result, the assets that AI cannot provide become relatively more valuable: industry knowledge, customer relationships, trust and established workflows.
This is precisely what Constellation's products are built upon. Leonard once described vertical software as a conversation between vendor and customer conducted over many years, during which implicit industry knowledge is translated into highly specific digital processes. AI can rewrite the code of a utility billing system. But it does not know the regulatory exceptions of individual states that have been incorporated into that system over two decades. Added to this is the economics of the niche: It is difficult to raise venture capital to disrupt a ten-million-dollar market.
The second line of defence lies with the customer. Around half of group revenue comes from the public sector – highly regulated, documentation-intensive, heavily dependent on regulatory requirements and technologically conservative customers with long procurement cycles. An administrative employee has no incentive to replace a functioning system with a new one and assume the associated migration risk. Where certifications are also required, as in US healthcare, the barrier for a new provider rises significantly further.
So far, the empirical evidence does not contradict this view. At the Annual General Meeting in May 2026, CFO Jamal Baksh reported that he monitors organic growth across approximately 900 business units and investigates anomalies – so far, none have been linked to AI. Jeff Bender, head of the Harris Operating Group, added immediately afterwards that there was still no concrete evidence of significant AI-related customer churn – although he expected it to emerge eventually.
Two developments would change our assessment. First, the pricing structure: If AI replaces human users, revenue from per-seat billing will decline – Bender himself identifies this as a cannibalisation risk. Second, an AI agent could position itself between the customer and the system, weakening the direct relationship and thereby shifting pricing power. As long as this does not happen, the niches remain defensible.
Further Risks
The Size Problem. In the coming years, increasing cash flows will have to be invested at minimum required returns. Chief Investment Officer Bernie Anzarouth openly acknowledged at the same Annual General Meeting just how difficult this has become: Around USD 3 billion in uninvested funds were available, and deploying them was proving difficult – despite the large number of available software companies. This is precisely what gave rise to PEMS (Permanent Engaged Minority Shareholder): significant minority investments in publicly listed software companies, including a seat on the Board of Directors, instead of full acquisitions. Asseco and Sabre are the first examples.
Key Person and Cultural Risk. Leonard was the intellectual architect of the capital allocation strategy. Whether this discipline can be maintained without him and at the company's current scale is the central unanswered question. The post-Leonard era has only just begun.
Transparency and Balance Sheet Complexity. There have been no regular quarterly calls since 2018, and shareholder letters have become less frequent. At the same time, spin-offs, minority interests and the IRGA liability make the financial figures considerably more difficult to understand.
What this means for us as investors: The share price decline of approximately 50% has brought expectations down from flawless execution to a moderate level. This is precisely where the appeal lies: We are getting an exceptional business model at a price that already reflects a meaningful degree of uncertainty. The other side of the equation is equally clear – the AI question remains unresolved, and nobody can provide a credible answer to it today.
Valuation Scenarios 2030 – Own Estimates
As a basis, we use 2025 Equity FCF of approximately USD 2.29 billion. Both scenarios are deliberately conservative: Growth of 10% and 15% per year, respectively, is significantly below historical growth and takes declining reinvestment returns into account. The exit multiples of 18x and 22x, respectively, are also below historical valuation levels and leave room for the unresolved AI question.
Figure 1: Stock price of Constellation Software Inc.; found on www.tickr.com
This article does not constitute a buy recommendation or investment advice.
Ott, Hagen & Partner AG
© Copyright Ott, Hagen & Partner AG 2026. All rights reserved.