Welcome to the eighth issue of “On The Radar”,

If you want to know what this series is all about, I recommend reading the first issue. Let me save you the click with a short summary: the goal is to share interesting companies that I’ve looked at but have not researched in-depth yet (at least not to the extent that allows me to gain sufficient conviction). Therefore, the “On The Radar” series should be understood as a well-curated list of companies that fit my investment philosophy and that might (in the future) make it into the portfolio. For you, it might be a great source of ideas!

Out of the 22 companies that have been profiled thus far in the series, two have made it into the portfolio, and both have generated substantial alpha compared to the S&P 500. The 22 “On The Radar” companies have on average returned 10% and have provided a mid-single digit alpha compared to the main US indices since the series was launched.

Today’s issue brings yet another three companies (one for free and the remaining reserved for paid subscribers) that I’ve been looking into or that I plan to look into in the coming weeks/months:

1. A global leader in in-store technology whose stock has almost halved despite record results (there’s a reason, of course)

2. A serial acquirer’s new roll-up trading below every price at which it has raised equity since 2025 to get the flywheel going

3. A picks and shovels play in one of the fastest-growing areas of oncology

Note that all of these businesses get included in the “On The Radar” tab of the Portfolio Management Tool with their respective price evolution since the issue.

As this is the eighth issue, free subscribers have had access to 8 companies (one for free per issue) whereas paid subscribers have had access to 25.

So, without further ado, let’s jump into the first company.

Vusion (EPA:VU)

Brief description of the business

Vusion (formerly known as “SES-imagotag“) is the global leader in electronic shelf labels (ESLs). You’ve most likely come across an ESL because they increasingly are replacing paper labels in stores:

Source: Vusion H1 2026 results presentation

The ESL is only Vusion’s entry point into the store, which is what potentially makes this business interesting. Vusion’s pitch is that it sells an IoT platform made up of labels, smart rails and AI shelf cameras, all connected to VusionCloud. This platform now runs circa 52,000 stores and 522 million connected devices, up 2.4x in a year.

On top of this platform sits what management calls VAS (value-added services), made up of software, cloud subscriptions, computer vision (Captana, ~200,000 shelf cameras) and retail media (through a recent acquisition). VAS revenue was €125M in H1 2026 (+39% YoY), of which €61M can be considered recurring (+73% YoY).

Despite the significant growth of VAS revenue, Vusion’s business model is still mostly project-based and hardware focused. The sales motion ultimately works as follows:

1. A retailer signs a rollout

2. Vusion installs labels across its stores over one to three years. Manufacturing is outsourced to BOE, the Chinese display giant that also owns ~24% of Vusion, plus a Jabil base in Vietnam for the US.

3. Vusion expands its relationship over time through VAS, taking advantage of its installed base.

The contract that has defined Vusion’s fate over the past couple of years has been Walmart in the US, signed in 2023. Walmart received 1.76M warrants, exercisable as its spending with Vusion went from $700M to $3B. Even though this ramp up has benefitted Vusion enormously (not only its numbers but also its credibility as an IoT supplier), the reality is that it can backfire through weaker 2027 (and beyond) numbers. I’ll briefly touch on this topic later.

Why it caught my attention

Vusion came up in one of my automated screens because the numbers and the share price seemed to be telling opposite stories. Despite Vusion’s strong growth track record (20%+ CAGR over the last 15 years as per management) continuing in FY 2026 and margins expanding considerably, the stock is down almost 50% from its highs:

This apparent discrepancy led me to dig deeper, and I ultimately found out that what the market seems to be pricing in is the end of the Walmart rollout. Initially expected to run into 2028-2029, management confirmed that the EdgeSense deployment at Walmart will be largely completed by the end of 2026. This pull forward likely created the rise to ATHs in 2025 and posterior drop as the rollout was deemed to be less sustainable (so much for forward-looking markets!).

Seeing that the valuation has derated due to a potential 2027 revenue cliff is something that I actually view positively, not only because it creates a duration opportunity but also because it demonstrates that Walmart saw value in Vusion’s offering.

Let’s imagine that 2027 is indeed a revenue cliff post-Walmart rollout (even though Vision’s pipeline seems to be filling the void quite nicely)...does a weak 2027 mean the business is worth 50% less? Likely not, which also doesn’t mean that the stock is currently cheap!

Vusion does seem optically cheap, though. At ~€1.9B of enterprise value, the stock trades at roughly 9x LTM adjusted EBIT (evidently EBIT is influenced by the Walmart US rollout). My research into Vusion is ultimately a quest to answer the following question:

Is 2027 a cliff, or just a pause before the next leg of growth?

Note that it might not even be a cliff in the sense that growth goes negative (which the multiple does seem to imply), but rather a revenue growth deceleration which honestly seems like a given at this point.

What I like

There are many things I like about Vusion, in no particular order...

• Clear market leadership. The company has a scale that is twice that of any competitor (the main ones being Hanshow, SoluM and Pricer), and its VAS revenue is several times their combined non-ESL revenue. This is important because scale seems to be a significant competitive advantage in its transition towards a razor-blade business model.

• The mix shift will most likely improve the quality of the business. VAS should reach ~€400M pro forma revenue in 2026 (including the In-Store Media acquisition), four times the levels of 2022, and management targets VAS at close to 30% of revenue in 2027. VAS carries a much higher variable cost margin than ESL, which is already beginning to become visible in margins (and it’s also more recurring).

• The penetration runway is long. Management estimates ESL penetration at around 20% of the market and much lower for computer vision and retail media. Walmart makes up a significant chunk of the market and Vusion has a head start relationship with the company, with future Walmart rollouts expected in Canada and Mexico.

• Capital-light manufacturing. Group-financed Capex is ~3.6% of sales as large rollouts are typically co-funded by customers (e.g., Walmart). This creates a “capital light” business mode, but also comes with large cash flow swings.

• Balance sheet and capital allocation. The company enjoys net cash of €197M even after the Walmart down-payment reversals and has a long-tenured CEO at the helm.

• Optionality in vision AI and retail media. This is already starting to become apparent in VAS, but it’s the next big wave of in-store technology. The fifth generation of its computer vision technology is set to scale in 2027 and AI is only as useful as the data one has (i.e., Vusion’s scale seems non-trivial).

So, all in all, seems like a pretty decent business trading for a low-double digit EBIT multiple.

What gives me pause

Even though the stock does seem cheap, the ~50% drawdown has happened for a reason:

• Customer concentration and the 2027 cliff. Walmart has driven most of the US growth (Americas/APAC +39% in H1 vs EMEA +6%). Replacing this contract requires a big step-up in orders in H2 which management still seems confident about. It does seem that the pipeline is filling up nicely, but the timing/magnitude are the two unknowns.

• Dependence on BOE, which is both the main manufacturer and the largest shareholder, with two board seats.

• The ESL itself is hardware. Chinese competitors compete hard on price, and the ESL can become a commodity if the platform doesn’t keep ahead.

What I am watching

The main thing I’m trying to do here is answer two questions:

1. Is the 2027 cliff real? It seems to be discounted in the stock price

2. What’s the opportunity ahead post-2027? Is the market large enough to continue growing for many years?

Let’s jump to the second company, honestly the one I like the most thus far.

Read more

Danaher reported Q3 earnings last week but the numbers were the least interesting part of the release. Q3 was Julie Sawyer Montgomery’s first earnings call as CEO (with Matt Gugino also debuting as CFO), so I (and probably many investors) was expecting a quarter in which the new management team would set the tone for what’s to come. Would expectations get a reset? Would we get a new strategy? We got a mix.

The quarter itself was okay. Danaher delivered a beat on the top and bottom lines, raised its EPS range for the year, and management alluded to an acceleration on a “questionable” metric. After having enjoyed respiratory tailwinds and not stripping these out from core growth, management started reporting core growth ex-respiratory once tailwinds turned into headwinds. You know how this works: tails I win, heads, you lose!

I’ll go over three things in this article:

1. A quick summary of the numbers (and a couple of things worth considering)

2. Why I found the call to be a good chunk of corporate BS and why the new CEO didn’t excite me one bit

3. Danaher and AI (the core of the article): what management had to say about AI, the autonomous lab, and the NVIDIA partnership

Without further ado, let’s jump right into the numbers.

Danaher’s Q3

Q3 was a decent quarter by all means (or at least what Danaher investors can consider “decent” after reset expectations). Danaher met top-line estimates, reported double-digit adjusted EPS growth (not all growth was organic), and strong cash generation:

Danaher hadn’t reported such strong revenue growth (+8.5% YoY) since 2021, but around 650 bps of this growth came from Masimo, which closed in Q2.

Core revenue (i.e., organic) grew 3%, with respiratory once again acting as a ~250 bps headwind. The “good” news (if you love adjusted metrics, that is) was that core revenue growth ex-respiratory came in at 5.5% YoY. This ultimately means that this metric has accelerated for three consecutive quarters:

The other good news was that core growth was relatively broad-based across segments:

• Biotechnology (core +5%): bioprocessing grew towards the upper end of MSD, slightly ahead of management’s expectations. Consumables revenue grew MSD while equipment revenue grew only LSD. Good news could be found in orders (again). Bioprocessing orders grew mid-teens (led by upstream consumables and resins), with equipment orders continuing to grow at a “strong double-digit range“ year-to-date.

• Life Sciences (core +4.5%): Q3 was the third quarter in a row in which Life Sciences beat management’s internal expectations. Who would’ve said this a year ago? Probably not many.

• Diagnostics (core +0.5%, +7% ex-respiratory): respiratory was a 650 bps headwind (around $350 million vs around $500 million last year). Cepheid’s non-respiratory business grew LDD and Beckman Coulter Diagnostics delivered its best quarter in more than two years.

Masimo continued tracking ahead of expectations, delivering HSD growth in its first full quarter within Danaher and tracking similarly YTD (you can read my Masimo article here).

Management proved to be relatively optimistic about the state of their end markets...

...which is a positive read-through for some portfolio companies like Judges Scientific (JDG.L) that have been punished by weak AcaGov (academic and government) demand.

The guide and the 2027 “starting point”

Management raised the FY 2026 adjusted EPS guide to $8.57-$8.62 (from $8.45-$8.60), mostly to reflect the Q3 earnings beat. Danaher guided for core growth ex-respiratory of 4%-5% (4.5% midpoint) in Q4. This would be a 100 bps deceleration compared to Q3 but it’s also true they are coming against tougher comps (4% in Q4 2025 vs 2.5% in Q3 2025). On a two-year stack basis, the Q4 guide implies 8%-9% growth, roughly in line with (or slightly better than) the 8% delivered in Q3.

They also trimmed their respiratory forecast yet again: they now expect ~$1.5 billion this year (around $100 million less than before and down from ~$1.9 billion in 2025).

The most relevant comment, however, was the “first look“ into FY 2027:

My first look into their first look is...”yikes.” I honestly didn’t expect that a new CEO/CFO combo would come up with an optimistic outlook in their first call, but this begs the question:

Tycho Peterson (Jefferies analyst) summarized pretty well what many investors were probably thinking...

The answer was pretty much “that’s what we’ve been growing at, and we want to go through the planning process.” So, again, why make and apparent sandbag when there’s no need to? No idea!

Management expects respiratory to land at ~$1.5 billion again in 2027, so (assuming an “average” flu season) core growth and core growth ex-respiratory should pretty much be the same number next year. On the one hand, this means the debate about the ex-respiratory metric goes away. On the other hand, it means that 4%-5% is the total core growth starting point for a company whose long-term ambition is HSD core growth (i.e., not great). It’s true that once Masimo comes into core growth then Danaher’s core growth should naturally accelerate, but let’s do some quick math to understand where next year could land.

Masimo generated $1.52 billion in revenue in FY 2025, so if it grows HSD (say 8%) in both 2026 and 2027, it would generate around $1.64 billion in 2026 and $1.78 billion in 2027. Since the deal closed in June 2026, Danaher will have only consolidated around $0.93 billion of Masimo revenue this year (roughly $90 million in Q2, $390 million in Q3, and $440 million in Q4), which means Masimo should add around $850 million in incremental revenue in 2027.

Doing the same exercise for StatLab ($250 million in FY 2025, closed in early Q4) adds another $220 million or so. On an estimated FY 2026 revenue base of around $26.3 billion, M&A should contribute roughly 4 points of growth in 2027. Adding this to the 4%-5% core growth starting point, Danaher’s FX-neutral total revenue growth would land somewhere around 8%-9% next year.

(Note these are my own estimates there might be a small overlap (around 25 bps) if management’s 4%-5% starting point already includes Masimo’s organic growth after it anniversaries in mid-2027.)

Not terrible on the surface, but organic growth has disappointed for a pretty long while and shows no sign of ticking up. This would be Danaher’s fourth year well below its HSD “aspiration” (some would call this aspiration a “dream” by now).

A lot of corporate BS in the call

To jump straight to the point...I thought a good chunk of the call was corporate BS, and the new CEO didn’t excite me at all, neither in the prepared remarks nor in the Q&A which seemed yet another prepared remark.

A good example of this was capital allocation. When asked about M&A priorities, the new CEO answered the following:

“Strategically compounding M&A” came up several times during the call and I honestly still don’t know what it means but sounds like ROIC is going to be low because “strategic assets” rarely come cheap.

The main problem with this is that this is likely not what investors wanted to hear after several years of disappointing growth. Danaher’s core growth was -1.5% in 2024, +2% in 2025, and is now expected to land at ~2.5% in 2026. What investors (speaking for myself here) want is self-criticism and a concrete plan on how growth is going to accelerate towards the company’s ambition.

To be fair, there was a small dose of the former, but this is ultimately criticism on what Rainer Blair did, so it’s “easy” criticism:

She also acknowledged that...

I think the intentions are the right ones, but the “how” was nowhere to be found. When analysts asked management what metrics investors should use to track progress over the next 12-18 months, the answer was “unambiguous share gain is our report card.“ Great corporate BS, but there were no targets, no timelines, and nothing we can hold management accountable for next year. I’ll give them that it’s still early and they might be becoming comfortable with the business.

Anyways, the call kicked off with a long preamble by Julie with a good chunk of corporate BS in which the main highlight was that she wants to pursue more horizontal integration across the OpCos (operating companies):

She claimed that this approach has already generated “several hundred millions of dollars of opportunity in a short time“ in Diagnostics, and she even said she’s willing to double pay sales incentives to make it work:

This makes sense on paper (customers buy across “3, s6, 12 operating companies“), and it’s likely the most tangible change she’s bringing to the table. This said, Danaher’s decentralized OpCo model has historically been one of its strengths, so it will most likely be a balancing act.

There was, however, a strong proof point of this strategy being put to work: the AI lab.

AI, the autonomous lab, and NVIDIA

AI was, by far, the most interesting topic of the call, the only one where management gave us something concrete, and likely the one that made Danaher’s stock rise the day of the release.

Before going into the details, it’s worth understanding how management frames the opportunity. AI matters to Danaher in two ways:

1. Externally: AI is changing how customers do science (and drug discovery in particular), creating demand for new products and workflows. This is something I discussed in an article I published a while ago and is the reason why I like the healthcare exposure

2. Internally: AI as a productivity tool layered on top of DBS (Danaher Business System)

Analysts asked Julie about AI’s implications for Danaher, and her answer was probably the only thing that was not corporate BS in the call. Her argument is that Danaher’s breadth becomes even more valuable in an AI world:

I could buy this argument. The thought process here is that if AI-driven science requires connecting the “brain” (models) with the “hands” (instruments, consumables, and automation that actually run experiments), then a company that sells all of these pieces and can make them talk to each other should be in a privileged position. Very few players have this breadth and AI makes the “connected Danaher” strategy discussed above much more tangible.

The autonomous lab

The first proof point is Danaher’s first AI-powered autonomous lab. Julie framed it in a way that should feel reminiscent of my Robotics Report:

There are two things I find interesting here. The first one is that the lab brings together six different OpCos, which makes it the most tangible example of the “connected Danaher” strategy. The second one is that the lab is being used internally first (to discover reagents at Abcam), with the idea of later making it available to third parties:

The “10x higher volume” claim is the part that matters most from our POV, but it’s worth being precise about what it means, though. The “volume” refers to the lab’s output (validated reagents produced at Abcam), not to what customers consume, and management also framed it as a “10x productivity enhancement.”

More output per unit of input could cut both ways for Danaher’s razor-blade model. On the one hand, if customers adopt autonomous labs and run many more experiments, they should consume more reagents and consumables. On the other hand, a smarter closed loop could reach the right answer with fewer failed experiments, which would mean fewer consumables per discovery. My guess is that the first effect wins (cheaper experiments usually lead to more experiments...Jevons Paradox anyone?), but we’ll have to wait and see.

But what are “reagents”? In Abcam’s case these are mostly research antibodies and assay kits (i.e., the tools scientists use to detect and measure a biological target in their experiments). They sit at the very start of the drug development chain (discovery), well before manufacturing (Cytiva) comes into play. Developing a reliable antibody is slow trial and error, and AI-driven discovery needs exactly this kind of high-quality wet-lab data. If Danaher can produce it faster than anyone else, it strengthens Abcam and supposedly positions Danaher as the go-to partner for AI-native drug developers.

