The Number That Explains Everything
In the spring of 2023, Constellation Software completed its 100th acquisition of the year — and the year wasn't half over. The target was not a household name. It was not a hypergrowth SaaS darling with a $500 million ARR deck and a Tiger Global term sheet. It was, in all probability, a software company serving a market so specific — cemetery management, transit scheduling, dental clinic administration, agricultural equipment dealerships — that most venture capitalists would have scrolled past the pitch deck in under four seconds. The purchase price was likely between $2 million and $30 million. The integration would be handled by one of Constellation's six operating groups, staffed by a small team that had done this exact thing dozens of times before, working from a playbook refined over nearly three decades and more than 900 cumulative acquisitions. The seller, in most cases a founder approaching retirement or a private equity fund that had extracted what it could, received a fair price and the assurance that the business would not be gutted, rebranded, or merged into oblivion. The employees kept their jobs. The customers kept their software. The acquirer kept compounding.
This is the machine that
Mark Leonard built: a $75 billion company, listed on the Toronto Stock Exchange, that most people in technology have never heard of — a permanent capital vehicle for buying, holding, and gently improving small vertical market software businesses, doing it hundreds of times a year, across dozens of countries, in markets so obscure they barely register as markets at all. It is the anti-unicorn. It does not disrupt. It does not pivot. It does not pursue total addressable markets measured in the hundreds of billions. It accumulates. And the accumulation, compounded over 29 years and running, has produced one of the most extraordinary shareholder return records in the history of public markets: a share price that has risen from roughly C$1.65 at its 2006 IPO to over C$4,400 by late 2025 — a return exceeding 260,000%.
The question that Constellation Software poses to anyone who studies it is deceptively simple: Can you build a world-class business by doing something boring, extremely well, thousands of times?
The answer, it turns out, is the most interesting story in software.
By the Numbers
The Constellation Empire
$75B+Market capitalization (late 2025)
$9.8BTrailing twelve-month revenue (2024)
900+Acquisitions completed since founding
100+Acquisitions per year (recent pace)
125,000+Customers worldwide
49,000+Employees across six operating groups
~260,000%Total return since 2006 IPO
$0Dividends paid since 2017 (all capital redeployed)
The Venture Capitalist Who Stopped Believing in Exits
Mark Leonard does not give interviews. He does not appear on CNBC. He does not attend Davos. He does not tweet. For decades, his annual letter to shareholders — dense, self-critical, occasionally scathing, always analytical — was the only public window into the mind running one of the most successful compounding machines in the world. There are no authorized biographies, no profiles in Fortune, no TED talks. The Forbes billionaire page for him is spartan to the point of comedy. In an industry that rewards self-promotion with venture term sheets and conference keynotes, Leonard built a company worth more than Shopify by staying invisible.
He arrived at the insight that would define Constellation through the long way around. Born in 1956, Leonard trained as an engineer, then pivoted to venture capital in the 1980s, working for a Canadian firm investing in technology companies. He was, by the standards of the era, competent but frustrated — the hit-driven economics of venture capital grated against something in his temperament, the low base rates of success, the dependence on a single exit to justify a decade of losses. What he noticed, though, was the particular durability of one subcategory within his portfolio: small vertical market software companies. These were businesses that sold specialized applications to narrow industries — transit authorities, utilities, public safety agencies, clubs and associations — and once embedded in a customer's workflow, they almost never got ripped out. Retention rates north of 90%. Recurring revenue. Low capital intensity. The economics were extraordinary. The problem was that nobody cared. The markets were too small to attract serious venture investment, the growth rates too modest to generate IPO-worthy narratives, and the founders too often lacked the management depth to professionalize the operations. These were, in the argot of Silicon Valley, "lifestyle businesses." Leonard looked at them and saw gold.
In 1995, at age 39, he incorporated Constellation Software Inc. in Ontario, Canada, backed by a small group of investors, with a thesis so unusual that it barely registered as a thesis at all: he would buy these vertical market software companies, one at a time, hold them permanently, improve their operations using a standardized set of best practices, and redeploy their cash flows into more acquisitions. No exits. No platform consolidation. No "synergies" achieved by firing half the staff. Each acquisition would remain autonomous — its own brand, its own customers, its own management team — while benefiting from Constellation's growing institutional knowledge about how to run a small software business well.
