Deep Dive Brief: Constellation Software and AI - Part #2

Deconstructing the AI Threat and Opportunity for Constellation Software

Sep 20, 2025

Expanse Stocks x Finchat (now Fiscal AI) Partnership!

🎁 Get 20% off + 2 months free on any plan! →🔗 Claim Discount

Welcome back Explorer!

In this thirteenth edition of my Deep Dive Briefs series and second installment of the Constellation Software (CSU) trilogy, I’m continuing my research journey on this serial acquirer, this time, we’re diving into a big, hot topic:

After the success of my first installment on this CSU trilogy, 🔗 Constellation Software Capital Allocation Outlook - Part #1, the next topic to cover felt pretty obvious to me, but also harder to assess.

One of the most frequent questions investors ask themselves these days revolves around Artificial Intelligence and its impact on established business models. Every industry seems to be in the crosshairs, and investors are rightly asking: who are the winners, and who are the value traps?

So, where does a company like Constellation fit in?

For years, CSU has been the gold standard of disciplined capital allocation, a quiet compounder that buys and holds hundreds of "boring" vertical market software (VMS) businesses. It’s the antithesis of the typical high-growth, cash-burning tech story. On the surface, its vast portfolio of legacy software seems like a prime target for disruption by nimble, AI-native startups.

After digging into this topic, my conclusion points in the opposite direction. I believe Constellation Software’s model is not structurally at risk from AI; rather, AI is an accelerant *if* managed correctly.

Topics We’ll Cover

🔹 A Quick Refresher on the CSU Machine
🔹 The AI Disruption Narrative
🔹 Mapping the AI Battleground
🔹 The "Vertical AI" Thesis
🔹 The Potential Evolving Playbook
🔹 The Real Risk Isn't AI Disruption
🔹 Comparing CSI with its Spin-Offs
🔹 Conclusion & What I'm Watching

This article couldn’t come at a better time. Constellation Software and Mark Leonard just announced an open-discussion shareholder meeting on September 22nd at 9:00 E.T. Management, along with their AI specialist team, will be walking through how AI could impact their business. Exciting, to say the least, don’t you think?

If you haven’t found the link yet, here’s the official announcement:

🔗 Constellation Software Inc. Announces Conference Call to Discuss AI’s Impact on Software Businesses

With that in mind, I thought it would be interesting to share my own take ahead of the call. That way, we can contrast my thoughts with management’s perspective. Let’s dive in!

📰 What’s New at Expanse Stocks?

📊 Quarterly Update - Portfolio Composition (by Industry and Geography) + Valuation Metrics → 🔗 Behind-The-Scenes [Free access]

📚 Articles

💸 General Investing
💎 Hidden Gems Series
✨ Annual Specials – Annual Letters, Investing Philosophy, Industry Write-ups & Top Picks of the Year
📚 Resources for Investors
📢 Latest Stock News
💬 Join My Chat → [📎 Learn more]

🔐 Paid Subscriber Exclusives

💼 Portfolio Corner – Holdings, valuation models, trades, performance & more!
🤫 Exclusive Sections – From select articles, Deep Dive Briefs & Hidden Gems

👀 Recent Releases!

👀 Coming Soon

💎 Unlocking Japan - Part #2: A Hidden Gem
✨ The Rareness of Moaty Industrials: A Selection of 6 Gems

A Quick Refresher on the CSU Machine

Before we can analyze the impact of AI, we must first appreciate what makes CSU so unique. Unlike horizontal SaaS giants that chase massive markets, CSU’s model is built on acquiring and holding mission-critical software for niche industries.

Think about the software that runs a local library, manages a municipal court's case files, or handles the billing for a rural utility company. This is CSU’s turf. These VMS businesses thrive on a different set of economics:

Deeply Embedded Workflows: This software is the central nervous system of a customer's operation. It's not a discretionary tool, it's a "must-have."
High Switching Costs: Migrating years of historical data, retraining entire teams, and ensuring regulatory compliance makes ripping out a VMS a painful, expensive, and risky proposition.
These switching costs aren't just financial, they are operational as well. When your software manages patient records, inventory, or regulatory compliance, the risk of something going wrong during a transition can be business-threatening. Why would you jeopardize your entire operation to save a few hundred dollars monthly?
Durable, Predictable Revenue: While the market for any single VMS is small, its revenue is sticky. Customers rarely leave, leading to high retention and predictable cash flows.