And in comes Nvidia

About a week before earnings, Danaher announced its collaboration with NVIDIA. The goal of the collaboration is to basically make the autonomous lab possible...

...but not only that. It involves:

• Running NVIDIA DGX Spark hardware at the edge (i.e., next to the instruments) so that data can be processed where it’s generated

• Using NVIDIA RAPIDS libraries to speed up data processing

• Evaluating foundation models like NVIDIA BioNeMo for real-time analysis

• Exploring NVIDIA Cosmos world models and Omniverse digital twins to simulate instrument behavior and lab conditions

The lab is a Danaher-run showcase at Abcam, whereas the NVIDIA partnership is about embedding AI into Danaher’s instruments themselves (so they can process data and adapt experiments in real time) and building digital twins to simulate experiments before running them. In theory, this could extend across the instrument portfolio and into customers’ labs, not just Danaher’s own. Management framed both as part of the same “autonomous science” vision, but for now the NVIDIA work is a pilot on a limited set of instruments.

Now, it’s important to keep expectations in check (Danaher investors know this by now). The initial pilot covers a limited set of instruments (not the whole portfolio), no financial terms were disclosed, and customers will be free to choose their own AI models and data standards (i.e., they won’t be locked into NVIDIA’s stack). This last point is relevant because Danaher seems to be positioning itself as an open, “plug-and-play” hardware layer rather than trying to build a walled garden. Julie also mentioned that Danaher is working with Anthropic on standards to make instruments easier to connect to AI agents and to other instruments:

The common thread is interoperability. If Danaher’s instruments become the easiest ones to plug into AI workflows, then customers would have a reason to standardize their processes on Danaher.

This seems like something encouraging but also something that will take a bit of time to materialize in the numbers. Management was honest with it at least:

The early demand is showing up in automation at Beckman Coulter Life Sciences, which saw “increased demand for AI-enabled solutions and services“ during the quarter. Management also mentioned that they’re approaching the new “tech bios” (AI-native drug development companies) as a connected Danaher from the start, and that they like the traction they’re seeing.

I’ll admit that this was the part of the call that I enjoyed the most (or maybe the only one haha). Unlike the strategic priorities that sound like corporate BS, the AI strategy does seem tangible. I do believe Danaher is one of the best-positioned companies in the industry to benefit from AI-driven science (the downstream healthcare industry seems like a good place to be), but TBD when we’ll see this in the numbers.

Final thoughts

Danaher’s Q3 can be summarized in one short sentence:

Nothing in this quarter changes my view of Danaher as a defensible, high-quality business, but Julie did not give me a reason to get more excited about it either.

Have a great day,

Leandro

The stock ran significantly above $600 after Q2, so August Troendle (Medpace’s founder and CEO) believed it was a good idea to sell tens of millions worth of his shares while not repurchasing a single share in Q3. Something that seems obvious (not repurchasing while the CEO is selling a large amount of stock) is something that other companies do regularly. One could go as far as to say that many management teams are using shareholder’s money to repurchase their own shares...but that’s not common practice at Medpace.

If we combine the lack of material repurchases with Medpace’s cash-generative nature (enjoys negative working capital dynamics), we get a cash pile that keeps on growing: from $497 million at the end of 2025 to $705 million at the end of September. And this growth comes after having spent $295 million on buybacks in Q1 at an average of roughly $417 per share (or 36% lower than where the stock sits today).

The chart below is always great to judge whether a management team knows what its business is worth. Needless to say, Medpace’s management does seem to know what their business is worth and uses buybacks opportunistically:

Now, I was slightly wrong with Medpace this quarter. Insider selling, a rising cash pile, and an elevated stock price made me think that management wanted to “create” a dip (i.e., that they would not be precisely optimistic during the call). This did not happen to the extent I thought I would, but they’ve nevertheless given themselves plenty of firepower for when it eventually comes.

Effective October 20th, the board terminated the old repurchase authorization ($527 million remaining and set to expire at the end of 2026) and approved a new one of up to $1 billion that runs through December 2028. That’s roughly 5–6% of the current market cap, and the cash on hand already covers about 70% of the authorization. Owning a cash generative business with a management team that deploys capital into buybacks opportunistically is an underrated advantage through drawdowns. Medpace’s stock might fall significantly, but you know that management will be “returning” cash to shareholders at the best possible time (evidently this assumes the business has not been impaired).

This pattern has held for Medpace for a long time: when there’s a sell-off, the buyback king strikes again. What made this quarter interesting is that it was somewhat the opposite of what we’ve been accustomed to. The leading indicators finally turned while the lagging indicators started to slow down, which means that the feared revenue cliff might have to wait a bit longer.

Let’s get on with it.

The numbers

The headlines looked mixed on the surface, with a relatively “weak” top line and extremely strong margin performance. Revenue grew 8.5% to $716 million, the slowest growth in quite a while and minimally outpacing expectations. This shouldn’t have been a surprise, though, as management’s guide already implied a sharp slowdown in H2 which will supposedly continue in Q4 (comps have been getting tougher through the year):

Part of the deceleration is “optical”, though. Reimbursed out-of-pocket expenses (pass-through costs) grew only 3.8%, but if we ignore them we can see that service revenue grew about 12%, the 6th consecutive quarter of double digit service revenue growth:

The question here is:

Because they are significantly dependent on project phasing. These are costs that Medpace pays on the client’s behalf while a trial is running. As Medpace always bills them back at cost, they count as revenue but carry no margin (in any other business these would be stripped out with the resulting figure being called “net revenue”). The nature of these costs ties them to the patient-heavy phase of a trial, which means that they ramp when many sites are enrolling and treating patients, but fall away as these close out.

Medpace experienced in 2025 a wave of large, enrollment-heavy studies (many of them metabolic as a result of the GLP-1 boom) which hit the “books” at the same time, creating a significant ramp up in pass-throughs. Now that those studies are maturing and the mix is moving toward oncology (more on this later), pass-throughs are normalizing. The good news is that service revenue (Medpace’s actual work in which the company makes a margin) keeps growing steadily at around 11–12% per year. In short: service revenue is a better gauge of the underlying business.

Does that mean EBITDA simply tracks service revenue? Yes to an extent, but it’s even better. Since pass-throughs carry no margin, EBITDA is ultimately earned on service revenue. This is why the reported EBITDA margin looks better when pass-throughs fall (24.0% this quarter). Now, the reality is that margins are expanding “organically” nonetheless. On a service-revenue basis the margin was about 40% in Q3, vs 39% a year ago and EBITDA has also grown faster than service revenue (+19.5% vs +11%) over the first nine months. That gap is operating leverage:

The caveat is accounting treatments. Medpace doesn’t book a contract’s fee as it bills it but rather books it in proportion to how far along the trial is. To measure the progress, the company uses the cost-to-cost method: costs incurred so far divided by total expected costs. The key detail is that those costs include investigator payments, which are pass-throughs.

Let’s take a look at this with a simple example. Say a trial is expected to cost $100 in total, of which $60 is investigator payments (i.e., pass throughs). If a big wave of patients enrolls in one quarter and Medpace pays $20 to sites, the trial suddenly looks 20% further along. This means Medpace books an extra 20% of its service fee that quarter, even if its own staff did no more work than usual. The bottom line is that pass-throughs earn no margin themselves, but they can shift when service revenue is recognized. They pull it forward when enrollment is heavy and slow it down when site activity fades on a contract by contract basis. The total fee over the life of the contract doesn’t change, but the quarter in which it lands does.

The CFO somewhat pointed at this during the call:

So the bottom line is that, over a full year, EBITDA should broadly follow service revenue but there will be noise from quarter to quarter.

Profitability was a very pleasant highlight

Margins were one of the highlights of the quarter. Even though Medpace barely beat the consensus revenue figure, it significantly beat margin consensus.

EBITDA margin reached 24.0% (22.5% a year ago), the second-highest quarter ever since Medpace adopted ASC 606. This (together with a growing cash pile and higher interest rates), helped net income grow 23.7%, more than double the pace of revenue.

Management attributed the margin performance to good utilization, low attrition and revenue growing faster than headcount. The most critical aspect of all of this, though, is that they don’t see it as a one-off:

This is “provable” with the guide The midpoint implies Q4 EBITDA of about $170 million on roughly $715 million of revenue or a margin near 24%. There’s one known headwind in 2027, though. The Cincinnati campus expansion adds about $20 million of operating costs (roughly 0.7 points of margin on ~$3 billion of revenue).

The leading indicators

The part that we should care more about is what’s coming, and it was an excellent quarter in terms of leading indicators (and forward-looking commentary). Let’s start with some highlights...

• Net new awards hit an all-time high of $873.4 million, up 10.6% YoY. Trailing twelve-month awards reached $3.02 billion, up 23.9%.

• Book-to-bill came in at 1.22x, the highest in at least seven quarters and up from 0.88x in Q1 and 1.12x in Q2. In Q2 I said the book-to-bill should go bananas. We are not quite at bananas levels yet, but we are getting there.

• Backlog grew 5.6% to $3.17 billion, and growth has accelerated every quarter this year (2.9% in Q1, 4.9% in Q2). We did not get any quantitative comment about the pre-backlog which, if you have read my prior articles, should know why it’s quite relevant.

• The slice of backlog expected to convert to revenue in the next twelve months is now about $1.99 billion, up roughly 8% YoY (from about $1.84 billion). This seems like a decent floor for 2027 revenue before any new award is considered.

Now, we got even more (and more positive) forward-looking commentary during the call.

2027 and the pipeline

I’d say that the most important part of the call was August’s commentary on the pipeline:

Simply put, August believes that the business is booming! Management added that decision timelines are faster than last year, and that clients want to start projects now, “not sometime in the future“.

Despite the very positive comments, an analyst noted that August sounded a bit less upbeat about Q4 than he did about H2 last quarter (these sell-siders just can’t get enough!). August answer actually was even more positive than anyone could’ve expected:

So, what does this imply? Plenty of things, and all of them positive:

1. “Another record“ means Q4 net awards above Q3’s $873 million. Implied Q4 revenue from the guide is only about $715 million, so book-to-bill will most likely increase beyond its current level (1.22x). August doesn’t like this metric, but I’d say the market does.

2. “Trailing twelve-month awards keep rising“. Q4 2025 awards were $737 million, so a record Q4 replaces it with something at least $135 million larger. TTM awards would end the year near $3.16 billion, up around 19% YoY.

3. “More awards moving into backlog”. He expects awards entering backlog to keep rising sequentially, which means backlog growth should keep accelerating into 2027.

August, however, made the same caveat he always makes: cancellations. Everything ultimately depends on this metric which is something management does not have much visibility on.

No cliff and a source of upside

Management will only guide “officially” for 2027 in February, but August opened the call with this:

Consensus is at around 6% revenue growth for 2027, which (you guessed it) would be a pretty significant deceleration compared to 2025. Now, there’s a pretty large caveat here because management ended up admitting that this view is supported by what they have today in the backlog (haha, classic). This ultimately means that if Q4 is strong and the current environment continues, then Medpace could experience significantly stronger growth in 2027 than what the market expects:

August continued:

Given everything else on the call, I think it’s pretty reasonable to think that “moved“ in this context means “moved upward“.

So here’s what we know about 2027. Management’s internal numbers are roughly at consensus (~6%) BUT (seems like a pretty relevant “but”) rising month to month. The next-twelve-month backlog is already up ~8% YoY, so if Q4 is a record and cancellations behave, the bar looks low.

There’s also a decision that doesn’t fit with falling demand...management’s decision to expand the Cincinnati campus in 2027. This is maybe reading too much into it, but you don’t spend $20 million on more office space if you expect demand to drop. An analysts also noted that headcount grew high single digits. This is probably the best signal of what management expects in terms of demand in a labor-driven business.

There was this fear (probably still alive to an extent), that Medpace would suffer a revenue cliff in 2027. The fear stemmed from two places:

1. The not so great book to bill and backlog growth.

2. The fact that Medpace had benefitted from a huge wave of cardio-metabolic trials

I believe both were “put to rest” this quarter. I’ve already gone over the first one, but what about the second? Management claimed that the mix kept shifting towards oncology:

Even in metabolic, things seem to be far from over. Notifications and awards are down, but August said RFPs show “quite a bit of metabolic work again“, with a lot of pending metabolic opportunities.

Full service, with some flexibility

There was an interesting discussion around Medpace’s current full-service model. One analyst asked whether clients are increasingly carving out biometrics or asking for FSP (functional service provider) models. August claimed he hasn’t seen a meaningful change and that Medpace still prefers full-service work although will take a project as long as it adds real value:

Pure FSP work is still a no-go for Medpace, but there’s some flexibility to cater to customer needs.

What about the valuation?

I’ve updated my three Medpace scenarios with a couple of changes.

Read more

The revenue update comes almost exactly a month after I discussed in Is Hermès “there” yet? whether the stock was finally cheap. Headline organic growth slowed to +4.3%, but the details tell a more interesting story. We are going to probably be bombarded with...

...because it’s the “easy” take (and who knows, maybe even a reasonable one in hindsight), but I believe this misses a good chunk of the story (hopefully you’ll understand why after reading this article).

I’ll start by outlining the numbers of the quarter, by region and by métier. Then I’ll go through what I believe matters most from the call and the qualitatives, ending with what I think about the whole situation.

Without further ado, let’s get started.

The numbers

Hermès grew Q3 revenue +4.3% at constant currency to €4.1bn. This might not seem impressive in and of itself, but I believe it’s pretty decent in the context of what the luxury industry has been going through:

Hermes’ largest peers (mainly LVMH, which is likely a better barometer of the industry’s health) continue to grow slower despite enjoying pretty easy comps from a lackluster 2025 in which Hermes was growing close to double digits.

Reported growth was +5.5% and the fact that it was higher than constant currency growth is more relevant than many people think. Q3 was the first quarter this year where FX helped rather than hurt the numbers. Q3 sales were roughly equal to Q2 in absolute value as well, portraying that Hermes might have reached some kind of stabilization. Management confirmed that there had been no change in trends (i.e., that things did not get worse) during early Q4:

This is encouraging considering that things had indeed gotten worse from Q2 to Q3.

Let’s go with some highlights.

1. Leather accelerated despite the broad-based slowdown

In my view, leather is the métier that best represents Hermès and its investment thesis because it’s the metier that remains the most in management’s control. Leather grew +10.8% organically in Q3, ahead of the +10.2% nine-month pace (wow). Management claimed that no single region drove the strength, that all regions grew (implying that China did as well), and that every bag produced was sold:

Ready-to-wear posted its first quarterly decline in about a decade and management attributed it to China and a tough comparison base after +80% growth over four years. Note that the overall tone was the following (I’m paraphrasing here or rather sharing how I interpreted it):

LVMH tells the opposite story in leather. Its Fashion & Leather Goods division fell 5% organically in Q3 as Chinese demand weakened, missing analyst expectations:

I don’t know exactly what analysts expected for Hermes’ leather metier, but judging by the questions asked in the call I’d say that it beat everyone’s expectations (including mine).

Management mentioned that leather supply keeps growing (rose about 6% in 2026) and that every single bag produced was sold. Hermes’ leather metier continues to be desirable.

2. The geographical breakdown and the wealth effect

The strongest regions were the Americas (+13.1% over 9M), Japan (+10.5%) and Europe ex-France (+9.0%):

That makes (to an extent) sense when we think about how luxury demand is created. Wealthy customers (or even the middle class in some countries) feel wealthier when equity indices and real estate valuations are at ATHs. Equity indices are at ATHs in Japan and the US, although I must say that Europe ex-France has been a surprise.

China is literally the mirror image. The country’s equity indices are far from ATHs and the company has suffered a property crash on top. Want to add more fuel to the fire? Let’s talk about taxes. China has introduced a new tax on foreign assets, which hits the wealthy customers who shop for luxury in other geographies. Management gave their POV during the call:

Those back taxes have to be paid quickly, which likely weighed on spending in Q3. Now, despite all the bad news in China, the country was stable for the first nine months and management alluded to stabilization running into Q4. Stabilization is not great (because it’s not growth), but I’ll take it.

There were two other pain points (geographically speaking). Firs, the Middle East (−4.3% over 9M). The Middle East was the only geography reflecting a decline due to the prolonged Iran conflict. And then we have France which has been all over the news lately.

France grew only +2.4% over the first 9M, and management credited the strengt to...

If you are not blind you might have noticed that tourist-heavy Paris isn’t on that list. Management shared that Paris footfall fell over the summer and blamed on the heatwave, claiming that September and early October improved. There were also some encouraging comments on the riots (especially after to what happened to Carrefour this week):

The open question and one that we are unlikely to have an answer to anytime soon is whether recent events keep tourists away from Paris/France in Q4.