It was, in essence, Berkshire Hathaway for software. Except that
Warren Buffett bought large businesses in mature industries and held them. Leonard would buy tiny businesses in niche industries and hold them — dozens, then hundreds, then over a thousand. The compounding engine would not come from the growth of any single acquisition but from the relentless redeployment of free cash flow into new ones, each generating its own cash, each feeding the next acquisition, the whole thing spinning faster as the organization learned to evaluate, close, and integrate deals with increasing velocity and decreasing friction.
We are not in the business of buying software companies. We are in the business of deploying capital at attractive rates of return, and software companies happen to be the best vehicles we've found for doing that.
Trapeze: The First Star in the Constellation
The very first acquisition tells you everything you need to know about the model. In the mid-1990s, Leonard acquired Trapeze Software, a company that made scheduling and dispatch software for public transit agencies. Trapeze had been cofounded by people from SAGE, a transit software provider acquired by Teleride — and one of those people was Mark Miller, a young engineer who had joined SAGE as an intern and stayed through the acquisition before leaving to start Trapeze.
Miller's career trajectory would mirror the company's. He became CEO of Trapeze Group after Constellation acquired it, then rose to Chief Operating Officer of Constellation itself, then CEO of Volaris Group — one of Constellation's six major operating groups — and finally, in 2025, President and COO of the entire enterprise when Leonard stepped down as CEO for health reasons. The pipeline from acquired-company founder to operating group leader to C-suite executive is not an accident. It is the system.
Trapeze was small, deeply embedded in the transit industry, profitable, and — critically — boring. No venture capitalist in Silicon Valley would have funded a transit scheduling software company in the 1990s. The total addressable market was limited. The sales cycles were long, involving government procurement processes that could stretch years. But once a transit authority adopted Trapeze, switching to a competitor meant retraining hundreds of dispatchers, migrating years of scheduling data, and risking service disruptions that could strand riders. The cost of switching dwarfed any conceivable savings from a competing product. Retention was essentially permanent.
Leonard paid a sensible price — a low multiple of earnings, financed largely from Constellation's existing capital — and left the management team in place. He did not attempt to merge Trapeze with another portfolio company. He did not rebrand it. He deployed a set of operational metrics — revenue per employee, maintenance revenue as a percentage of total revenue, customer retention rates, organic growth, return on invested capital — and used those metrics to coach, not command, the management team toward higher performance. This was the template. It would be applied, with remarkable consistency, over 900 more times.
How the first acquisition established the model
1995Constellation Software incorporated; Leonard raises initial capital from private investors.
~1996Acquires Trapeze Software, a transit scheduling business — the first acquisition.
1996–2005Builds portfolio through steady acquisitions, establishing the decentralized operating model and standardized performance metrics.
2006IPOs on the Toronto Stock Exchange at roughly C$1.65 per share; uses public listing primarily as a capital structure tool, not a liquidity event.
The Architecture of Obscurity
To understand why Constellation's model works, you first need to understand the market it operates in — or rather, the thousands of markets it operates in simultaneously.
Vertical market software (VMS) is a term of art for software designed to serve a specific industry's unique workflows. Horizontal software — think Salesforce, Microsoft Office, SAP — sells across industries. Vertical software sells within one. Cemetery management. Marina slip reservation. Court case management for county governments. Fleet maintenance for waste haulers. K-12 school nutrition tracking. The markets are so narrow that many of them have only three or four viable software providers serving them worldwide.
These businesses share a set of characteristics that make them exceptional acquisition targets. First, customer retention rates typically exceed 90% annually, and often exceed 95%. When a county government builds its entire property tax assessment workflow around a specific software platform, the cost of switching — retraining staff, migrating data, risking operational failures during transition — is orders of magnitude greater than the annual license fee. The software becomes infrastructure. Second, the businesses generate substantial recurring revenue through maintenance contracts, hosting fees, and now increasingly SaaS subscriptions, creating predictable cash flow streams. Third, they require minimal ongoing capital expenditure — the software is already built, and enhancements are funded from operating cash flow. Fourth, the markets are too small and too specialized to attract the attention of large platform companies. Google is not going to build cemetery management software. Salesforce is not going to enter the transit scheduling market. The obscurity is the moat.