CSU's genius has been to create a decentralized M&A engine that hunts for these small, durable businesses, acquires them with cash flow at disciplined prices, and holds them forever.

The AI Disruption Narrative

The AI disruption story has become the boogeyman haunting every software investor's dreams. For Constellation Software and its spin-offs, this narrative goes something like this: generative AI will commoditize niche software, lower development costs will invite new competitors, and legacy code bases will become sitting ducks for AI-powered substitutes.

The Bear Case: Why People Are Worried

The fear around CSU and AI stems from a few key assumptions:

Commoditization: The belief that large, horizontal AI platforms from Microsoft, Google, or OpenAI could commoditize the specific workflows that CSU's niche software currently manages.
Lower Barriers to Entry: The idea that generative AI will make software development so cheap and easy that new entrants can quickly build competing products.
Legacy Code Risk: The concern that CSU’s older, less glamorous software is vulnerable to being replaced by slicker, AI-powered alternatives developed in house.

While this sounds good, the reality seems to point in the opposite direction:

Notes shared by user: @zeroxpectation on X

While these concerns are logical on the surface, they miss the fundamental nature of Constellation's moat and what makes its business model so durable.

Mapping the AI Battleground: Not All Verticals Are Created Equal

The first mistake is to treat AI as a monolith. It’s not.

Take the term “mission critical.” It gets thrown around in VMS discussions but often misunderstood. You don’t just “vibe code” the software that runs a medical practice or manages public transit. Writing code is just one ingredient. The real moat lies in domain expertise, proprietary data built over years, and deep knowledge of regulatory requirements and workflows. AI can accelerate coding, sure, but replicating those advantages is a different kind of beast.

That’s why I don’t see all CSU verticals as equally exposed. I group them into three buckets:

1. The Fortress Verticals
Public administration, healthcare. Here the moats are strongest. Regulatory compliance, data security (HIPAA, etc.), decades of trust, and slow procurement cycles create walls AI startups can’t easily breach.

2. The Contested Verticals
Automotive dealer management, field service management. These remain sticky systems of record, but a well-funded AI-native could chip away at specific modules (e.g. pricing engines, route optimization, etc.) CSU may need to adapt, but the core system isn’t likely to vanish.

3. The At-Risk Verticals
Hospitality, attractions management. Workflows are simpler, switching costs lower, user experience more central. An AI-native offering dynamic pricing and personalized guest experiences could realistically displace legacy systems.

The takeaway: CSU’s “Fortress” verticals make up a large share of the portfolio and remain highly defensible. The rest may require more adaptation, but the narrative of “AI will wipe them all out” simply doesn’t hold water.

The "Vertical AI" Thesis: Why CSU Holds the Winning Hand

My optimism is heavily influenced by a framework I find incredibly useful: the "Vertical AI" thesis laid out on this article 🔗 The future of AI is vertical by Bessemer Venture Partners (a MUST READ), which argues that AI's most enduring value won't come from generalized models, but from specialized "co-pilots" and AI agents that are deeply integrated into specific industry workflows.

Winning in the AI era comes down to three things:

Data + Workflow + Distribution

This is exactly where Constellation’s edge shines. An AI agent is only as strong as the data it learns from and the workflow it’s tied to. So ask yourself:

Who holds decades of proprietary, process-bound data across thousands of niche businesses? → CSU.
Who owns the mission-critical workflow software that customers can’t operate without? → CSU.
Who already has the trusted distribution channels and relationships to deliver it? → CSU.

An AI startup can build a flashy feature, but it can’t replicate the institutional knowledge embedded in a CSU business that has been serving a niche for 25+ years.

CSU doesn’t need to create the world’s best foundational model, it just needs to fine-tune smaller, efficient models on its proprietary data to solve real customer problems.

I expect some of these questions to come up in the upcoming AI shareholder meeting, and my hunch is that my assumptions won’t be far from how management sees things.

The Potential Evolving Playbook: AI as an Accelerant

AI will potentially force CSU’s capital allocation model to evolve, but it presents more opportunities than challenges.

First, the M&A playbook will adapt:

If the market wrongly buys into the "AI disruption" narrative, it could depress the valuations of the very VMS businesses CSU loves to buy. Constellation could find itself acquiring fantastic, durable businesses at even better prices from founders who fear a disruption that may never come.
On the other hand, AI hype could lead to inflated valuations, which, in turn, may also lead to an expansion of the deal funnel. A previously "stagnant" VMS business with a great market position and rich data now becomes an "augmentable" target. CSU can acquire it and inject AI to unlock entirely new avenues for growth and efficiency.