3. The valuation: currency, and what one is really paying for

Q3 was the first quarter of 2026 where reported growth (+5.5%) outpaced constant-currency growth (+4.3%). After roughly €310m of FX drag over nine months, management’s tone sounded more confident about future currency headwinds (albeit they are totally unpredictable). Even though currency might seem unimportant, I believe it carries more significant implications than many believe it does (especially those taking valuation multiples at face value).

It’s exactly the scenario I described a month ago in Is Hermès “there” yet?. Hermès hedges its currency exposure to protect margins, but the company reports fewer euros of revenue and operating profit when the euro appreciates. When the euro gives some back, both metrics march higher even if organic growth is unchanged. As you might have imagined, the reported multiple is calculated on reported metrics, not on organic ones.

The stock currently trades at an LTM P/E of about 30x on LTM reported revenue growth of only 2.7%. Much of that apparent slowdown has been currency, not the business itself. As FX turns (who knows if it will next year), reported earnings rise with basically no change in business performance, so the multiple compresses mechanically at a given share price (and assuming everything else stays the same).

And we’ve seen this movie before but with an opposite impact. The euro slid against the dollar toward parity and briefly below it in 2022. This made Hermès’ reported figures run well ahead of the underlying business (i.e., the organic growth): revenue grew about 29% at current rates versus roughly 23% at constant rates, and recurring operating income rose about a third.

A more similar scenario took place in 2015, when the euro’s sharp fall against the dollar pushed reported revenue growth to around 17%, roughly double the ~8% constant-currency figure. In both cases a good chunk of the earnings growth came from translation rather than from selling more bags, which is precisely why reading a P/E without adjusting for FX (or the cash position) can mislead in either direction. Now, why is this tough to do? Because we ultimately don’t have a clue how FX will move (but at least we should be aware of it).

Of course, we’re now going to get bombard with people who claim that the multiple is too rich “for 4% growth,” but I don’t think this connection makes much sense. I don’t think many people own Hermes at current prices for 4% growth, they likely own it for the ability to keep growing when the industry is not doing great. Hermès has grown every year through four difficult years for luxury, and the company experienced acceleration in its core metier.

Still, all this said, there’s no denying that the luxury industry is going through an important bump and the shares will most likely remain out of favor for a while. We’ll see. I do think that the report was much better than the share price reaction implies, but maybe that’s just me.

Have a great day,

Leandro

Eurofins reported its Q3 and 9M revenue update today. European businesses are not legally required to report earnings every three months (the requirement is semi-annual) so they either (1) don’t report a single thing in Q1 and Q3, or (2) release a short revenue update. Eurofins falls in this second group.

The quarter was not incredible, but probably not too far from what the market expected. The earnings call was pretty encouraging (you’ll learn below why) which is likely part of the reason why the stock reacted positively despite having run up significantly over the past couple of months:

Note that here I’ll focus on the quarter (not the 9 months) in this article, as I published an H1 2026 article not long ago which you can below:

Let’s jump right in.

The numbers

The numeric highlight (in my view) was that growth continued accelerating (reported numbers were aided by an FX tailwind), but maybe not at the pace many expected. The company grew Q3 2026 revenue 6.1% to €1,913m (right around the market’s expectations), with organic growth accelerating to 3.5%.

Q3 was the third straight quarter in which organic growth picked up by half a point: 2.5% in Q1, 3.0% in Q2, 3.5% in Q3.

Even though the organic growth acceleration is welcome, it fell short of what the company required to achieve to meet its FY 2026 objective of MSD (mid-single-digit) organic growth. I believe this “guidance downgrade” is something that many people assumed back when Eurofins reported H1 2026 earnings. This is what I wrote in my H1 2026 update:

So, in my opinion, what occurred in Q3 served simply to confirm what most analysts had already anticipated and honestly to confirm something that seemed inevitable.

Just to put some numbers on this...Q4 organic growth would have to reach roughly 6.9% to hit 4% organic growth for FY 2026, 10.7% to hit 5% and 14.6% to hit 6%. Even the bottom of the range means doubling Q3’s 3.5% organic growth rate in a single quarter. This ultimately means that the guidance cut was a question of when, not if. These are my own estimates and, even though projecting organic growth is tough, it should be enough to demonstrate that the company needed a significant acceleration in H2 that the market did not really buy into.

The “good” news

There were a couple of pieces of “good” news on the growth front which I’d summarize as follows:

• Management claimed that FY 2027 organic growth should return to MSD. I don’t know if this will happen, but the organic growth trend is encouraging and so were management’s comments (more on them below)

Gilles framed the 2027 objective as follows:

• Inorganic growth is tracking ahead of management’s expectations, meaning that the combination of “organic + inorganic” growth might well end up in the same place despite organic falling short of their expectations

• Biopharma seems to have considerably accelerated in Q3, albeit not even management knows whether the acceleration is sustainable.

• A non-trivial portion of the organic growth headwind is self-inflicted and temporary. Eurofins is still rolling out its proprietary LIMS (laboratory information management system) across its network. Management claims this has been disrupting the labs and costing them growth but that, once the deployment is complete, that drag should fade and teams can refocus on winning new clients.

Management attributed the still subpar performance of biopharma to delays in clinical programmes (these sit in ancillary services).

Which is not something new...

In layman’s terms...the restart has moved from “soon“ to “H2 2026“ to “not in 2026.” Management also claims that projects can be lumpy and that the base effect has now been absorbed, which potentially means that growth can return pretty fast.

Gilles confirmed the delay today but sounded significantly more optimistic about 2027:

Something that seems relevant is that next year’s MSD target doesn’t assume these programmes kick in, so they’re optionality on top of what has been “guided” to:

Something that I don’t know whether to interpret as good or bad news was the geographical distribution of growth. The source of organic “weakness” remains concentrated in Europe (not a surprise), with North America trending within the MSD objective and the “rest of the world” far above it:

Management alluded to “macroeconomic” pressures in Europe which, as a European, I can somewhat confirm from anecdotal evidence! Things are definitely not trending in the right direction in Europe. The Rest of the world is growing close to double digits organically, and it’s no surprise considering that Gilles considers Asia to be a significant growth opportunity for Eurofins:

Gilles was also explicit on the call noting that they believe they are past the worse:

So, overall, a mixed story on the growth front but with some encouraging forward-looking comments. I’m still not done with these comments and I’ll share a bit more later on!

Still a profitability story

Even though we did not get profitability numbers, I’d consider profitability a highlight in the quarter. Despite claiming that they’d be under-delivering on the organic growth front and over-delivering on the M&A front (typically a bad mix for margins), management shared that they expect profitability and free cash flow to come in line with their expectations at the beginning of the year.

This must mean that the organic profitability of the business must have improved materially. Recall that Eurofins targets an Adjusted EBITDA margin of at least 24% in FY 2027 and they were already pretty close to it in H1:

Considering that H1 2026 came in close to 24% despite a poorer organic performance than what we should expect in H2, I’m inclined to believe that the 24% adjusted EBITDA margin objective is beatable this year. We’ll see, but Eurofins has been a story of underdelivering on growth and overdelivering on profitability for a while.

This has two potential readings, one optimistic and one pessimistic. The optimistic one is...what will happen to margins when/if organic growth accelerates? The pessimistic is...is Eurofins just a “milk the cow” story? Both are valid at this point in time although I still believe that organic growth re-accelerating to target is upon us.

The post 2027 period and AI optionality

Even though management stopped in 2027 when guiding investors numerically, they did give some qualitative commentary on what the post-2027 period might look like:

The only caveat I’d make here is that it’s not the first time that Gilles talks about the end of a Capex period which he then chooses to follow with a new one. He also discussed AI optionality in the call:

It seems like the company has enough physical capacity to enjoy this optionality even though it’s not baked into the long-term growth guidance. This should help us understand the magnitude of the capacity that the company has added in the most recent Capex period.

Updating the model

Gilles answer the following to a question on buybacks:

Eurofins is up around 30% since I published my in-depth report, which begs the question...is the stock still cheap?

Let’s take a look.

Read more

You can read this article (almost) entirely for free. If you like what you read, consider becoming a paid member to get access to…

• All in-depth reports (16 companies profiled thus far, and growing)

• Earnings follow-ups

• Other investment related content

• A community of like-minded investors

• Complete access to my portfolio and transactions

• Occasional webinars

You’ll also get access to the 30+ page detailed report on the robotics industry.

The in-depth reports of Stevanato and Deere are also free to read to gauge the quality of the research. Other in-depth reports have been written about…

• Deere (DE)

• FRP Holdings (FRPH)

• Five Below (FIVE)

• Diageo (DEO)

• Hermès (HESAY)

• Atlas Copco (ATCO.B)

• Stevanato (STVN)

• Keysight (KEYS)

• Zoetis (ZTS)

• Judges Scientific (JDG.L)

• Medpace (MEDP)

• Trupanion (TRUP)

• AAON (AAON)

• BBB Foods (TBBB)

• Eurofins (EPA.ERF)

• Sabre (SABR)

• Rosebank Industries (ROSE.L)

• Undisclosed position

• Undisclosed position

It was yet another interesting week in the market with “heat” appearing in the AI trade. I discuss what led to it and what lessons it can teach investors (myself included).

Without further ado, let’s get on with it.

Articles of the week

I published two articles this week. The first one was part 2 of my Rollins (ROL) vs Verisure (VSURE) series.

I go over Verisure and explain why I believe it’s a better business than it seems at first glance. I also explain how it compares to Rollins side by side; they are actually pretty similar businesses but at the same time very different (strange, I know, but I explain why in the articles). You can read part 1 here.

The second article of the week was my ASML earnings digest.

The company reported pretty good earnings. I go over those in detail and update my 2028 estimates with the new information management gave. I also triangulated what orders/backlog might look like through the first 3 quarters of the year even though we got no official reporting. You can find this in the X post below:

Without further ado, let’s see what the markets did this week.

Market Overview

It was a pretty negative week for financial markets. The S&P 500 (SPY) and the Nasdaq (QQQ) both fell pretty significantly, especially the latter:

The drop was much more pronounced in the Nasdaq because of its heavier exposure to the AI trade, and there were several culprits…

• Hotter-than-expected inflation data revived fears of rate hikes

• Financing costs for AI infrastructure kept rising with CoreWeave’s bonds now yielding over 15%. Remember that in a previous NOTW, I discussed how it was “strange” to see rates going up with the AI trade staying strong because much of the infrastructure needs to be financed with debt (and the cost of debt was rising). Now, this CAN happen if returns still stay above the cost of capital and if much of the debt is issued by investment grade players (i.e., the hyperscalers). I don’t know if this is the case for neoclouds.

The most relevant news, though, was that Anthropic gave investors its first real look at the numbers ahead of its IPO (an IPO which is now rumoured to be delayed due to AI safety concerns). Q3 revenue came in at $16.8 billion, up roughly 45% quarter-over-quarter, and the company claimed it was already profitable on an adjusted basis (adjusted doing the heavy lifting here). Impressive at first glance, but not by the standard the market had priced in. The reality is that growth is slowing and the IPO valuation being discussed reportedly dropped from $2 trillion to $1.5 trillion. When expectations are sky-high, “great” can still disappoint. A tale as old as time in financial markets. Stocks that had run up considerably YTD sold off hard even though the magnitude of these drops is never as visible in the indices as it is under the hood. The unwind, however, has been severe enough to put some very large funds under pressure.

Rumors circulated this week that the VARA AI fund had blown up. This has not been confirmed (a person familiar with the fund said it’s “doing fine”), but the story is fascinating regardless. At the end of June, VARA reported $26.4 billion in AUM (assets under management) just 16 months after launching (the kind of things you don’t see at market lows tbf), run by a team of around 9-20 people. The fund lost 44% in July alone on this portfolio…

The above brings several lessons in diversification. Even though the portfolio held 60+ companies, it was all exposed to the same factor. And leverage always makes things more interesting. The fund’s latest filing shows around $9 billion in regulatory AUM, while the 13F reports roughly $40 billion in notional positions (largely through call options on Google, Amazon, and Nvidia). That’s 4x+ exposure to a single theme. This is something that a lot of people heavily exposed to the AI theme will learn eventually: diversification by name count means nothing if every position is the same bet.

Whether or not VARA has actually blown up, there’s an important lesson here about Wall Street. Even after July’s loss, the fund was still up 65% for the year, so the early LPs did fine. The ones getting hurt are the ones who arrived after the triple-digit returns, which is precisely when the AUM ballooned. The managers, meanwhile, are already rich. Why, because they get paid based on the size of the pot, and the pot was the biggest right before the drawdown. Where am I going with all of this? The richest people in this industry are not necessarily those who add the most value but those who were in the right place at the right time. In asset management, it’s enough to win big once because incentives are terribly misaligned with LPs and sustained value creation (especially during “mania” periods), but it is what it is.

This isn’t the first time we’ve seen this movie this year. In July, Leopold Aschenbrenner’s Situational Awareness lost 67% in a single month, ran into margin calls with leverage of nearly 4x, and had to forcefully sell its entire public stock portfolio to Citadel. The fund was still up 80% for the year afterward and kept its private holdings. Some people are now calling Leopold “unlucky”:

Running a concentrated, single-theme book at nearly 4x leverage and getting margin-called when the theme turns is not bad luck. It’s the predictable outcome of the bet and those who are making it probably know it! I don’t doubt that Leopold is incredibly talented (who am I to say otherwise), and to his credit, he told investors he’d removed all leverage. But the system rewarded the approach handsomely right up until it didn’t, and capital rushed in at the top. One shouldn’t feel too much pity for these types of managers. They are very rich, and the people who lost money were sophisticated investors who should have known what they were investing in.

Does all of this mean the AI trade is over? I have absolutely no clue. Anthropic’s numbers suggest demand for AI is very real, especially since Anthropic is now the only player in the AI-trade. This said, this week was a reminder that the businesses themselves can keep performing while the leveraged bets on them get wiped out (it’s always been this way!). Interestingly, money rotated into defensive sectors like healthcare and staples (my portfolio was green with markets significantly red), and if the AI trade does eventually go under, we’ll find out, as Buffett put it, who’s been swimming naked. That’s precisely why I do have exposure to the AI trade but have not gone crazy over it (surviving is objective number 1).

The industry map was very red in tech and green pretty much everywhere else:

The fear and greed index dropped significantly and entered extreme fear territory:

Company-specific news

Read more

IBM is the fastest way available for customers to expand capacity and therefore it shouldn’t be a surprise to see IBM sales grow fast throughout supply crunches. This is good news for ASML and its shareholders because IBM sales allow ASML to profit from an upcycle relatively fast despite being a long lead-time business. IBM came in at €3.3 billion (€400 million above guidance) and management attributed a good chunk of the gross margin beat to it (no surprises here).

Increased volume is also helping ASML achieve considerable operating leverage. Operating margin reached 44.4% this quarter (up from 37.1% in Q2) and roughly half of the 730 bps gain came from Opex (not COGS), and nearly all of it from R&D...Sales grew 25% QoQ while R&D actually fell (from €1.28 billion to €1.17 billion), taking R&D from 13.7% to 10.1% of sales:

Even though this low figure has only presented itself in Q3 2026 and therefore shouldn’t be considered a trend, it’s something that shouldn’t surprise us because ASML is in the early stages of monetizing last decade’s R&D. The company has been investing heavily into its EUV program over the past decade and EUV is expected to ramp up considerably over the next couple of years. The way to think about this is that ASML has been running an R&D program over the last decade that it expects to monetize now.

This doesn’t mean that R&D will not go up anymore in our lifetimes. We should also be mindful about new R&D cycles like hyper-NA EUV, even though system commonality might help keep R&D proportionally low. I believe the “common platform” will be quite a big deal in terms of R&D and COGS leverage:

In short, ASML’s Opex could be considered somewhat of a fixed cost, meaning that sales increases typically fall to the bottom line at very nice incrementals.

I’ve also seen some people quote ASML’s strong EPS beat, but it needs context. Net income of €5.0 billion includes a one-off gain from Mistral AI...

Stripping this revaluation gain out, net income would’ve been around €4.45 billion and basic EPS around €11.60 rather than the reported €12.95. It’s an outstanding quarter nevertheless and even though the fact that the EPS figure (and therefore the P/E) are misleading seems so obvious, we shouldn’t underestimate many people’s ability to fixate on a P/E multiple or a headline beat!

ASML’s China business: warranted skepticism?

I was quite surprised to see a lot of people point out the China/DUV outperformance in Q3 (as a bad thing, that is). China went from 14% of net system sales in Q2 to 23% in Q3, and ArFi (immersion DUV) units jumped from 23 to 40:

Source: ASML Investor Presentation

I believe many are extrapolating these numbers to a new change in trend, but the reality seems very different. China accounted for roughly €4.0 billion of the €21.2 billion of net system sales YTD, which is about 19% of total sales. In other words, China is tracking right where management expected it to be, around 20% of total system sales in 2026. As EUV ramps up over the coming years, China as a percentage of total sales should moderate further going forward (it’s just a matter of mix).