But obscurity also creates a problem: these businesses are hard to find, hard to evaluate, and hard to buy. The founders are often technical experts who built the product themselves and have run the company for decades. They are not seeking venture capital or preparing for IPOs. When they want to sell — usually because of retirement, fatigue, or succession challenges — their options are limited. Private equity firms want growth stories and exit timelines. Strategic acquirers want to consolidate and cut costs. Neither option appeals to a founder who cares about the survival of the product and the welfare of the employees.
Enter Constellation, with a pitch that is singular in its simplicity: We will buy your company. We will pay a fair price. We will keep it running as an independent entity. We will keep your employees. We will keep your brand. We will never sell it. And we will help you run it better, using the operational playbook we've developed from hundreds of similar businesses.
That pitch, it turns out, is worth billions.
The Organizational Fractal
Constellation's corporate structure is unlike anything else in the software industry — or, for that matter, in most industries. It is a fractal: each level of the organization replicates the structure and operating philosophy of the level above it, all the way down to individual portfolio companies.
At the top sits Constellation Software Inc., the publicly listed parent. Below it are six major operating groups, each of which functions as an autonomous entity with its own CEO, its own acquisition team, its own portfolio of businesses, and its own P&L. These groups are:
- Volaris Group — the largest, with over 200 businesses spanning agriculture, communications, hospitality, and field service industries
- Harris Computer — focused on utilities, healthcare, education, and local government
- Jonas Software — serving fitness, hospitality, and construction industries
- Perseus Group — specializing in financial services, real estate, and insurance software
- Vela Software — a newer operating group focused on smaller "micro-VMS" acquisitions
- Topicus.com — spun out in early 2021, focused on European vertical market software, publicly traded on the TSX Venture Exchange (Constellation retains a controlling interest)
Each operating group, in turn, contains multiple business units and portfolio managers — sometimes called "operating group managers" or "portfolio leaders" — who oversee clusters of acquired companies. These portfolio managers have significant autonomy: they identify acquisition targets, negotiate deals, manage integrations, and coach their portfolio company leaders on operational improvement. The whole thing operates as a vast, decentralized network held together not by top-down directives but by shared metrics, shared incentive structures, and a shared cultural DNA transmitted through what Constellation calls its "best practices" — a continuously refined body of knowledge about how to run small software businesses.
The decentralization is not cosmetic. Portfolio companies retain their own brands, their own customer relationships, their own product roadmaps. A cemetery software company in rural Ohio and a transit scheduling firm in Melbourne may both be owned by Constellation, but they will never share a sales team, a product platform, or even a corporate email domain. The connection exists primarily at the level of capital allocation, operational metrics, and talent development.
Constellation is a software conglomerate that owns more than 500 vertical software businesses. What makes it remarkable is not just the number of acquisitions, but the consistency of the operating model applied across all of them.
This structure creates something unusual: an organization that can scale its acquisition volume without proportionally scaling its headquarters complexity. Because each operating group runs its own deal pipeline, Constellation can execute 100+ acquisitions per year without bottlenecking at the top. Leonard and his small headquarters team focus on capital allocation, cultural transmission, and setting the hurdle rates that determine which acquisitions get approved. Everything else is pushed down.
The Hurdle Rate as Philosophy
The single most important number in Constellation's operating system is not revenue, not EBITDA, not customer count. It is the hurdle rate — the minimum return on invested capital that an acquisition must clear to be approved. For most of Constellation's history, this rate has been in the range of 20–30% annualized, depending on the risk profile of the target. This is not a soft guideline. It is a hard constraint, enforced with religious discipline, and it explains both the extraordinary returns the company has generated and the existential challenge it now faces.