Second, and perhaps most importantly, is the opportunity for operating leverage and margins enhancement. This is a key offset to external pressures. AI can make CSU’s businesses more profitable by:

Automating Support: AI-powered knowledge bases and chatbots can handle a significant portion of customer support inquiries, reducing headcount.
Streamlining Services: AI can automate data migration and system configuration, shortening implementation cycles.
Increasing R&D Productivity: Generative AI for code generation can help maintain and enhance legacy codebases more efficiently.

For a $3 million annual revenue company, these types of operational improvements can really push margins. AI tools might enable additional efficiency gains, perhaps needing fewer employees through AI chatbots handling customer support, automated code documentation, or other productivity enhancements.

This shift from a playbook focused purely on cost discipline to one focused on AI-driven value creation may become the next chapter in the Constellation story.

If Mark Leonard and co. are as “paranoid about AI” as they claim to be, they’ve probably already started the CSU evolution in this direction. Again, I expect management to provide more color on this in the conference call.

The Real Risks

If there's an AI risk for CSU, it's not existential disruption, it's margin pressure. If customers start believing AI makes software cheaper to build, they may resist price hikes. CSU’s model leans heavily on pricing power from stickiness and switching costs, so a shift in perception about software value could slow margin expansion.

That said, CSU management has been consistent: they don’t rely on aggressive or systematic price increases to drive efficiency. Pricing power is there, but it’s not the main lever. If anything, AI looks more like a potential tailwind for margins than a threat as explained in the previous section.

The other concern is talent attraction. CSU doesn't position itself as a bleeding-edge tech company, it's fundamentally a capital allocation and operational excellence story. If AI capabilities become table stakes for retaining customers, CSU's subsidiaries need to adopt quickly enough to stay relevant.

Comparing CSI with its Spin-Offs

When I look at CSI versus its spin-offs Topicus and Lumine through an AI lens, the defensive characteristics become even more apparent.

Constellation Software sits in the sweet spot of maximum diversification to dilute risks. With hundreds of businesses across dozens of verticals, it's the ultimate hedge against any single AI breakthrough.

But, Topicus might actually have the best AI upside. European public sector customers are conservative adopters, but they're also underserved by existing AI solutions. TOI can add AI capabilities to citizen service portals, healthcare systems, and educational platforms. These are markets where trust and regulatory compliance matter more than cutting-edge features.

Lumine seems to be the wild card, operating in telecom and media where bigger platforms already compete directly. But even here, the niche focus on specific operator workflows provides protection that broader platform plays can't match.

Conclusion & What I'm Watching

Constellation Software is not immune to technological change, but it is far from being a victim of it. The company's core advantages: embedded workflows, proprietary data, and deep customer trust may not only be defensible in the age of AI but also the ingredients needed to build valuable, specialized AI solutions *if* used wisely.

The model may need to evolve, and, with it, R&D spend may tick up. But, in my view, the fundamental engine of acquiring and compounding durable cash flows remains intact, now with the added tailwind of AI-driven productivity.

As I monitor the company, here’s what I’ll be watching closely over the next 5 years:

Organic Growth & Net Revenue Retention: Is CSU successfully using AI features to drive modest price increases and upsell new modules, pushing NRR above its historical baseline?
Operating Margins: Are the promised AI-driven efficiencies showing up in margin expansion at the segment level?
M&A Activity: Is CSU successfully acquiring these "augmentable" targets, or is it getting priced out of the market? I’ll be watching the size, frequency, and returns on new deals.
Management Commentary: I’ll be listening carefully for any shifts in how management talks about the role of organic growth, M&A criteria, and the development of any shared AI platforms.

Thanks for following along,

—Nikotes

➕ Unlock Premium Content – For just $0.39/day ($12/month) or $0.27/day ($100/year)!

🔗 Full Portfolio Content - 🗓️ Monthly Updates (Last Update: 11-Sep-2025)

📈 Biweekly Report – Portfolio updates, recent moves & watchlist additions
💼 Full Portfolio Access – Holdings, valuation metrics & performance tracking
📊 Valuation Tools – DCF, reverse DCF, capital efficiency & growth models
🎧 Podcast Picks – Biweekly handpicked episodes on business & investing
🔎 Stock Picking Framework – My methodology & investing philosophy
🎯 Swing Trading – Short-term strategy & latest swing trades
💬 Private Community Chat - 📎 Learn more
🔐 Full access to all my articles