I can totally understand why the market is worried about China, though. Let’s talk a bit about the MATCH Act. The MATCH Act is a new bill with bipartisan support that promises to make ASML’s shipments (and servicing) to China even more restrictive than it currently is. As one could’ve correctly imagined, The Netherlands is not precisely happy about it...

Dutch Foreign Trade Minister Sjoerd Sjoerdsma called the US MATCH Act coercive in June, while claiming that the Netherlands would favor “highly targeted proportionate measures“ of its own (i.e., they want independence to set restrictions, good luck with that). The MATCH Act wants to ban ASML from selling DUV machines to China and from servicing its installed base there. This would be a big step up from today’s restrictions. EUV has always been restricted in China, and since 2023-2024 only the most advanced immersion DUV tools (NXT:2000i and above, plus the NXT:1970i/1980i) need export licenses, so ASML can still sell mid-range immersion and dry DUV to China and service most of the installed base. The MATCH Act would extend the ban to virtually all DUV plus servicing which would be a double hit to ASML’s China business.

The good news is that the potential impact of this risk should diminish as EUV ramps up and the reality is that we still don’t know whether the MATCH Act will pass and under what conditions. So, I do believe that it’s rational for the market to be worried about ASML’s China exposure, but at the same time I find “interesting” how this was much more of a risk when the stock was trading at 52 week lows rather than near 52 week highs when in fact the potential impact of the risk got worse! Something something, price drives sentiment.

+30% and +30% is not enough

“Capacity” was once again the star word of ASML’s earnings release. Management laid out a plan in Q2 that called to raise unit output by 30% in 2027 and (aspirationally) by another 30% in 2028. Management shared in Q3 that this plan will probably not be enough to satisfy demand:

(This +30% and +30%, by the way, refers to both low-NA EUV and DUV immersion capacity.)

Due to ASML’s complex supply chain and its long lead times, 2027 unit capacity looks locked at +30%, while it does seem possible to expand capacity beyond +30% in 2028. This doesn’t mean that there’s no upside to 2027 numbers, but if there’s any, it’ll likely come from the other two capacity levers at ASML’s disposal:

1. IBM/upgrades: customers get fast incremental capacity within their existing footprint, and ASML is “very aggressively“ developing more upgrades. This is pretty profitable revenue, so a 2027 beat driven by IBM would once again fall through to the bottom line at pretty nice incremental margins

2. Productivity of new systems: the NXE:3800F (260 wph vs 230 wph today) arrives in 2027 “in small quantities”, with the real impact in 2028 and beyond. Critical immersion tools shipped in 2027 will also be upgraded to 330 wph. The bottom line is that, not only is ASML going to ship more units, but these are going to be more productive

There was a tone change this quarter as well in terms of the visibility of demand. Management shared that customers are telling them that they don’t expect their supply to meet their demand “at least until 2028.” I’m not sure what industry expectations were with regards to a potential supply/demand imbalance extending beyond 2028, but this doesn’t seem great for the “bottleneck thesis” that many have been pitching for the AI industry (i.e., “supply constrained beyond 2030“). It’s consistent with what I wrote in Q2, though:

Now, the real question is whether the bottleneck will be solved through a cooling demand or through a heating supply. I believe this matters because both situations end up in widely different economics for the supply chain. ASML believes it will be the latter (supply has to catch up because “demand visibility is getting stronger”):

Even though I believe this commentary is great and that it’s better to hear it than not to hear it (duh), I also believe that “demand visibility” has to be contextualized here. Industry participants talk about visibility of demand as if it were a given, but the reality is that future compute demand is all but a given and depends to a great extent on things outside of their control.

You can have your arguments in favor or against future compute demand being incrementally much higher in 2030 and also about where that value ends up accruing, but it seems obvious that orders being signed for 2029 and 2030 might evaporate if demand “falters” (although there’s no way of knowing the terms they were signed on). What seems pretty obvious right now is that the supply chain remains confident in the durability of the AI demand wave; whether that remains true in 2029 and 2030 when these orders are due is still TBD. I’m less worried about 2028 orders because they are pretty close and prepayments are likely being made, but thinking out to 2029 and 2030 seems like a bit of a stretch.

Another “problem” we have here is that we are currently getting only qualitative comments from ASML. We’ll get the backlog figure next quarter and then we’ll be able to really see how this positive sentiment is translating into the company’s order book. The good news for ASML is that its demand sources are now much more fragmented than they once were (Intel, memory...), BUT they are all dependent on the same AI growth thesis so really the demand source currently remains concentrated.

High-NA and metrology also show green shoots

The quarter brought a lot of news in terms of high-NA EUV, not all released in the quarter’s press release per se:

1. TSMC asked ASML to lead a consortium to bring 12-inch reticles (photomasks) to the industry sometime in the first half of the next decade (with all major customers joining). TSMC also confirmed it will introduce High-NA in HVM around 2030, which seemed a done deal considering that the investment in the 12-inch reticle is being made because they expect high-NA to run at scale

2. Samsung will start using High-NA in HVM in 2028

3. Intel is already running High-NA in HVM and is increasing the number of layers

It seems like the fears about high-NA adoption were greatly overblown (who knew?).

Metrology is also shaping up to become a source of upside as ASML continues to vertically integrate around its holistic lithography strategy. Management shared the following:

The new YieldStar 550B (overlay metrology) is ready for the 14 angstrom logic node and 1D DRAM and brings a 60% productivity improvement. Metrology & inspection was a €825 million business in 2025 (+28% YoY), so it’s still small in the grand scheme of things, but it’s growing fast as customers need more and more metrology to get performance out of advanced nodes.

In overlay metrology the main competitor is KLA (with its Archer platform) but ASML’s edge is “holistic lithography”: YieldStar data feeds directly into the scanner to correct overlay in real time, something a standalone metrology vendor can’t replicate to perfection.

Revisiting my 2028 exercise

Management upped the 2026 sales guide midpoint from €44 billion to €44.5 billion. This implies a 36% YoY increase which, according to management, is “largely driven by EUV.“ (the YoY increase, not the raised guide). The breakdown they gave for 2026 is:

• EUV: growing over 45% YoY

• Non-EUV (DUV + metrology & inspection): growing around 20% YoY

• Installed Base Management: growing around 50% YoY

This means that ASML is going to beat (almost certainly) the low end of its 2030 guide (€44 billion) in 2026.

In last quarter’s article, I shared an exercise in which I estimated that ASML could land around €70 billion in revenue in 2028 (there was a mistake in my numbers so I corrected that), 17% ahead of the high end of its 2030 guide, two years in advance. I recommend going back to that article to understand how the exercise was built because I’m going to revisit it below with the new information. The new info is the following:

• ASML is working towards a 2028 capacity increase of more than +30%. I’ll use +35% and +40% scenarios although I’ll anchor in the low end because it’s an aspiration and will be challenging to achieve. This translates into 115-119 low-NA EUV systems (vs my Q2 estimate of 110) and 228-237 immersion DUV systems (vs 220). Note that management specifically claimed that there’s upside to 2028 capacity but not for 2027. As 2027 will most likely continue to be a constrained supply environment, I modeled their comments through higher IBM sales.

• IBM is (not surprisingly) running much hotter than I modeled (€12.2 billion in 2026 vs my €11 billion estimate) and we might get a further boost in 2027 if unit volume is not enough to satisfy demand (see comment above). I’ll use 25% growth in 2027 (vs 20% I used in my Q2 exercise) and +20% in 2028.

• Pricing conversations are done and the full effect lands in 2028. Last quarter I used a 15% blended ASP increase, which solely reflected the more productive mix (management expected the mix to add ~15% of capacity on top of the +30% in units) and ASML pricing to value (i.e., it simply assumed that ASML charged the same for every increased unit of productivity). Outright price increases should come on top of that, so I’ll use a 20% blended increase for low-NA EUV (15% capacity increase + 5% pricing). I also added a 10% blended ASP increase to immersion DUV (some reports point to this being the ASP increase that ASML is already implementing on some customers, but we don’t have official confirmation from the company). For High-NA EUV I keep the same 15% ASP uplift I applied in Q2 as I doubt the company will raise prices outright so early in its adoption curve.

• Metrology is seeing “major, major growth“ although I’ll be conservative and will not update my numbers here

Everything else stays exactly as it was in Q2. I’ll keep dry DUV flat at €2.5 billion and metrology & inspection at €1.5 billion for conservativeness sake. Below you can find the three scenarios:

This gets ASML to €77-81 billion in revenue in 2028, 29-36% above the high end of the current 2030 guide (€60 billion, which likely gets raised in the next Capital Markets Day) and 17-23% ahead of 2028 consensus revenue of €66 billion (which in fairness has not been updated with the new info). Compared to my corrected 2028 estimate shared in Q2, the changes break down as follows:

• IBM: +€3.3 billion, together with low-NA EUV the largest driver (higher 2026 base and faster 2027 growth)

• Low-NA EUV: +€2.6-3.7 billion, split roughly evenly between more units and the outright price increases

• Immersion DUV: +€2.2-2.8 billion from more units and the 10% ASP increase

• High-NA EUV: unchanged at €5.75-8.05 billion, with the same 15% ASP uplift as in Q2

In terms of margins, I anchored on 60% gross margins last quarter. ASML is guiding for 56% this year (57.5% in Q3) before any of the price increases show up, and outright price increases should fall almost entirely to the gross margin line, so 60% now looks conservative but I’ll leave it as is.

Management guided for €6 billion in R&D and €1.7 billion in SG&A in their 2030 scenarios and I’ll keep these as is as well. Operating income at 60% gross margins and under this scenario would be somewhere between €38.7 billion and €41.2 billion.

With the stock at around €1,530 (~€590 billion market cap and ~€580 billion EV considering ~€10 billion in net cash), ASML is trading at around 14-15x its potential 2028 EBIT (60% gross margin case), compared to 18x with my Q2 assumptions. Valuing the business at 25x EBIT at the end of 2028, the potential IRR would be around 26-30%; at 20x it compresses to 14-17% (not considering dividends and buybacks, which should add 1-2%). This seems like a pretty appealing return, but there are certain caveats...

• I do think that ASML is a fairly high visibility company over the next few years due to its long lead times, but if there’s one thing that I’m sure about is that the numbers above are most likely going to be wrong in some way

• The increase “beyond +30%” in capacity in 2028 is something that’s under analysis, but ASML might not be able to conduct the capacity increase as they like. This means that the numbers I estimated in Q2 might be a better reflection of what ASML might achieve, although these should also not be considered a given. We must also be aware that if they can’t satisfy demand through increased units, we might see upside in IBM sales which is also good for ASML

• AI demand can falter along the way and orders might get pushed out or outright cancelled

Semicap investors should not forget about cyclicality. ASML is undoubtedly a cyclical business and I believe it will remain so over the long term. The only difference today is that we seem to be living in an unprecedented supply crunch across the industry, but historically these have ended in “unprecedented” supply gluts. I highly doubt we’ll live in an imbalance forever and the market seems to be increasingly skeptical about what demand might look like post-2028. We’ll see.

Anyways, I do believe that ASML seems to offer a pretty appealing return despite the stock having run up significantly. The facts have changed and the order books are starting to get filled, proving ASML with good visibility for the next 2 years. I remain more skeptical on the 2029 and 2030 numbers and they are more important than they look because those years will most likely inform the exit multiple.

I will continue to hold my position.

Have a great day,

Leandro

In this article I’ll provide some thoughts on Verisure, the “unknown compounder”.

Verisure is an interesting business because it doesn’t get the same attention as Rollins does despite arguably enjoying somewhat more interesting characteristics. This (not a lot of attention + attractive unit economics) tends to be an interesting set-up, especially since we can compare the current Verisure to what the company might look like in 5 years (i.e., we somewhat have an idea of what the market might be willing to pay for the company in 5 years).

While Rollins doesn’t tell you what it costs to “stand still,” Verisure does provide all the information necessary. In the remainder of this article I’ll go through Verisure’s unit economics, what each customer is really worth, the risks that come with its model, its valuation, and how it stacks up against Rollins head to head.

What Verisure does

Verisure is the European leader in monitored alarms. The company operates in 17 countries across Europe and Latin America, and ended Q2 2026 with 6.38 million subscribers (+9.4% YoY) paying an average of €48.2 per month (i.e., monthly ARPU). Revenue was €3.75 billion in 2025 (+9.9% YoY) with annualized recurring revenue now standing at €3.6 billion.

Let’s dig a bit deeper into what the company actually does. In a nutshell, the company sells professionally installed and monitored alarms to homes (82% of customers) and small businesses (18%), mostly in Southern Europe: Iberia, Italy and France alone are around 63% of revenue. Verisure typically receives two payments from its customers:

1. A one-time upfront installation fee to cover the hardware

2. A recurring monthly subscription that covers 24/7 monitoring from Verisure’s own monitoring centers, verification of each alarm (photo, video or audio), response (calling the police or sending a guard), maintenance and customer care.

Management summarizes what Verisure offers in four words: deter, detect, verify and intervene. The key point here is that, despite subsidizing hardware, Verisure is a service business, not a hardware business. This makes it more comparable to Rollins without being identical. The way I like to see this is that around 90% of revenue is recurring, with the hardware acting like the entry ticket to the subscription. The company is the undisputed leader in 13 of its 17 countries, with its portfolio being more than five times larger than the second player across its footprint. This seems somewhat relevant in an industry characterized by scale economies.

One thing that stands out from the outset is how management “discloses” the business. While Rollins bundles both activities (the acquisition machine + the recurring business), Verisure choose to split itself into the two halves of the “machine”:

• Portfolio Services (i.e., the existing customer base): ~87% of revenue, with a 73.7% adjusted EBITDA margin in 2025 (74.1% in Q2 2026). This is the company’s cash cow and the asset that likely drives most of the current valuation. Unlike Rollins’ relatively high churn, Verisure reports a churn rate that ranges between 6%-8%.

• Customer Acquisition (i.e., the acquisition machine): this segment is loss-making by design (−€723 million of adjusted EBITDA in 2025) because installation fees are kept low and most of the cost of winning a customer is paid upfront. This segment’s goal is to bring customers into “portfolio services” where the company can “milk” them for an average of 13-15 years.

• Adjacencies: smaller products sold outside the core alarm, ~3% of revenue and not really relevant for today’s discussion.

Management is ultimately telling investors how much the portfolio earns, how much it spends to grow it, and how much it spends just to replace the customers who leave. This means we can actually answer the “two questions that matter” which if you recall are...

1. How many customers leave each year?

2. How much does it cost to replace them?

CPA vs RMC, the two key variables

The entire Verisure business model comes down to two numbers, with the good news being that management is pretty conservative in how it calculates both:

• CPA (cost per acquisition): what Verisure spends, net of installation fees, to put one new alarm on a wall. It includes sales, marketing, the hardware and the installation. It’s basically an “all-in” metric that shows how much the company spends on acquiring a new customer

• RMC (recurring monthly cost): measures what it costs to serve that customer each month (monitoring, guards, maintenance, customer service). ARPU minus RMC results in EPC (Earnings Per Customer), the monthly adjusted EBITDA per customer.

RMC is not as conservative as it should probably be. RMC is an operating cost, so it doesn’t include maintenance Capex. Replacing equipment at existing customers (including the ongoing 2G/3G upgrades) is reported separately as Portfolio Services Capex, below EBITDA. This means the ~74% Portfolio Services margin overstates what the base really earns on a steady state. Deducting roughly €2.7 per customer per month of portfolio Capex (2025) from an EPC of €35.7 (Q2 2026) leaves an “all-in” portfolio margin of roughly 68% of ARPU. It’s also true that this number is somewhat depressed by the 2G/3G upgrades, which should fade after 2030, but the conclusion stands: Verisure owns a very profitable base business that continues to enjoy positive margin incrementals.

By dividing CPA by a year of EPC we get to what management calls the Acquisition Multiple, which is equivalent to the payback period in years. Acquisition multiples have historically ranged between 3.6x-3.8x meaning that Verisure takes around 4 years to pay back its CPA:

The nice thing about the business comes after those 3.8 years. Customers tend to stay on average 13-15 years with the company, so Verisure ultimately earns 9-11 years of EPC as “pure” profit. This customer longevity results in pretty appealing returns.

As you might have already imagined, the variable that makes everything “click” and that makes this business different to Rollins is churn. With 7.4% disclosed attrition, management estimates the average customer stays around 15 years. They claim this translates into a ~20% IRR on each new acquired customer, a claim that we can actually verify. Think of each new customer as a stream of cash pulled in two directions:

1. The cash per customer grows every year thanks to price increases and upselling (~2.5% ARPU growth over time)

2. On the other hand, the odds that the customer is still around shrink every year because of attrition (7.4%).

If we net both out, the cash a cohort of customers generates shrinks by roughly 5% per year. This means that a simple model that treats each customer as a slowly shrinking annuity should help us understand whether management’s returns claim makes sense. Using Q2 2026 EPC (€35.7) minus ~€2.7/month of portfolio capex (2025, including 2G/3G upgrades), each customer generates roughly €396 a year. Against a €1,644 CPA, 7.4% attrition and 2.5% ARPU growth, that’s a ~19% pre-tax IRR, right in line with management’s claim:

The problem is that EPC and CPA are moving variables, so we must stress-test the IRR to understand how sensitive it is to these variables:

The good news is that CPA would need to rise more than 60% (to ~€2,650) before a new customer stops clearing a 10% hurdle rate. This is ultimately the cushion that Verisure has at its disposal thanks to its customer longevity (15 years). The case for Rollins is markedly different. With its much higher churn, a similar rise in acquisition costs would hurt the unit economics a lot more because the pest control giant spends considerably more on replacing churn than Verisure does.