The logic is simple but ruthless. If Constellation pays 1x revenue for a software company generating 25% operating margins, and that company's recurring revenue base provides predictable cash flows, the payback period is roughly four years. After that, the acquisition is generating pure incremental return on the original invested capital, and those returns compound as the cash is redeployed into new acquisitions. If the company can also grow organically — even at modest single-digit rates — the returns improve further. The key is discipline: overpaying for an acquisition destroys the compounding math, because the capital redeployed into the next deal must clear the same hurdle, and every dollar wasted on an overpriced deal is a dollar that can't compound at 20%+.
This discipline has kept Constellation out of bidding wars. It has kept them away from "hot" sectors where multiples are inflated. It has kept them focused on the unglamorous, underloved corners of the software market where sellers are realistic about valuation because they have limited alternatives. And it has created a self-reinforcing reputation: founders know that Constellation will not offer the highest price, but it will offer a fair one, it will close quickly, it will keep the business intact, and it will never flip it. That reputation, compounding over decades, has become itself a competitive advantage — an informational edge and a reputational moat that is extraordinarily difficult for competitors to replicate.
But the hurdle rate also creates a problem. As Constellation has grown — from hundreds of millions in deployable capital to billions — finding enough small acquisitions that clear a 20%+ return threshold becomes exponentially harder. You can buy two hundred $10 million deals a year and deploy $2 billion, but finding two hundred deals that all meet your quality and return criteria is a monumental organizational challenge. The company has responded to this constraint in two ways: by dramatically increasing the number of acquisitions it completes annually (from a few dozen to over 100), and by beginning to consider larger acquisitions — deals in the hundreds of millions or even billions — where the competitive dynamics and return profiles are different.
This tension — between the discipline that built the machine and the scale that now threatens to overwhelm it — is the central strategic question facing Constellation Software.
The IPO That Wasn't Really an IPO
Constellation went public on the Toronto Stock Exchange in May 2006, but the IPO was less a coming-out party than an administrative decision. Leonard did not need the capital — the business was already profitable and self-funding. He did not want the publicity. What he wanted was a liquid market for employee shares, a currency for potential acquisitions, and a tax-efficient structure for his existing investors.
The listing was priced at roughly C$1.65 per share (adjusted for subsequent splits). There was no roadshow circus, no first-day pop, no CNBC bell-ringing ceremony. The company issued a prospectus, listed its shares, and went back to buying software companies. For years afterward, Constellation remained one of the most thinly traded stocks on the TSX, covered by almost no sell-side analysts, held primarily by a small group of value-oriented institutional investors who had done the work to understand the model.
This obscurity was, like so much else about Constellation, a feature rather than a bug. A low profile meant no activist shareholders agitating for "strategic alternatives." It meant no quarterly guidance game. It meant Leonard could write his annual shareholder letters — the Rosetta Stone for understanding the company — in the dense, self-lacerating style of a capital allocation philosopher rather than a corporate pitchman. Those letters, which run thousands of words and routinely address the company's mistakes as thoroughly as its successes, became cult documents in the value investing community.
We have been unable to find large acquisitions at reasonable prices. This is a failure of effort and perhaps of imagination, and we need to address it.
The stock's subsequent trajectory speaks for itself. From C$1.65 in 2006 to over C$4,400 by late 2025. A dollar invested at the IPO was worth over $2,600 nineteen years later. This performance — which rivals Berkshire Hathaway's best decades and exceeds the returns of virtually every technology stock over the same period — was achieved without a single transformative acquisition, without a single viral product, without a single quarter of hypergrowth. It was achieved through relentless, disciplined compounding at the smallest possible unit level, aggregated across hundreds of businesses, year after year after year.
The Employee as Owner, the Manager as Entrepreneur
Compensation at Constellation is, in the words of one analyst, "the gold standard" for aligning employee incentives with long-term shareholder value. The structure is deceptively simple but its effects are profound.
