Constellation Software (CSU)

Constellation Software reported very good earnings today. As you may know, the company’s stock has not been immune to the _SaaSpocalypse_ narrative, but investors’ fears should’ve been calmed a bit post-earnings. For several reasons…

1. There’s no noticeable impact of AI in the numbers yet

2. Management provided interesting (albeit qualitative and subjective) comments as to what they expect the AI impact to be

3. AI fears have actually enlarged Constellation’s capital deployment opportunity

I believe there were several interesting things in the numbers worth discussing.

Revenue grew at a good pace in Q4 and FY 2025 and Constellation should be now considered as an “_accelerating top line_.” Interestingly, the company enjoyed somewhat significant margin expansion. Expenses grew 100 bps slower than revenue in Q4 and 200 bps slower for the full year:

What I found particularly interesting about the margin expansion was its source. “Staff” makes a disproportionate amount of the company’s expenses, and these grew considerably slower (+11%) than revenue (+15%) for the full year:

The optimist in me wants to believe that the differential here was driven by increased AI efficiency. This would mean that AI is currently being accretive to the numbers (which would be somewhat of a narrative violation for many). Now, the realistic in me is aware that Constellation comes from a period of slightly slower capital deployment through which the business typically gets more efficient (either way, good news for shareholders).

Investors probably had their eyes on organic growth to gauge the impact of AI. Organic growth (constant currency) was 2%, but organic growth (again, constant currency) in maintenance and recurring was 6%. This was in line with previous quarters:

Several things worth noting here. First, Mark Miller claimed during the call (first call since 2018) that AI has not driven an organic acceleration (at least not yet), but that it had not resulted in loss revenue either (i.e., it’s business at usual at Constellation for the time being):

I would personally completely ignore P&L earnings and focus on cash flows. P&L earnings were negatively impacted by the IRGA revaluation charge and the reclassification of the Asseco stake. Add amortization of intangibles, and P&L earnings become entirely misleading for CSU.

Operating Cash Flow grew a whopping 24% in 2025:

Free Cash Flow available to shareholders was a different story; it contracted in Q4 and “only” grew 14% in FY 2025. Now, as I discussed in my article ‘_Hidden Value in a Liability’,_ we should ignore the IRGA liability revaluation charge to correctly assess Constellation’s earnings power. The IRGA grew significantly in 2025 and pressured FCFA2S, which would’ve grown 27% YoY excluding this item, ahead of Operating Cash Flow.

All in all, the financials don’t seem to portray that this is a dying business, but in all fairness to the skeptics, the AI threat “promises” to be a long-term threat in nature rather than one of immediate disruption. FCFA2S excluding the IRGA was $2.12 billion in 2025. At Constellation’s current EV of $46 billion, the company is trading at an EV/FCF yield of 4.6% while growing cash flows above mid-teens and potentially accelerating capital deployment.

Let’s talk about this a bit more.

The main highlight (besides no noticeable AI impact) for Constellation was capital deployment. Let me briefly share some data. Constellation deployed (these are all-inclusive numbers, so CSU+TOI+LMN)…

* >$2.1 billion into acquisitions and investments in FY 2025. Considering that all-inclusive FCFA2S (excluding the IRGA and before NCI) was $2.37 billion, that’s a reinvestment rate of 90%. Not too bad.

Management was evidently asked about their thoughts on the AI threat. Even though they did not outright provide thoughts about the implications of AI in setting terminal values, I believe that the PMS strategy sort of confirms their view on the AI impact on VMS (i.e., the market might have gone too far in some cases). Mark Miller also opened the call with some thoughts on the topic:

I want to end this earnings recap with some thoughts on stock buybacks. There were some relevant news on this front. Constellation is now considering buybacks as a valid capital deployment option. This is a significant departure from Mark Leonard’s strategy. The only “problem” is that “they have a number,” but they are not yet hitting that number and have plenty of available opportunities elsewhere (i.e., they don’t think it’s in the best interests of shareholders to repurchase shares):

Now, many people typically rush to interpret these comments as an admission that the stock is expensive. I disagree with this for a very simple reason: Constellation’s hurdle rate is likely far higher than the hurdle rate of most Constellation’s shareholders. Considering that the company targets 20%+ hurdle rates, the fact that they are not doing buybacks here probably means that they consider the IRR on Constellation’s shares to be lower than 20%, which shouldn’t necessarily make the stock expensive if your hurdle rate is lower.

All in all, a good quarter from Constellation (and Topicus and Lumine).