All this said, the CPA trend is worth watching as the Acquisition Multiple has drifted from 3.6x to 3.8x. It does seem like a slow squeeze on a very profitable investment and not a threat to the model per se, but CPA seems to be rising for everyone due to AI-driven search. Recall that, in part 1, I explained why AI-led search seems less of a risk to Verisure:

The interesting tug-of-war between CPA and EPC

What makes these two variables interesting is their interplay: CPA is growing faster than EPC. In Q2 2026 CPA rose 9.8% at constant currency (6.7% excluding the Spanish rebrand), while EPC rose ~4%. Management blamed media inflation and lower operating leverage on fixed sales costs (branches, sales salaries) but it should nevertheless sound familiar because it’s the same digital-lead inflation that hit Rollins. The key difference lies in what happens to Verisure’s base.

Verisure’s RMC has been remarkably stable over decades. It was roughly at the same level it is today in 2004 (wow) which means that, in inflation-adjusted terms, RMC has come down very significantly over time.

Management guides for roughly 1% annual cost-out over the long term through scale, its “Fit for Growth” program, and several AI initiatives. There was a small uptick in Q2 (+1.2%) but it needs context; it came from integrating ADT Mexico. Excluding Mexico, RMC fell YoY, again. So while acquiring customers is definitely getting more expensive for Verisure, serving its installed base is getting cheaper in real terms.

ARPU has kept rising 2-3% per year, which one could consider in line or slightly below inflation. For any other business this would be not great, but for Verisure that 2-3% can be considered a REAL price increase (due to RMC remaining flat) and therefore pretty profitable. What does this mean in layman terms? That Verisure’s base of customers is becoming more profitable over time even as the cost to acquire new customers rises. This is relevant because each new customer is acquired at the portfolio services margin, meaning that the company can somewhat protect itself from a rising CPA using its RMC leverage.

The capital intensity problem

One question should’ve crossed your mind by now...

If the unit economics are this good, why isn’t Verisure already a “cash machine”?

The answer lies in the capital intensity of the business model; Verisure needs to eat a lot of capital up front to grow organically. Group Capex was 26.2% of revenue in 2025 (29.8% in 2022), and that’s only the part that gets capitalized. If we add what Verisure expenses through the income statement, the company spent €1.69 billion gross (€1.32 billion net of installation fees) acquiring customers in 2025. That’s 35% of revenue, net, spent on the sales engine.

The beauty of Verisure’s disclosure (which is something one doesn’t get with Rollins) is that it splits the spend into two buckets:

1. Replacing customers who left (analogous to Maintenance Capex)

2. Growing the base (Growth Capex)

Here’s how cash flow should be interpreted:

A few things should stand out:

• €656 million went to attrition replacement. We can simply triangulate this figure by multiplying cancellations times CPA. That’s 34% of the cash the portfolio generated before customer acquisition, or 17.5% of revenue. This is Verisure’s true maintenance Capex, and it’s exactly the number Rollins doesn’t disclose outright, something we don’t really need to know that it’s proportionally higher.

• Another €665 million was pure growth investment. Verisure could, in theory, turn this off and milk its customer base. The interpretation is that, if the company had chosen not to grow organically, pre-growth cash generation over the last twelve months would’ve been ~€1.4 billion. In FY 2025, pre-growth operating cash generation was just under €1.3 billion.

• The portfolio is outgrowing its own appetite for growth (good news). The Portfolio Reinvestment Rate (acquisition spend divided by portfolio cash flow) went from 72% in 2019 to 61% in 2024, 59.6% in 2025 and 57.3% in Q2 2026. The bigger the base gets, the more cash is left over after funding growth even when facing CPA increases. What does this ultimately mean? That the CPA/RMC dynamics we discussed above seem favorable to Verisure.

Management has been quite vocal about the reinvestment rate as a cash flow lever. The CFO called the portfolio reinvestment rate “an important metric for us“ because...

He laid this out in a recent call (Q4 2025):

It’s one of the three “structural improvements“ (together with working capital and financing costs) behind the guidance to be free cash flow positive in 2026. In the Q2 2026 call, the CEO also commented that the improvements should also fund “additional shareholder returns beyond 2026“. The inflection is becoming increasingly apparent in the numbers:

Besides for the interest rate, we don’t really have much guidance on how these will move, but the trend seems pretty obvious. So much so that management has hinted at increasing capital returns to shareholders, both through dividends (first discussed in the Q2 2026 call) and maybe buybacks (still TBD):

This leads to what I believe should become the central “dilemma” for anyone looking at Verisure (it is for me at least): it is an incredibly cash generative business if it stops growing, but it’s very unlikely to stop growing anytime soon. If management keeps reinvesting ~€650-700 million a year to grow the base, the cash generation capacity of thus business never really shows up in shareholders’ pockets. If it stops reinvesting, the cash shows up but growth disappears.

So at first glance it feels like you can only have one of the two and you really don’t know what will make the market/investors happy. What a long-term investor should want is pretty obvious...If Verisure is able to continue earning a 19% pre-tax IRR on its customer acquisition activities, you’d want the management team to reinvest as much as possible. Why? Because it’s tax efficient and it’s a pretty decent return! Simultaneously, you’ll likely not want to enter the continuous loop of never seeing the company’s real cash generative capacity. Recall that a business that grows 10% per year, returns no cash to shareholders, and goes bust in year 10 is worth...zero. I don’t think the “dilemma” between growth and cash generation is unique to Verisure (every company has it), but it does seem a tad more pronounced due to its business model.

The good news for Verisure’s shareholders is that the company is gradually proving that you don’t need to choose one or the other, you can maybe have both! Free Cash Flow is inflecting as the portfolio scales, so Verisure is increasingly reinvesting aggressively while it generates incrementally more cash.

For now, investors have to trust the reinvestment. The value creation shows up in a bigger and more profitable portfolio rather than in reported free cash flow, and a sponsor-controlled company has every incentive to keep pushing for growth. The growth runway remains long, with penetration of monitored alarms in Europe remaining still at only around 5% of homes (evidently the TAM is not 100% of homes and I’ll discuss this in more detail later).

In my humble opinion, this interplay between reinvestment and cash flow is a feature rather than a bug, but only as long as the IRR on new customers stays well above the cost of capital. The day CPA inflation erodes that spread, the right answer flips from “grow” to “harvest”, and one would expect management to be disciplined enough to make that call.

A unique set of risks

I think that, as a recent IPO, dividing the company’s risks between those specific to the stock and those specific to the business makes sense. Let’s get the “stock risk” out of the way first.

The sponsor overhang

Verisure is still controlled by its pre-IPO owners. The company listed on 8 October 2025 at €13.25 per share, but only around 26% of the company was sold in the IPO. As of 31 December 2025, the largest shareholders were:

• Hellman & Friedman: 43.7%

• Eiffel Investment (linked to Singapore’s GIC): 16.0%

• Alba (Corporación Financiera Alba): 6.3%

• Securholds: 5.4%. This is the vehicle of Luis Gil, a long-time Verisure executive and current board member. It bought about €185 million of shares in the IPO at €13.25, and the lock-up specifically allowed it to pledge its shares as collateral for the debt used to finance that purchase. This is a stake worth watching with the stock around 40% below the IPO price: share-backed loans usually come with loan-to-value covenants, so a further drop could force a sale regardless of what the fundamentals say.

• Management and former employees: about 3% (the CEO alone holds about 1%)

The above simply shows that roughly 70% of the shares are held by investors who will, sooner or later, want to sell (or at least have the option to). At today’s share price, H&F’s stake alone is worth around €3.7 billion, close to half of Verisure’s market cap. The lock-up calendar looks like this:

• 180 days for the large shareholders (H&F, Eiffel, Alba, Securholds): expired in early April 2026, so they can already sell. It’s a “known issue” by now.

• 365 days for board members and senior management: expires right about now (October 2026), with other management shares released in phases over the following 3-12 months.

• Orderly marketing agreement until October 2027 (extendable in six-month increments up to October 2029): sales by existing investors need H&F’s consent and are typically coordinated through block trades.

The practical implication is that supply is likely to come in chunks through block trades, which are usually priced at a discount and might weigh on the share price for a while. Until H&F (and probably Eiffel) are meaningfully out, the stock is likely to trade with a cloud over it. This doesn’t (or shouldn’t) make it uninvestable (does anyone here remember what happened once TBBB’s lock-up expired?), and there are two mitigating factors. First, H&F chose not to sell in the IPO (most of the deal was a primary raise used to pay down debt). This means that, if they weren’t in a rush to sell at €13, they may not be in a rush to sell at €8. Second, every sale increases the free float, which over time improves liquidity and index eligibility. This is a risk to the timing and the share price, not to the business, and from a long-term investor’s POV it may well be the reason the stock is cheap in the first place.

Rollins has a version of the same problem. The Rollins family group still controls about 38% of the shares, and it has started to sell: LOR (the family vehicle) sold 20 million shares at $57.50 in November 2025 (Rollins bought back $200 million of them which in hindsight doesn’t seem ideal), after a first secondary in 2023. That sale came with a 365-day lock-up that expires in November 2026. The difference is one of scale: Rollins has a much larger free float to absorb sales, while at Verisure most of the company could still change hands.

The economics: front-loaded costs, back-loaded returns

The second risk is somewhat built into the company’s operating model. Verisure pays about €1,650 upfront for every new customer and needs almost four years of EBITDA to get its investment back. The real money is made through customer longevity, as these tend to stay 15 years on average. This means that the cost is front-loaded and the payback is back-loaded, which ultimately means that anything that shortens customer life hits value that has already been paid for.

That said, I’d make a couple of caveats:

• The model is fairly robust when looked at per customer. Because the payback (3.8 years) is so short compared with the customer life (15 years), the IRR doesn’t collapse if attrition rises. Attrition going from 7.4% to 9% only takes the IRR from 19% to about 18%. This makes CPA the much more relevant variable. The real damage comes from customers who leave before the four-year payback, which is why minimum contract periods (12-36 months in many markets) and the upfront fee matter.

• The real impact comes at the group level. Every extra point of attrition on a 6.4 million customer base means about 64,000 more cancellations, and replacing them at today’s CPA costs roughly €105 million more per year just to stand still. In the table below you can see how this impacts “no-growth owner earnings” (i.e., the cash Verisure would generate in a steady state):

In short, a jump in attrition wouldn’t destroy the unit economics, but combined with the company’s leverage it could wipe out a third of the cash flow available to shareholders, and that’s before any CPA inflation. In all fairness, attrition has been quite stable throughout Verisure’s history, but some things that could trigger it are a deep recession (management points to Spain’s resilience during the financial crisis, but the base is now much more international), cheaper DIY alternatives (cameras like Ring, or Verisure’s own Arlo), or service quality slipping as the company pushes for margin. The mitigant is that Verisure could cut its €665 million of growth investment quickly to protect cash, a lever Rollins doesn’t really have without shrinking.

TAM and what ADT tells us about mature state economics

Before I hinted at what Verisure should look like when it matures: the no-growth owner earnings. The interesting thing is that we have a pretty good example on the other side of the Atlantic to really understand whether our approximation is directionally correct. ADT is the largest monitored security company in the US, has almost exactly the same number of customers as Verisure (~6.1 million vs ~6.3 million), but their financial statements and KPIs look like night and day:

ADT’s revenue grew ~2% in the last year and its monitoring revenue (the equivalent of Verisure’s portfolio) barely grew 1.4% in 2025. The subscriber base is actually shrinking (6.1 million vs 6.4 million a year ago, partly due to a divestiture). Four potential reasons explain this growth trajectory:

• Churn eats almost all the growth. ADT loses 13.1% of its revenue base every year, almost double that of Verisure. It added ~190,000 customers in Q2 2026 (~760,000 annualized), while 13% attrition on 6.1 million customers means roughly 800,000 cancellations every year. In layman terms, ADT is running on the same treadmill as Rollins: most of what it spends on sales just replaces customers who left. Relocations (Americans move much more than Europeans), non-payment and price-driven cancellations are likely the main culprits.

• A mature and crowded market. Monitored alarm penetration in the US is around 23% according to Verisure’s own estimates (vs ~5% of homes in its European footprint), and ADT competes with pro-installed alternatives (SimpliSafe, Xfinity, Vivint, Brinks) as well as DIY players like Ring (Amazon), SimpliSafe, Arlo or Wyze.

You might wonder why the penetration gap is so large between the US and Europe. There are a couple of reasons I can think of, but interestingly crime isn’t one of them. According to Verisure, many of its European markets actually have higher burglary rates than the US.

The gap comes down mostly to structural reasons. First, housing type. Most Americans live in detached houses with many entry points, while around half of Europeans live in flats (even more in Spain or Italy), which feel safer and have historically been harder to address. Second, insurance: US insurers commonly offer premium discounts for monitored alarms. Third, price and history: the US market developed decades earlier, with a huge dealer network and cheaper monitoring plans, which turned alarms into a mass-market product. Lastly, police rules: several European countries require alarms to be verified before police are dispatched, which raises the cost of the service and limits low-cost offers.

• Expensive growth channels. A good chunk of ADT’s additions came from dealers, affiliates and bulk purchases of customer accounts, which are more expensive and less sticky. Management is now actively pulling back from these channels.

• A cash cow mindset. ADT returned $684 million to shareholders in the first half of 2026 alone (buybacks and dividends, including buying part of Apollo’s stake as it exited). The business is run for cash rather than for growth, which is rational when the returns on growth are lower.

The most telling data point is that ADT has launched “ADT Blue”, a cheaper DIY product line sold on Amazon, to compete for “value-conscious” customers it had never targeted before. Even the US leader has had to go down-market to defend itself against Ring! Its medium-term targets (1 million more subscribers by 2030, attrition down to 11%) show that the strategy is now to stop the bleeding rather than to grow aggressively. Everything that I’ve discussed here about ADT is important to keep in mind for later when we take a look at the valuation.

Other competitors

Beyond DIY (Ring), Verisure competes with two main groups. The first is other professionally-monitored players, a fragmented category with more than 70 providers across its footprint (Sector Alarm, Movistar Prosegur Alarmas, EPS Homiris, Nexecur, IMA Protect and Orange Maison Protégée in Europe, ADT and Prosegur in Latin America), many of them affiliated with banks or telcos. The numbers seem to clearly favor Verisure against these players: management estimates it held around 37% of the installed base of professionally-monitored systems across its footprint at the end of 2024 (up from 28% in 2019), and across its top 10 countries it’s on average 2.7x larger than its nearest competitor. Despite these being company estimates, the direction seems obvious and scale seems to be reinforcing: the leader can outspend everyone else on marketing and innovation while spreading monitoring centers, guards and IT over a much larger base (hence the falling RMC discussed earlier).

The second is telcos, banks and insurers, which already have a relationship with the customer and could bundle security into it at a lower monthly price (Telefónica and Orange have re-entered the category in recent years in Spain and France). Verisure has turned part of this threat into a new sales channel: its partnership with BPCE (one of the largest French banking groups) has scaled nationally, and in April 2026 it launched a partnership with MasOrange in Spain. Management says customers coming through these alliances come in at Verisure pricing, with attrition in line with or better than the group’s (they are pre-qualified and often bundled with a mortgage or an insurance policy).

So Verisure doesn’t face a “long tail” of local competitors the way Rollins does, and its main competitors in each market are few and smaller. One would think that entry barriers to alarms is higher than it is to pest control, and while this is true, it also comes with a greater deal of technological disruption risk.

The DIY risk

The most relevant thing that comes with the ADT analogue is that it shows what happens when a market matures and cheap DIY alternatives become good enough: the low end of the market moves to DIY, the cost to win a professionally monitored customer goes up and growth stalls. The 7.4% vs 13.1% churn gap suggests that Europe is structurally different, but the question is whether the same could happen to Verisure as its markets mature.

In Europe, DIY is, imho, the most interesting competitive threat Verisure faces. Cheap connected cameras, video doorbells and self-installed kits (Ring, Google Nest and a growing list of Chinese manufacturers) let people monitor their home from their phone for a fraction of the price and without a contract. Verisure itself flags in its prospectus that a shift in consumer preferences towards self-monitoring could lead to pricing pressure, higher attrition and a higher CPA, and that big tech (Amazon, Google, Apple) could decide to push further into the category.