Operating group leaders and portfolio managers are compensated primarily based on the return on invested capital generated by the businesses they oversee. Not revenue growth. Not EBITDA. Not share price performance. Return on invested capital — the metric that most directly captures whether capital has been allocated wisely. A portfolio manager who acquires a company that generates a 30% return on invested capital over five years will be compensated handsomely. A portfolio manager who overpays for an acquisition — even one that grows revenue rapidly — will see it reflected in their compensation. The alignment is direct and unforgiving.
But the genius of the system is that it cascades. Operating group leaders are incentivized to identify and develop portfolio managers who can make good capital allocation decisions. Portfolio managers are incentivated to identify and develop general managers within their portfolio companies who can improve operational performance. General managers are incentivated to retain customers, grow recurring revenue, and run efficient operations. Each level of the fractal reinforces the level above and below it, creating a self-reinforcing cycle of talent development and capital discipline.
Constellation also requires its senior leaders to invest a meaningful portion of their own net worth in the company's stock, and restricts the use of stock-based compensation in favor of cash bonuses tied to ROIC. This means leaders are shareholders first and managers second — they feel the consequences of capital allocation decisions in their own portfolios, not just their annual reviews. The effect is a management culture that thinks in decades, not quarters, and that treats every acquisition as a personal financial commitment rather than an abstract corporate transaction.
Mark Miller — the engineer-turned-executive who rose from Trapeze intern to Constellation President — embodies this system. In a 2025 conversation, Miller described his own evolution from technical specialist to capital allocator as a process of learning by doing, guided by mentors within the Constellation ecosystem who had made the same journey. The pipeline from acquired-company founder to operating group CEO is not a recruiting pitch; it is the organizational metabolism.
The Spin-Off and the Scaling Problem
In February 2021, Constellation spun off Topicus.com, a collection of its European vertical market software businesses, as a separately listed public company on the TSX Venture Exchange. Constellation retained a controlling interest — roughly 30% economic ownership but majority voting control — while giving Topicus its own public equity currency and operational autonomy.
The move was revealing. It signaled that Constellation's leadership recognized a structural constraint: as the parent company grew larger, its ability to deploy capital at attractive returns in small deals was becoming harder, not because the deals didn't exist but because the organizational complexity of managing hundreds of tiny acquisitions across multiple continents from a single public entity was creating friction. By creating a separately listed European vehicle, Constellation could push decision-making and capital allocation closer to the markets where deals were being sourced, while maintaining the cultural and operational DNA of the parent.
Topicus was, in effect, a laboratory for solving the scaling problem. If it worked — if a smaller, more focused, locally managed entity could deploy capital at higher returns than the parent — the template could be replicated. More spin-offs, more operating groups with their own public listings, more nodes in the fractal, each one compounding independently while feeding cash back to the parent. The vision, never stated explicitly by Leonard but legible in the architecture, was something like a confederation of compounding machines — a decentralized capital allocation network where the center sets the philosophy and the hurdle rates but the periphery executes the deals.
Topicus has performed well since the spin-off, acquiring aggressively in European markets where Constellation's brand recognition was less developed and where the competitive landscape for acquiring small VMS businesses was less crowded than in North America. But it also underscored the central tension: even with the spin-off, Constellation's capital generation was growing faster than its ability to deploy that capital at attractive returns. The company was, in a sense, too successful — its cash flow machine was producing more raw material than the acquisition engine could process at its quality standards.
The Large-Acquisition Dilemma
For years, Leonard publicly agonized over Constellation's inability to complete large acquisitions. His annual letters returned to the theme with the insistence of a composer working a difficult passage — each year probing the same problem from a different angle, each year arriving at the same uncomfortable conclusion: the company had not yet found a way to deploy capital in $500 million or billion-dollar increments at returns comparable to its small-deal engine.
The math was unforgiving. Small VMS acquisitions — $5 million to $50 million — typically traded at 1–3x revenue, or 5–10x earnings, because the seller pool was fragmented, the buyers were few, and the businesses were too small and obscure to attract competitive bidding. Large software companies — $500 million and up — traded at 5–15x revenue, because they attracted attention from private equity mega-funds, strategic acquirers, and public market investors willing to pay premium multiples for scale. At those prices, Constellation's hurdle rate math fell apart.