There are a few reasons why I believe this threat is real but manageable:

• It’s a different product. A camera tells you something happened; Verisure’s offering is built to deter, verify and send someone (a guard or the police). The customer who wants the latter is unlikely to settle for a notification on his phone at 3 a.m. although I must say that ADT proves that DIY disruption should 100% be taken seriously.

• Regulation works in Verisure’s favor. In many of its markets, alarm hardware must be certified, installation must be done by a certified professional and maintenance is mandatory. On top of that, in France the police only respond to alarms that have been professionally verified, and in Spain they require video, audio or personal verification (otherwise they only respond after three alarms in 30 minutes). A self-monitored kit simply doesn’t get you the response that you are looking for.

• Verisure owns a DIY brand itself. if part of the market moves to self-monitoring, Verisure participates through Arlo, even though it would most likely be at worse economics. The fact that management acquired Arlo is likely a confirmation that the DIY risk is real.

Where I think DIY should be a real concern is at the low end of the market. The price-sensitive customer who might have bought a cheap monitored alarm may now just buy a couple of cameras and, while this is unlikely to show up in the existing base; it would show up as a slowly rising CPA (more effort to convince each customer) and fewer gross additions in Verisure’s more mature markets.

Is penetration perennially low?

Verisure’s bull case relies on the fact that only ~5% of European homes have a monitored alarm, vs ~23% in the US. But a low penetration rate is not the same thing as a large addressable market even if management makes it seem this way. There are a few things that make me somewhat skeptical of the “untapped market” narrative:

• The category itself isn’t growing that fast. According to Verisure’s own estimates, the installed base of monitored alarms in Europe grew at around 4-6% per year between 2019 and 2024. Verisure has grown faster mainly because it captured around two-thirds of the category’s growth, i.e. by taking share. In Spain, its most mature market, it already holds 63% of the installed base, so share gains can’t continue forever and the underlying industry is not in hyper-growth mode.

• There may be cultural and economic reasons behind the low penetration. A ~€48/month subscription is a significant expense for many European households, homeownership patterns differ (more apartments, which have historically been harder to address) and the perceived need for security varies a lot by country. Penetration has indeed been rising for decades, but it has done so slowly.

• Is the company’s expansion outside of the core a tell? If the opportunity in core markets were truly unlimited, why buy ADT Mexico (from Johnson Controls, not from ADT Inc.) (~125,000 customers) and expand in Latin America? Even though this is a fair concern, Latin America does make sense in its own merits. Penetration is even lower than in Europe and safety concerns are higher, which led to category growth of ~11% per year between 2019 and 2024. With ADT Mexico, Verisure bought an immediate #1 position for €207.5 million. This gave management 125,000 customers and resulted in a CPA of €1,660 which is right in line with what they spend organically. This is definitely not the case for Rollins, which tends to pay considerably more for customers acquired inorganically.

The realistic scenario probably lands somewhere in between a mature state “local” business and a fast-growing international business. Verisure’s European business probably converges slowly towards a more mature profile (lower growth, more cash returned), while Latin America, new product categories (LockGuard, Guardian, Seniors) and pricing keep total growth going for a while. The good news is that even the “mature” end-state, as ADT shows, is a ~50% EBITDA margin business that returns a lot of cash. The bad news is that the market won’t pay a compounder multiple for it, so the key question for the investment case is how long Verisure can keep growing at ~10% before it starts to look like ADT. Let’s address this.

Is Verisure a buy after its 40% drop?

Verisure listed at €13.25 per share in October 2025 and has had a rough first year as a publicly traded company. The stock trades at around €8.20, near its all-time low, which puts the market cap at ~€8.4 billion and the enterprise value at ~€13.3 billion (with €4.9 billion of net debt).

Read more

You can read this article (almost) entirely for free. If you like what you read, consider becoming a paid member to get access to…

• All in-depth reports (16 companies profiled thus far, and growing)

• Earnings follow-ups

• Other investment related content

• A community of like-minded investors

• Complete access to my portfolio and transactions

• Occasional webinars

You’ll also get access to the 30+ page detailed report on the robotics industry.

The in-depth reports of Stevanato and Deere are also free to read to gauge the quality of the research. Other in-depth reports have been written about…

• Deere (DE)

• FRP Holdings (FRPH)

• Five Below (FIVE)

• Diageo (DEO)

• Hermès (HESAY)

• Atlas Copco (ATCO.B)

• Stevanato (STVN)

• Keysight (KEYS)

• Zoetis (ZTS)

• Judges Scientific (JDG.L)

• Medpace (MEDP)

• Trupanion (TRUP)

• AAON (AAON)

• BBB Foods (TBBB)

• Eurofins (EPA.ERF)

• Sabre (SABR)

• Rosebank Industries (ROSE.L)

• Undisclosed position

• Undisclosed position

Join today:

Subscribe now

It was yet another interesting week in financial markets. The indices did not move in the same direction and the “AI trade” started to show some “cracks.”

Without further ado, let’s get on with it.

Articles of the week

I published one article this week: Rollins (ROL), Verisure (VSURE), and the cost of standing still (Part 1).

I began looking at Rollins (ROL) after its significant price drop…

…and I must say that I was not very much impressed with the stock even after the drop. I explain why in the article, which is part 1 of a mini series in which I’ll also go over Verisure. Verisure is a very different and similar company at the same time, and I’ll most likely publish part 2 this upcoming week.

Without further ado, let’s see what the markets did this week.

Market Overview

Markets were mixed this week. The S&P 500 rose whereas the Nasdaq dropped:

It was yet another interesting week in financial markets (it always is). The “AI trade” showed some “cracks” when OpenAI disclosed an annualized revenue figure of $50 billion. This was considerably below what the market was anchored to ($70 billion), but OpenAI later made it seem as if it were a confusion between gross vs net revenue. Regardless of whether you want to believe that the labs are decelerating or not, the reality is that the “AI trade” seems to be taking a breather. This is one example of many but after an incredible run, the KOSPI is down 30% from the highs it marked in June and has gone nowhere since:

It’s definitely true that the KOSPI is exposed to memory and that the AI trade is much more than memory, but it’s nevertheless interesting to see that some AI-related stocks are not moving much despite posting spectacular numbers. The market is a bit (being kind here) bipolar so we might as well see the narrative shift soon (or not, who knows).

We are currently living in a very interesting market environment. Rates are rising and remain pretty high compared to recent history, but this hasn’t stopped stock indices from being very close to ATHs. Warren Buffett used to say that interest rates are “like gravity to asset prices,” and while this is true, there is one exception. Rates and stocks can remain high simultaneously if the market anticipates strong earnings growth. This may well be the case here if one considers that hyperscalers and AI capex beneficiaries make up a good portion of the index. Hyperscalers are investing mouthwatering amounts into Capex which (if one believes they’ll earn somewhat decent returns) should result in a considerable earnings increase. At the same time, hyperscaler Capex is being spent on Capex beneficiaries which also make up a good portion of the indices, meaning that the earnings of those companies should rise too!

One thing that does “worry” me is that much of this AI buildout is going to be financed with debt, so seeing a higher cost of debt is maybe not ideal. Truth is though, if that returns are going to be 15-20% minimum then it makes sense to continue issuing debt to invest in AI at current rates.

I also worry that indices might have become a bit “fragile” in the sense that they’ve become pretty concentrated. If the AI trade plays out well, indices will most likely do fine, but if it doesn’t play out well…then passive investors might be in for a bit of trouble. It’s no secret that indices have become very concentrated around a given number of names, but it’s also true that the proportion of earnings that one can attribute to these businesses is likely similar to the weight they have in the index. What does this mean? That the indices might be “acting” rationally by overweighting these companies. One thing that seems true, though, is that a lot of passive investors probably don’t realize what they own.

I believe all of this situation also exposes what’s great about indices. When stocks do well in an index they typically rise to the top, meaning that the index eventually overweight the winners. As the winners typically keep winning and momentum has been a relevant factor in markets over the past years…well, indices have done pretty well. People who claim that an equal-weight index is better or safer are imho doing exactly what Peter Lynch claimed that investors should never do…”cutting the flowers and watering the weeds.” This doesn’t mean that the equal-weight index can’t outperform the market-cap weighted index; it definitely will if AI does not do well. This would be the ideal environment for active managers not overly exposed to AI.

The industry map was mixed this week, although it was pretty much green outside of semis. Agricultural equipment did not do great either, and I’ll discuss why in the company-specific news section:

The fear and greed index improved and entered neutral territory:

Company-specific news

Read more

It’s highly likely that you have come across the kinds of products Rollins and Verisure sell. The similarities between both businesses begin in the nature of the product/service they provide: pest control and alarms are considered critical expenses for homeowners because they make up a low portion of a total household’s budget but help protect a valuable asset. These characteristics translate into somewhat steady demand and relatively low churn (”relatively” is doing the heavy lifting here). Even though these businesses are unlikely to meet many investors’ description of a subscription-like business, I must say they look a lot like one.

Rollins and Verisure are characterized by a similar compounding machine/business model that can be subdivided into two “engines”:

1. A base of customers who pay regularly

2. A sales engine that has to keep refilling that base as customers leave

The core concept is identical across both businesses but the interplay between the variables that matter varies markedly and dramatically changes the unit economics. This is why I believe that putting both side by side is a very interesting exercise.

Yet another thing that’s markedly different is the valuation of both businesses. Comparing earnings-based valuation multiples makes little sense here because both companies expense things differently, so I’ll have to graphically plot their EV/sales multiples. Even after its recent drop, Rollins trades at a 15-20% premium to Verisure both on an EV/sales and EV/(EBITDA-Capex) multiple basis. The premium was as large as 50%+ earlier during the year, but Rollins has derated to a greater extent than Verisure. This multiple compression has been likely driven by a similar concern which I’ll discuss in more depth later:

The interesting question here is whether Rollins’ valuation premium justified. Investors must try to answer two questions before aiming to answer this:

1. How much does it cost to win a customer?

2. How long does that customer stay?

Arriving at the answer to this question is not as straightforward as it seems. Both Rollins and Verisure have a “good” and a “bad” side just like all businesses do, and it should be up to investors to judge whether their “good” and “bad” justify the valuation premium/discount. The first thing that stands out is that one of these companies is willing to help you answer the questions above, whereas the other does not.

Rollins has undoubtedly been one of the great compounders of the last few decades, but the stock is down almost 50% over the last twelve months. The reason (imho) lies in a cost that most investors never look at: what does it cost Rollins to stand still? Verisure, which listed in Stockholm in October 2025, is the opposite case. It looks capital-intensive and leveraged at first glance, but it discloses its customer economics in so much detail that you can actually do the math and understand what you are paying for.

Today’s article focuses on Rollins, the known compounder. In part 2 of this mini-series I’ll go over Verisure, the unknown and new (to financial markets that is) compounder.

Rollins (ROL): outlining the unquestionable

Rollins is the largest pest control company in the US. Orkin is its flagship brand, but the company can really be considered a house/portfolio of brands (HomeTeam, Fox, Saela, Northwest and dozens of local names) serving more than two million residential and commercial customers. With 75% of revenue being “recurring” services spread across millions of customers, I believe one is not making a mistake by treating Rollins as a subscription-like business. This in and of itself doesn’t make a business great, though.

The company’s track record speaks for itself. Rollins grew its revenue from $1.57 billion in 2016 to $3.76 billion in 2025, a ~10% total revenue growth CAGR that was tilted to the organic side (organic growth CAGR was above 7% between 2022 and 2025). Acquisitions filled the delta and have consistently added around 2-3% of growth per year (has been higher lately). Rollins takes advantage of the fragmentation and local dynamics of the pest control industry when it comes to M&A, something that’s not as large as a growth lever for Verisure.

No matter how skeptical one wants to be, there’s plenty to like here. I believe that the “good” of this business has been widely discussed by many, so I’ll make a quick summary of what I believe to be the “good” and will jump right into what I believe many are missing.

Let’s start with the good:

• Good growth and resilient demand. The company has managed to grow at a double digit clip for many years while enjoying HSD organic growth. The demand for its services also seems very resilient (both in the residential and commercial side) through “tough” economic periods, so the company has not only been a good grower, but also a consistent one.

• The Rollins family is still involved with the business, although management is currently “professionalized” and the family has began periodically selling shares. These sales are not bad per se, but they can potentially create a supply overhang on the stock (this is a much more relevant topic for Verisure).

• Route density as the moat. Pest control is a local business with every extra customer on an existing route being almost pure margin. The rationale is that the truck, the technician and the branch are fixed costs so, in theory, the bigger you are in a given zip code, the harder you are to beat on cost. While this is true, I believe people have mixed up these dynamics with operating leverage. I believe Rollins has less operating leverage than many appreciate, and I’ll explain why later.

• A fragmented industry. The fact that the industry is made up of thousands of mom-and-pops (more than 30,000 according to Rollins’ CEO) means there’s a long M&A runway for a consolidator like Rollins. In 2025 alone Rollins closed Saela plus 26 other deals.

• It looks very capital-light (”looks” is doing the heavy lifting here). Rollins generated $650 million of free cash flow in 2025 on $855 million of adjusted EBITDA (22.7% margin). Reported Capex was roughly $28 million, less than 1% of revenue. At first glance it looks like this business doesn’t need capital to grow, but one would be very wrong in assuming this is true.

• A conservative balance sheet. Leverage stands at ~1x EBITDA, which seems pretty low for such a recurring business. While I don’t expect Rollins to “lever up” significantly, I believe the low leverage does increase the runway to return capital to shareholders.

This is exactly the type of business I like at first glance, but the problem is that I don’t necessarily agree with several of the above, at least not to the extent many people do. This (imho) makes Rollins a worse business than many believe it is.

What’s not to like?... A couple of things

Even though one is constantly bombarded on social media about how cheap Rollins is after the recent fall, the reality is that it might well still be expensive. The halo effect dominates financial markets: when a stock is doing well, everything from the management team to the moat seems perfect. On the flipside, when the stock is doing poorly, everything seems to be going terrible. Aristotle used to say that “virtue lies in the middle” and I believe this is largely applicable to investing. The only caveat is that “stock price bias” is real and therefore we might be more inclined to think negatively about a stock when price is heading south. With the goal of being transparent, I will tell you that this might well be the case here! Still, I do think the business model is more fragile than many people think it is.

Let’s start with the most obvious concern, albeit not the most worrying.

Capital-light… or simply expensed?

Rollins barely capitalizes the cost of winning a customer. Apart from a small amount of sales commissions ($34 million amortized in 2025), digital leads, advertising, sales salaries and door-to-door reps all run directly through the income statement. Selling and marketing alone was $485 million in 2025, 12.9% of revenue. Even though expending everything directly through the income statement is conservative accounting (i.e., good), it does imply that the sub-1% Capex figure hides the real reinvestment needs of the business. The most relevant implication, though, is that a big chunk of what’s spent/reinvested isn’t spent on growth but rather on replacing existing customers.

The way to think about this is that, if Rollins stopped its sales and marketing activities tomorrow, the customer base would start shrinking immediately at a pretty concerning clip. The money spent to keep the business flat should economically be considered maintenance capex, but for Rollins it just happens to sit in Opex. This said, the real “worrying” part lies in churn and the cost to replace the existing base.

Churn and the rising cost to replace customers

Rollins doesn’t report a retention rate in its filings. Management repeats every quarter that retention is “strong” and “stable” but they indirectly provide a number to triangulate churn. At the May 2026 Investor Day, the CFO said the following:

A four-to-five-year customer life implies residential churn of roughly 20-25% per year. Combining both residential and commercial results in a blended churn rate of around 15%-20%. This churn rate stands out (as high, that is) and management acknowledges the concern:

Residential pest control is a business where customers move, switch to DIY or simply cancel after a quiet season, so churn is structurally much higher than Rollins’ historical track record would suggest. I must say that it caught me by surprise when looking at the business. The key question is not whether churn is high, though (we know it is), but the key question is rather...

Because the business model is quite sensitive/exposed to it. The short summary is that, the higher the churn, the more of your sales effort goes into running to stand still. One could consider these investments “unproductive” in the sense that they are not made to grow and I’d say few shareholders would want to see a good chunk of money being reinvested into maintenance. Let’s do some simple math. Say Rollins wants to grow its business by 5% per year and currently experiences 15% churn. The company would have to grow 20%, with 75% of this growth being just “maintenance” growth:

The simple way to interpret this is that, with a residential churn of 20-25%, four out of every five residential customers Rollins acquires simply refill the base. This means that growth is very dependent on the acquisition engine running at full speed all the time, and the cost of that engine becomes the key variable of the whole investment case.

Rollins doesn’t disclose a CAC (customer acquisition cost), but we can more or less approximate it with the information we are given. The company spent around $2.07 per $1 dollar of organic growth in FY 2025. This cost has been steadily rising and marked a new high in H1 2026: $2.37:

The bottom line here is that Rollins is spending more and more to grow less and less. Selling and marketing grew 13-14% per year in 2024 and 2025 while organic growth decelerated, and the result is that the cost per dollar of organic revenue is up about 20% in two years. This dynamic is especially dangerous when one considers that a good chunk of said investments are “unproductive” in the sense that they simply refill the base. So, not only does Rollins need to spend a good chunk of S&M in standing still, but standing still is becoming more expensive! This is not the best news when thinking about incremental returns.

Management itself said at the Investor Day that it has increased selling and marketing by 200 basis points of revenue over the last few years, paid for by cutting G&A. G&A leverage is a valid source of funds...but it eventually faces a limit. Rollins is not the only company disclosing higher media costs by the way, but this trend matters deeply for Rollins due to its relatively high churn.

All of the above has become evident in Rollins’ EBITA incrementals, which have disappointed even the pessimistic...

...but the higher CAC is not the only reason behind this trend. I also believe that Rollins’ model is less scalable than what many people believe it is.

There’s operating leverage in route density, but it’s not unlimited

The route density advantage is real and I’m not going to be the one to negate it, but it doesn’t mean that the business model is very scalable. The rationale is that a technician can only do so many stops in a day and a truck only carries one technician. This means that, once a route is full, the next customer doesn’t come for free: it requires another technician, another truck and, eventually, another branch. So, ultimately, the cost base looks fixed while you fill a route, but over any meaningful period it becomes a step-variable, growing more or less in line with the number of customers you serve.

The numbers seem to back this up. Between 2023 and 2025 Rollins’ revenue grew 22%, yet gross margin only went from 52.2% to 52.8%, and employee costs in cost of services stayed flat at 31.0% of revenue in both 2024 and 2025. Operating margin barely moved (19.0% in 2023, 19.3% in 2025). The problem is that the reverse happens when volumes disappoint; the business suffers operating deleverage. In Q2 2026 “service salary de-leverage” was one of the main reasons gross margin fell 100 basis points. This doesn’t seem like the ideal asymmetry.

If every new route needs a new technician, keeping technicians becomes just as important as keeping customers (not really, but you get it, it’s important). Employee expenses in cost of services were $1.17 billion in 2025 (31% of revenue), and Rollins has around 22,000 employees. The problem is that it loses a significant chunk of its employees early on (retention in year 1 is estimated at 50%). In management’s own words...

The good news is that those who make it past year one tend to stay 10, 15 or 20+ years. At the 2026 Investor Day, the company put a “price” on these dynamics...

• Rollins spends roughly $15,000 to onboard each new teammate, money that is lost when they leave within the year.

• Retaining just 1,000 more teammates would save almost $15 million, and fixing technician turnover is worth $15-20 million a year, or 30-40 basis points of margin.

• First-year retention improved 8% in 2025 and nearly 18% since 2023, so progress is being made, but from a low base.

So labor is not only step-variable, it’s also leaky: to add one technician to a route, Rollins often has to hire more than one. In the pest business, employee churn feeds customer churn because it’s a relationship business, but the opposite could be said as well: employee retention feeds customer retention. Rollins itself links the two: talent retention “enables a better customer experience and improved customer retention”. But hold on, there’s news also on this front.

The FTC (Federal Trade Commission) added yet another “headwind” in 2026. In May 2026 Rollins agreed to an FTC consent order after the agency alleged that it required nearly all of its roughly 18,000 US workers (including technicians, sales staff and customer service reps) to sign non-competes barring them from working in pest control within a 75-mile radius for two years. Under the order, Rollins can only use non-competes for senior executives with equity, must rescind existing ones for everyone else, and its customer non-solicitation clauses have been narrowed, with compliance reporting for 10 years.

Management calls the FTC concern “totally overblown” and says the order won’t have a material impact. The CFO argued that first-year leavers quit because “this is a really tough job”, but that they typically don’t do so to start a competitor. That’s probably true for the first-year problem, but there’s a flipside: a tenured technician who knows the customers on his route can now more easily move to a competitor (or start his own business, which is how much of this fragmented industry was born) and take some of those relationships with them. With less legal protection, Rollins has to retain both technicians and customers through pay and culture, which means another cost line that is unlikely to go down. Yikes.

The next article will show why Verisure is very different in this regard. Once an alarm is installed, serving one more customer mostly means a bit more capacity at a monitoring center that is already running 24/7, plus guard response and maintenance only when something happens. This is the reason why Verisure’s recurring monthly cost per customer (RMC) is going down over time (€12.5 in 2024, €12.2 in 2025) while its Portfolio Services margin expands (72.7% in 2024, 73.7% in 2025, 74.1% in Q2 2026). Adding a customer to Verisure’s base adds revenue at around 74% margin, whereas adding one to a Rollins route adds revenue at roughly the same margin the company already earns. The heavy cost at Verisure is upfront (the CPA), but serving the customer once they’re in is a very scalable business. Even though this last part seems unimportant, it’s actually pretty relevant and likely the key difference between both businesses.

The model shows cracks in FY 2026

All of what I’ve discussed above is exactly where/how things went wrong this year for Rollins (so yes, I’m looking at it in hindsight obviously). Rollins grew organically 5.7% in Q2 2026, with residential organic growth shy of 4%. Management cut full-year organic growth guidance from 7-8% to “at least 6%” and incremental margins to “at least 10%”, a far cry from the 30%+ medium-term target presented at the May Investor Day. Adjusted EBITDA grew just 2.2% and the margin dropped from 23.1% to 21.9%. No bueno.

More than the numbers per se, the explanation was what was most telling. Retention and pricing were fine, but what worsened was the acquisition funnel. Orkin, which relies on search, digital media and inbound calls, saw leads dry up. And when asked whether spending more would have helped, CEO Jerry Gahlhoff answered:

In other words, the marginal cost of acquiring a customer through digital channels went up considerably, so much so that it wasn’t worth paying for! Management had already flagged in February that they are “constantly fighting increases” in the cost of generating digital leads and, in fairness, they are not the only company to flag this. AI-driven search might drive costs up, and I believe this is more of a problem for Rollins than for Verisure. Let me explain why.

When someone asks a search engine or a chatbot “who should I call about ants in my kitchen?“, the answer is increasingly a short list of names with an explanation rather than a list of blue links with Orkin as the top result (this was the “Google Search” era). For Rollins, that list will often include local operators with good reviews competing on equal footing with Orkin. Pest control is a local, fairly standardized service, and Rollins competes against a huge tail of small companies. An AI answer that compares options, prices and reviews in one place flattens the brand advantage Orkin has historically bought through advertising. Management is aware of this and they discussed it at the recent investor day, claiming that Orkin ranks first in AI visibility among pest control brands, and admitted AI “may put pressure on total lead quantity”. The encouraging news is that while leads are drying up, the quality of the leads seems to be improving (still TBD).

I believe (even though I don’t have proof) that Verisure is in a significantly different position. A monitored alarm with guard response isn’t something a local handyman can offer because it requires 24/7 monitoring centers, licensing (in Spain and France, for example, a connection to the police), a field network and scale. In most of Verisure’s markets there are only a handful of credible alternatives, and Verisure is usually the leader. It’s also a considered, high-trust purchase where brand matters more and where Verisure relies heavily on its own sales force and telesales rather than on winning a single search query. The bottom line is that an AI assistant asked for “the best home alarm in Spain” is far more likely to name Verisure than an assistant asked for “pest control near me” is to name Orkin. This hasn’t stopped Verisure from derating, though!

Verisure isn’t immune to Rollins’ concerns (it has also flagged digital media inflation), but the key difference lies in churn: with ~7% churn instead of 20-25%, Verisure simply needs to win far fewer customers each year to stand still. This means that any increase in the cost of a lead hurts it much less especially as its base becomes more profitable over time.

Let me be clear...it’s not the end of the world for Rollins

Rollins’ answer and counter-argument is to lean on the brands that are not dependent on digital leads. Fox (acquired 2023) and Saela (acquired in 2025) are growing double digits organically thanks to their door-to-door sales forces, and HomeTeam sells prevention through home builders. Management called these customers “nice sticky customers” that build efficient routes and their argument is that they help diversify the customer acquisition routes.

Door to door comes with several advantages. For starters, management claims that retention data on door-to-door is 1-2% better than customers from its other channels. This doesn’t mean that door-to-door is free. Rollins paid a somewhat hefty multiple for both Fox and Saela (Saela alone cost $207 million), and the model relies on large seasonal sales forces. So another relevant question comes to the mix...

Rollins doesn’t disclose this, but the rising selling and marketing cost per dollar of organic revenue we saw earlier might be a hint.

The margin “problem”

Because Rollins’ acquisition costs are fully expensed and it spends a considerable chunk on onboarding new customers, any further increase of its CAC hits margins one-for-one and pretty quickly. Here’s a sensitivity on the 2025 adjusted EBITDA margin of 22.7%, depending on how much of selling and marketing is really about winning new customers:

A 30% increase in CAC would wipe out several years of the margin expansion that management is promising (incremental margins of 30%+ over time). And the worst part is that you wouldn’t see it coming in the reported numbers until it was already there. While I am definitely not denying that Rollins is a good business (I actually believe it is), I believe that its business model is more fragile than many people think it is. It’s a subscription-like business, yes, but the company’s churn makes the customer acquisition engine pretty relevant for the thesis.

At what price is Rollins a buy?

Something that’s pretty evident is that the market has already punished the stock (shoot first, ask questions later). Rollins trades at around $30 per share, down roughly 48% over the last twelve months and more than 50% from highs:

This is a big drawdown for a business that most investors considered untouchable, but nothing unheard of in today’s “compounder” land where a lot of real (and perceived) high quality companies have suffered considerable drawdowns. While many of these companies were undeniably trading at pretty hefty valuations, it’s fair to think that (for some) the market might have overshot to the downside. Is this the case for Rollins?

Read more

I unveiled my portfolio management tool (aka: PMT) through a Webinar and in NOTW#93 (published in May this year). The PMT is a comprehensive spreadsheet (in HTML form) that contains features such as…

• My portfolio with weights and returns

• A transaction log

• Portfolio industry and factor exposure

• A watchlist

• …

By far the most relevant feature within the PMT is the risk/IRR matrix. This matrix plots my positions across two axes (one is risk and the other one is expected IRR) which eventually result in four different quadrants. The basic idea of this matrix is to help me maximize the risk-adjusted return of the portfolio by understanding that I need to allocate a greater amount to the bottom-right quadrant while “avoiding” the top-left quadrant. Here’s how it looks like after all the portfolio changes:

Worth noting that nothing in investing should be black or white, so I don’t think that simply allocating 100% of the portfolio in the bottom-right quadrant while completely ignoring the top-left also makes sense. The objective is to find a balance that maximizes the risk-reward return while not putting all eggs in one basket. I shared this not long ago, but a good chunk of my YTD winners have come from quadrants that I wouldn’t have expected. This is not unusual because valuation is a poor leading indicator of 12-24 month returns. Valuation always matters, we just don’t know when it’ll start to matter.

So, what I’ve done this week is reposition the portfolio to maximize the matrix you see above. You shouldn’t expect these kinds of portfolio adjustments to happen recurrently but rather very very rarely.

So without further ado, let’s jump right into the changes.

You can read this article (almost) entirely for free. If you like what you read, consider becoming a paid member to get access to…

• All in-depth reports (16 companies profiled thus far, and growing)

• Earnings follow-ups

• Other investment related content

• A community of like-minded investors

• Complete access to my portfolio and transactions

• Occasional webinars

You’ll also get access to the 30+ page detailed report on the robotics industry.

The in-depth reports of Stevanato and Deere are also free to read to gauge the quality of the research. Other in-depth reports have been written about…

• Deere (DE)

• FRP Holdings (FRPH)

• Five Below (FIVE)

• Diageo (DEO)

• Hermès (HESAY)

• Atlas Copco (ATCO.B)

• Stevanato (STVN)

• Keysight (KEYS)

• Zoetis (ZTS)

• Judges Scientific (JDG.L)

• Medpace (MEDP)

• Trupanion (TRUP)

• AAON (AAON)

• BBB Foods (TBBB)

• Eurofins (EPA.ERF)

• Sabre (SABR)

• Rosebank Industries (ROSE.L)

• Undisclosed position

• Undisclosed position

Join today with a 20% discount:

Claim the 20% discount

Read more

It was a matter of math eventually mathing, so this is exactly how it played out. Judges’ H1 was pretty abysmal, but some green shoots have started to emerge in H2. If we think about Judges holistically, we can come up with two main growth levers:

1. Organic growth: has averaged around 7% annually over time

2. Inorganic growth: Judges can be considered a serial acquirer

Both have suffered significant “headwinds” lately, which has translated into an abysmal H1 performance:

Judges has suffered many headwinds on the organic growth front. For example, the company is exposed to University funding in the US, which has virtually frozen during Trump’s mandate and his Government’s proposed cuts. Even though these cuts did not get implemented (at least not to the extent that the Government wanted), university Capex remained “frozen.” The Government is back at it for 2027, so the future doesn’t look rosier either.

Secondly, Judges is also suffering some headwinds in China. Orders had recovered strongly after a couple of weak periods, but order conversion to revenue is taking longer than anticipated. The reason is that China is finalising a new process for R&D tax exemptions (worth around 13%) on government-funded purchases of scientific equipment. This ultimately has translated to a Capex freeze, as many customers are waiting for this process to be effective to ultimately act on their orders. This also impacts order intake which has been delayed until this is resolved:

Thirdly, Judges has experienced something that could be considered organic/inorganic at the same time. After having acquired Geotek in 2022 with the expectation of enjoying one Coring contract per year or every one and a half years…the reality has been markedly different. Judges enjoyed a Coring contract in 2025, but is not going to enjoy one in 2026 nor 2027! This is a bummer and definitely portrays that Geotek might not have been the greatest investment, but the good news is that the market doesn’t seem to expect much from it either. The following commentary from management, however, seemed intriguing:

Does this imply that there might be any given year with more than one Coring contract? I don’t know, but it does seem that the Coring contract is significantly lumpier than previously anticipated.

Lastly but no less important, M&A has been non-existent. After historically acquiring around 2-3 companies per year, Judges has not been active on the M&A front for more than 2 years (the company’s last acquisition was Teer Coatings in August 2024).

So, what does all the above mean? That Judges has suffered the perfect storm, some of which was self-induced and some of which was outside of the company’s control. This would be my interpretation:

• Outside of the company’s control: US funding and China R&D freeze

• Self-induced/outside of the company’s control: Coring contract

• Self-induced: M&A

Not unexpectedly, the shares have fallen 70% from pretty inflated ATHs after the operational headwinds resulted in significant multiple compression:

Judges has always been a cyclical company, but I don’t think it has ever suffered a cycle to the same extent to the one it’s suffering today.

The interesting question is not whether H1 was bad because it’s pretty evident that it was but whether 200 pence of Adjusted EPS in 2026 represents the company’s new earnings power or an unusually depressed point in the cycle. I increasingly believe it is the latter and I’ll explain why below.

The green shoots

With tough comps behind us, things seem to have started to change. Positive news could be found on both growth levers (organic and inorganic), but let’s start with the organic side.

Despite the abysmal H1 numbers, management shared a couple of encouraging things. The first and most obvious one was that they maintained the full year guide. This is definitely encouraging but it’s important to understand that the guide is not risk-free. Meeting it is going to require a significant production ramp in Q4, so any small delay might carry revenue into the next year. This is nothing extremely worrying if you are focused on the long term (the end of the year is an arbitrary checkpoint), but would definitely not be great for the market’s perception. The good news, though, is that orders to achieve the guide seem to be “locked in” and therefore the problem is one of execution rather than demand.

The second good news on the organic side pertains to orders. Management shared something that I don’t think people really understood well from the press release (myself included). After a mid-teens LFL (like for like) decrease in orders in H1, management shared that the YTD (year to date) number was now closer to -1%. This strong recovery ultimately means that Q3 must have been an incredibly strong quarter in terms of orders (also easier comps). This is something that management confirmed during the call, more or less confirming that orders have grown at a 20%+ clip for 4 consecutive months. There’s good and bad news here:

• The bad news is that visibility in terms of how long order strength will last is very low

• The good news is that the order strength is being achieved without the US funding returning in any meaningful way and despite the R&D exemption delays in China

So, the good news is that there’s even upside to these encouraging numbers, but the bad news is that we don’t know if we’ve reached the inflection point yet. I would add yet another “good” news: I don’t really think the market is pricing that a strong inflection in orders is about to take place, so the upside doesn’t seem priced in.

The order chart that Judges shares does seem at least a tad encouraging. The LTM order growth line (black line) starting to inflect upwards for the first time since 2024:

Management is not expecting incrementally good news in terms of US funding. The exposure is considerable (around 15% of total Judges revenue is exposed to US funding) but not unmanageable. As for China, the management team does believe that orders are mostly delayed but that they’ll end up flowing through the books:

Now, there were also green shoots on the inorganic side. More than 2 years have passed since the company made the last acquisition, but Tim Prestige somewhat alluded to new acquisitions being closer:

I believe Judges’ management has always been conservative in terms of M&A, but they are maybe more conservative now after their Geotek experience. This is yet another potential source of upside but one that we shouldn’t really count on.