The company's response was characteristically methodical. It began exploring "large deal" pipelines, assigning dedicated teams to evaluate potential acquisitions in the $200 million to $1 billion+ range. It developed internal frameworks for assessing whether the operational improvement opportunities in a larger business could compensate for the higher purchase multiples. And it signaled — through its annual letters and its structural decisions — that it was willing to be patient, to wait years or decades if necessary for the right large deal at the right price.
This patience was tested in 2023 and 2024 as Constellation deployed more capital than ever — reportedly over $2 billion annually — across a mix of small and medium-sized acquisitions. The company's pace of dealmaking accelerated dramatically, with some estimates suggesting over 100 acquisitions per year, a velocity that would have seemed impossible a decade earlier. But the average deal size remained relatively small, suggesting that the large-acquisition breakthrough had not yet arrived.
The strategic implications are profound. If Constellation cannot solve the large-deal problem, it faces a future where its returns on invested capital gradually compress as it is forced to deploy ever-larger amounts of capital across the same pool of small deals — a pool that, while vast, is not infinite. If it can solve the problem — if it can find or create a way to acquire large software businesses at returns approaching its historical standards — the compounding runway extends for decades.
The AI Question, or: Will the Boring Software Empire Survive?
In September 2025, the share price of Constellation Software had declined roughly 30% from its all-time highs. The proximate cause was twofold: Mark Leonard's decision to step down as CEO for health reasons, and a broader market narrative that artificial intelligence would destroy the economics of vertical market software.
The AI bear case goes like this: if AI tools radically reduce the cost of building software, new competitors can enter Constellation's niche markets at a fraction of the historical development cost, undermining the switching costs and competitive moats that have protected its portfolio companies for decades. Why stay locked into a 20-year-old cemetery management system if a startup can build a better one in six months using AI-assisted development tools?
Leonard addressed this argument directly on a September 2025 investor call, opening with a parable that revealed both his analytical rigor and his temperamental skepticism about technological determinism:
In 2016, Jeff Hinton made a long-term forecast... that radiologists were going to be rapidly replaced by AI. In the intervening nine years, the number of radiologists has increased from 26,000 to 30,500. Jeff wasn't wrong about the applicability of AI to radiology. Where he was wrong was that the technology would replace people. Instead, it's augmented people.
The point was characteristically subtle. Leonard was not arguing that AI was irrelevant to Constellation's businesses — he confirmed that most of the company's thousand-plus portfolio companies had been experimenting with AI tools for years, measuring productivity gains, determining ROI. What he was arguing was that the prediction — AI will destroy vertical market software — was the same species of confident technological determinism that has been wrong, repeatedly, about the pace and mechanism by which new technologies transform established industries.
The bull case for Constellation's AI resilience rests on three pillars. First, building industry-specific software requires deep domain knowledge accumulated over decades of customer interaction — knowledge of workflows, regulatory requirements, edge cases, and institutional preferences that cannot be replicated by a language model trained on general-purpose data. Second, Constellation's competitive moat is not primarily technological; it is relational and operational — embedded in customer switching costs, long-term support relationships, and the institutional inertia of organizations that have built their operations around specific software platforms. Third, to the extent that AI tools do reduce software development costs, they benefit Constellation as much as any potential competitor — the company's portfolio managers can use those same tools to improve their existing products, reduce development costs, and expand margins.
David Poppe of Giverny Capital, a Constellation shareholder, offered the most concise version of the bull case: Constellation's businesses represent "one of the last places AI entrepreneurs will target" because they'd "be fighting for" markets so small and specialized that the economics of disruption simply don't pencil.
But the honest answer — the one Leonard himself seemed to be gesturing toward — is that nobody knows. The range of outcomes is genuinely wide, and the timeline is genuinely uncertain. What Constellation has, and what most of its potential AI-enabled competitors lack, is 29 years of embedded customer relationships, 125,000+ customers, institutional knowledge spanning hundreds of micro-industries, and an organizational machine capable of adapting iteratively rather than needing to make one big bet right.
The Succession and the System
When Mark Leonard stepped down as CEO in 2025, citing health reasons, the market reacted as though the company had lost its animating force. The stock dropped. Analysts questioned whether the operational philosophy could survive without its philosopher-architect. It was a reasonable concern — Leonard's intellectual fingerprints were on every aspect of Constellation's culture, from its hurdle rates to its compensation structure to its acquisition evaluation frameworks.
But the succession itself told a different story. Mark Miller, the former Trapeze engineer who had risen through every level of the organization over nearly three decades, assumed the role of President and COO. The transition was not a surprise to insiders — Miller had been the operational backbone of Constellation for years, managing the largest operating group (Volaris) and overseeing a portfolio of more than 200 businesses. His career was the literal proof of concept for Constellation's talent development model: an engineer who joined a tiny acquired company in the 1990s, learned the operating playbook, ascended through progressively larger responsibilities, and eventually took the reins of the entire enterprise.
The question the market was really asking was not whether Miller could run the company — anyone who had followed Constellation closely knew he could — but whether the system Leonard had built was robust enough to operate without its creator. Whether the hurdle rates, the incentive structures, the decentralized operating model, and the cultural DNA would continue to compound returns without the gravitational pull of Leonard's intellect and reputation holding it all together.
The evidence, so far, suggests yes. The acquisition pace has not slowed. The operating groups continue to function autonomously. The pipeline of portfolio managers and operating group leaders — trained in the Constellation way, compensated to think like owners, embedded in the system for years or decades — represents a deep bench of institutional continuity. If Constellation is more than one person — if it is truly a system — then the succession should be a non-event, a seamless transfer of stewardship from the architect to the apprentice he trained for exactly this purpose.
If it is not — if Leonard was the irreplaceable variable — then the next few years will reveal it in the numbers.
The Silent Compounding Machine
There is a passage in
Constellation Software: Reaching for the Stars that captures something essential about the company's nature. Unlike most corporate success stories, Constellation's narrative resists dramatization. There is no near-death experience, no pivotal product launch, no iconic marketing campaign, no rivalry with a charismatic competitor. There is, instead, the daily work of evaluation: a portfolio manager in Volaris reviewing the financials of a Spanish home-care scheduling software company; a Harris Computer team dialing into a demo with a mid-sized utility in Queensland; a Jonas Software analyst modeling the recurring revenue profile of a fitness studio management platform in Germany. Thousands of these interactions, happening simultaneously, across dozens of countries, each one individually unremarkable, collectively constituting one of the most effective capital allocation machines ever built.
The closest analogue is not another technology company. It is Berkshire Hathaway — the comparison Leonard has always resisted but which the market has always imposed, and which is, in its essentials, accurate. Both companies are permanent capital vehicles run by capital allocation obsessives. Both eschew debt-fueled empire building in favor of disciplined cash deployment. Both rely on decentralized operating structures that grant autonomy to subsidiary managers. Both generate returns through relentless compounding at attractive rates rather than through transformative bets or financial engineering. The difference — and it is a meaningful one — is that Buffett's model depends on the durability of the businesses he buys, while Leonard's model depends on the velocity of the buying itself. Berkshire compounds through the longevity of great businesses. Constellation compounds through the repetition of good acquisitions.
This is what the 30% drawdown in 2025 tests: not whether Constellation's model works — 29 years of evidence suggests it does — but whether the model's output can continue to clear its own hurdle rate as the machine scales beyond the ability of any single person to understand it in its entirety. Whether the fractal can keep replicating. Whether the silent compounding machine can keep doing its quiet, unglamorous, extraordinarily profitable work in a market that has decided, for the moment, that only AI counts.
In a warehouse in Cedar Rapids, Iowa — the registered U.S. address of Trapeze Software Group, the first company Constellation ever acquired — transit scheduling algorithms still optimize bus routes for municipal agencies. The code has been updated hundreds of times since 1996. The customers are still there. The business still generates cash. And that cash, like all the cash Constellation generates, flows upward through the fractal, through the operating group, through the parent, and back out again — deployed into the next acquisition, the next small software company in the next obscure industry, the next star in the constellation.
The machine does not stop.