Some discussions around management

There were also certain things in the management team worth discussing. The first one is that Brad Ormsby announced his decision to step down as CFO of the company. I don’t think this is worrying because he has served for more than 11 years as CFO of the company. With Tim Prestige becoming CEO, a change of CFO makes perfect sense.

I believe Tim Prestige’s alignment also deserves a word or two. He has been granted two packages of 60,000 options each which, at the current share count and considering full vesting, would result in a 2% stake in the business. These options are based on a 5% adjusted EPS CAGR, and while this doesn’t seem like very demanding (it’s not), the adjusted EPS level they were issued at differs materially from where we are now:

• The 2025 grant vests on January 2028 and would require 328.1p of adjusted EPS

• The 2026 grant vests on January 2029 and would require 318.7p of adjusted EPS

The company expects around 200.5p of adjusted EPS in 2026 meaning that, for Tim’s options to vest, the company’s Adjusted EPS must grow 64% in 2027 (for the 2025 grant) or compound at a 26% CAGR over two years for the 2026 grant. Note that this includes M&A, so seeing Tim being very conservative in the multiples they want to pay is good news.

The 2025 grant seems tough to achieve BUT we must also not forget that Judges is a fixed cost business; earnings can go up and down quite materially with demand inflections and M&A can also do its part here, so I would say it’s not impossible. The bottom line is that Tim Prestige does seem aligned with shareholders as his base salary is relatively low and a good chunk of his compensation package is tied to these option packages.

Some thoughts on the valuation

Judges is currently trading at a P/E of 18x (on Adjusted EPS that strips out amortization of intangibles). This seems pretty low for a cyclical fixed-cost business that is starting to experience some green shoots. When it comes to cyclical but secular businesses, I always come back to Jerome Dodson’s quote which, at the risk of overusing, it I share once again below:

I believe Judges fits this mold but I’ve also been wrong thus far in terms of “timing” the inflection. The company has suffered far more headwinds than I previously anticipated and the thesis is definitely not free of risks yet. The good news is that the market seems to be increasingly discounting these risks, which is interestingly happening as green shoots emerge. Note that, despite the abysmal H1, Judges met the market’s expectations (which probably tells you how expectations have adjusted). I also decided to do a brief exercise to try to understand what has happened to Judges’ earnings power.

The market expects around 200.5 bps of Adjusted EPS in 2026. The last time the business generated similarAdjusted EPS was in 2019, or 7 years ago. In that year, Judges generated 19.8 million pounds of subsidiary EBIT, which ignores central costs. Since then, the company has acquired around 17-18 million pounds worth of EBIT. This means that subsidiary EBIT should result in around 37-38 million pounds (19.8M + 18M) assuming 0% organic growth in the existing 2019 companies and everything that was acquired after that.

Judges has, however, grown its central costs significantly (with no return to show thus far, which is not great). If we strip out around 6 million pounds of central costs from my calculations of subsidiary operating income, we get to around 31 million pounds in operating income “power” assuming no organic growth. Just to make it clear, this should be the operating income that Judges should be generating if organic growth would’ve been 0% over the past 7 years.

We then strip out 3-4 million pounds in interest costs and assume a 23% tax rate and we get to adjusted net income of 22 million pounds. Divide this by 6.66 million shares and we get to around 315 to 330 pence of adjusted EPS. So, what does all of this mean? Even when assuming that Judges has not grown at all organically since 2019, we get to an adjusted EPS level that’s 58% above the 2026 consensus estimate.

Judges has historically been a 7% organic grower but it’s fair to think that those organic growth rates are not here to stay after all that’s happened. Even so, it seems like current earnings are significantly depressed and still the company trades at 18x current earnings. Has something structural changed for Judges? Maybe when it comes to US funding, but it seems likely that the market is not assuming much in terms of a cyclical recovery.

The risk-reward seems appealing here, although if the market has demonstrated anything over the past few years is that risk-rewards can become more and more appealing over the short term (i.e., it can continue to go down).

For all of the above, I will add to my position.

Have a great day,

Leandro

You can read this article (almost) entirely for free. If you like what you read, consider becoming a paid member to get access to…

• All in-depth reports (16 companies profiled thus far, and growing)

• Earnings follow-ups

• Other investment related content

• A community of like-minded investors

• Complete access to my portfolio and transactions

• Occasional webinars

You’ll also get access to the 30+ page detailed report on the robotics industry.

The in-depth reports of Stevanato and Deere are also free to read to gauge the quality of the research. Other in-depth reports have been written about…

• Deere (DE)

• FRP Holdings (FRPH)

• Five Below (FIVE)

• Diageo (DEO)

• Hermès (HESAY)

• Atlas Copco (ATCO.B)

• Stevanato (STVN)

• Keysight (KEYS)

• Zoetis (ZTS)

• Judges Scientific (JDG.L)

• Medpace (MEDP)

• Trupanion (TRUP)

• AAON (AAON)

• BBB Foods (TBBB)

• Eurofins (EPA.ERF)

• Sabre (SABR)

• Rosebank Industries (ROSE.L)

• Undisclosed position

• Undisclosed position

Join today:

Subscribe now

Markets were mixed this week, with macro concerns now starting to loom in the background. Will this finally be the time when markets will really crash due to macro? Well, one never knows, but one always knows that markets will eventually crash (it’s rarely up and to the right with no bumps).

Without further ado, let’s get on with it.

A webinar next week

This upcoming week I’ll host a webinar to discuss the rationale behind some portfolio changes that I will make early in the week. There will also be time for Q&A so if you are a paid subscriber feel free to leave your questions in the comments or send me a DM.

The webinar will be held exclusively for paid subscribers and will take place on Wednesday at 1:00 PM EST. If you can’t make it, don’t worry as it’ll be recorded and you’ll be able to watch it offline. I’ll share more details early during the week.

Articles of the week

I published one article this week: “Is Hermes there yet?”

The stock (not the company) continues to tumble, so I wanted to shed some light on how one should interpret the current reported numbers and the multiple. I discuss what’s going on in the industry, discuss my valuation scenarios, and share why I added to my position.

I was surprised (because I was not expecting it) to see that the article has been one of the most liked and viewed in a while.

Early next week I plan on publishing an update on Judges Scientific (JDG.L).

Without further ado, let’s see what the markets did this week.

Market Overview

It was a turbulent week in financial markets. The indices continued to shrug off what seems to be like pretty relevant macro news and ended the week relatively flat. Again, with their fair share of volatility as one should always expect:

Where should I start? All the events are pretty much connected.

The markets began demonstrating a truth that investors seem to forget time and time again: the variables that matter the most are typically the ones that are hardest to forecast. Things such as interest rates, wars, tech disruption, and politics are evidently important for the well-being of the economy and the stock market, but knowing that they matter is very different than knowing (or claiming to know) how they will play out. This is the perfect combo for volatility: relevance + unforecastability.

Let’s begin with the not so great news: treasury yields surged again, breaking considerably above 5% and even reaching 5.20% on Friday, their highest level since 2007. There’s a combination of factors that’s likely playing a role in this; for example the expectations of higher inflation and the hawkish stance by the Fed. Higher rates are relevant because they increase the hurdle rate for every financial asset across the economy (i.e., equity valuation multiples should compress according to theory). Of course, if the reason for the rising rates comes from a stronger economy, equities might be able to counter the “lower” valuation multiple with faster earnings growth, but the market seems worried about the potential environment that higher rates might create.

Higher treasury yields also result in a higher cost of capital. This seems particularly relevant in a moment when a good chunk of the AI buildout is being financed with debt. We shouldn’t forget that this buildout is likely responsible for a good chunk of economic growth and index resilience. I was this wild stat this week on X:

The market might be worried about the fact that rates are rising while several companies are reporting that the consumer is starting to show some cracks. If you were alive and investing in 2022 this word might sound familiar to you: stagflation. The reality is that I’ve heard that we are going into an economic/stock market crash many times over the past few years and it always sounded very convincing. Will it be the case this time around? No idea honestly.

Part of the inflation that’s supposedly coming is caused by the war in Iran and high oil prices. The conflict has escalated over the past few weeks and oil temporarily rose above $100 (it has since corrected). There was good news on this front, though: seems like the US and Iran are ready (again) to begin diplomatic talks. We don’t know where these will take us but it’s a good sign nonetheless.

The final relevant event this week was Trump’s meeting with Xi (China’s leader) in Washington. I don’t think there was much great news besides a two month extension to the tariff truce, but it did seem as if it lowered the tension between the two most significant countries in the world (which is always good). I have talked with many people who have told me that they believe that geopolitical tensions have never been as high as they are today. While this may be true (pretty much impossible to quantitatively put a number on it), I think it’s also caused by a tad of recency bias.

Everything that we are currently living seems to be very worrying, but this is typically the case because we have forgotten how we felt in previous crises. Let’s not forget that we had almost double digit inflation and China doing military exercises around Taiwan not long ago. I would say that pretty much nobody thinks about those today, why? Because we typically remember that we lived through a crisis but don’t remember with what intensity we did so!

Now, leaving macro aside, I also wanted to share some thoughts on Muse (don’t expect to read what will happen over the next decade). You probably know that Meta has recently launched a new personal agent called Muse. The rollout of the product (primarily thanks to Meta’s distribution) was a huge success with some sources pointing to almost 3 million downloads in its first 12 days, even outpacing the rollout of ChatGPT in 2022 (which is something). This gave the market yet another thing to worry about:

I must say that I was surprised (and not surprised at the same time) to see so many people jump to conclusions on day 1.

Many argued that this is automatically terrible for marketplaces/aggregators, Google Search, and Amazon (due to their ads business). While I can obviously understand the rationale behind these concerns, I think that we should just allow some time for things to play out before being categoric. There are great use cases for Muse (many of which have been widely discussed on social media), but there also seem people trying to assign use cases to Muse that likely don’t make any sense. If one reduces everything to “customers will want the lowest price and will not care about anything else,” the discourse seems pretty straightforward, but it’s unlikely that things will pan out the way we think today.

In similar “forecasting” scenarios, I always come back to Mark Leonard’s quote on a recent Constellation call discussing the impact AI might have on software:

Admittedly, Muse is an entirely different beast to software in the AI era, but this quote is pretty relevant when trying to make such long term forecasts categorically. Of course, everyone in financial markets is likely going to talk their book. Many Meta longs will tell you that Muse is going to disrupt everything and become one of the best products ever. On the other hand, non-Meta longs will tell you that the product is stupid and that it sucks. The reality will likely fall in between those two and I can see arguments in favor of both.

I mean, the product does seem game changing because it’s likely the first time that the general population feels agentic AI. Now, at the same time, it seems like some people are assigning it use cases where it adds little to no value while giving an unproven product access to all of their data. Muse will likely not be the only agent out there in the coming months/years, and nobody knows how people will react once others start to come out, so it’s also likely too soon to claim that it’ll be incredibly profitable for Meta (as it’s still given out for free and at a cost to Meta). So, the conclusion from all of this is that products such as Muse do change the playing field, but I prefer to take the conservative route and update the probabilities as the evidence arises rather than shooting on day one (the market typically opts for the latter and this is potentially a source of opportunity).

The industry map was mixed this week, with semiconductors doing particularly well due to the arrival of the agentic era:

The fear and greed index improved slightly but remained in fear territory:

Company-specific news

Read more

The stock was trading at €1.600 then, but it has come down a further 15%+ and is currently trading at €1.350 or so. Just for context…the stock had not seen similar levels since 2022, so the absolute and relative underperformance compared to the indices has been significant.

I ended that article with the following:

Is this time different? Let’s take a look.

First things first: the reasons for the drawdown. I believe that there might be several plausible reasons behind the recent drawdown, some of which remain unchanged since I published that article. I’d quickly summarize them as follows:

1. China: the sentiment in the most important luxury market in the world has not been the best and many luxury companies continue to struggle in China. Despite all these struggles, Hermes continues to grow in China, albeit slower

2. The conflict in Iran: what seemed to be a “short” conflict at first has been lengthened considerably, which most likely means that the impact to companies like Hermes will be more long-lasting than previously anticipated

I would add a third potential reason this time around, which funnily enough makes for a fascinating story: Hermes’ missing shares.

Nicolas Puech (one of Hermes’ heirs) had historically been one of the company’s most significant shareholders. The thing is that his original 5-6% stake was handled by a Swiss wealth manager who went by the name of Eric Freymond. Puech is now alleging that Freymond sold part of his stake to Bernard Arnault against his will.

Everything happened around the year 2010 when Bernard Arnault was acquiring Hermes shares using derivatives in an effort to acquire the company against the family’s will. LVMH surprised everyone by building a 23% stake in Hermes, but the company was eventually forced to distribute a good chunk of these shares to its shareholders. This is the well-known story and, until here, all good.

Now, in 2022, Puech supposedly started to see something “strange” in his accounts: a good chunk of his Hermes’ shares were missing. He pointed his finger at Freymond and filed a criminal complaint for fraud in Geneva in September 2023, which was ultimately dismissed. Now, here’s where things get REALLY interesting. In 2025, Puech sued Bernard Arnault and LVMH seeking compensation of around €14 billion as he alleged that his shares had been sold to LVMH back when the company was trying to build its Hermes stake. LVMH denied the allegations and (importantly) denied ever wanting to acquire Puech’s stake. This ultimately meant that Puech did not have a way to prove that this had happened…until recently.

Reuters reported this month that some court documents demonstrate that LVMH signed an agreement in 2002 to acquire Puech’s shares, which goes directly against what LVMH manifested in a 2026 court filing. The more interesting piece is Freymond. According to Reuters, Freymond had been paid more than $20 million throughout a period of 8 years (2001 - 2009) in connection with LVMH’s effort to build its Hermes stake. It’s tough to envision a scenario in which Freymond could act in the best interest of both simultaneously, making Puech’s allegation plausible. Uncoincidentally or not, Freymond committed suicide in 2025 while the matter was being investigated.

So, what does all the above mean? Not only is China weak, not only does the Middle East conflict continue, but now we also have to face a “conflict” between the two most relevant luxury companies in the world. A pretty interesting environment for luxury!

The key question here is whether all of the above justifies Hermes’ stock price decline. That’s a different question entirely and one that imho is typically not well answered by the many who ignore all the moving parts.

Let’s take a look.

The moving parts and their implications

The first thing we must be aware of is that Hermes’ reported numbers (both the financials and the multiples) require significant context. Let’s start with the why. This is something that I discussed with paid subscribers in a relatively recent NOTW, but ultimately, if you look at the reported figures for Hermes you are most likely going to see the following:

1. LTM (last twelve months) reported revenue growth of 2.7% YoY

2. An LTM P/E ratio of 31x

What would most “quick-glance” investors say about these numbers? REJECT. Fine. But there’s considerably more context needed to understand these. Let’s begin with revenue growth.

Hermes’ has faced considerable FX headwinds over the last twelve months due to the strong appreciation of the euro. So, in short, FX has considerably pressured its reported growth. The reality though is that, despite all the aforementioned headwinds which are not minor, the company has managed to grow its top line at an FX-adjusted rate of 6% (H1 2026) with revenue accelerating from Q1 (+6%) to Q2 (+7%). Is this great? It’s definitely not the best that Hermes has ever seen, but it’s not something unheard of for the company.

The most interesting thing, though, is the implication that FX has on the reported multiple. Hermes hedges its FX exposure to protect margins, but when the euro appreciates, the company ends up with fewer absolute revenue and operating profit dollars. The opposite happens when the euro depreciates: both revenue and operating profit march higher even if organic growth stays constant. This ultimately means that Hermes’ valuation multiple can’t be understood without taking FX into consideration.

Now, there’s good news and bad news. The good news is that this means that the valuation multiple would potentially contract simply if Hermes enjoys a favorable period of FX. The bad news, though, is that we don’t know whether that will happen. I don’t consider myself an FX expert by any means, but if I had to guess, I’d say that the euro has appreciated quite materially against many currencies and see no reason why it shouldn’t give up a bit of it. Looking back at Hermes’ reported growth vs FX-adjusted growth, it sure seems like a cyclical phenomenon:

All this said, because an FX tailwind/headwind is totally unpredictable, I’ll simply focus on the organic performance of the business.

Now, the other relevant topic here is taxes, which is where things potentially get “structurally” interesting. In 2025, the French government decided to come up with a “solidarity tax” (they always call them solidarity) through which companies generating more than $3 billion in French revenue would be required to pay a “special and non-recurring” tax. The name in French sounds better (“contribution exceptionelle sure les benefices des grandes enterprises”) but it’s the European playbook (higher taxes to pay excesses) all over again but with a cute name.

The tax is pretty significant: an additional 41% over the corporate income tax used as the tax base. So this means that Hermes is paying roughly an additional tax of 10% over the profits of the revenue it derives from France (which is significant). For FY 2025, this special tax added almost 500 basis points to the consolidated effective tax rate.

The bad (or one could say: “the expected”) news is that this special tax is likely going to be more recurring than previously envisioned. It was initially thought for only 2025, but it was subsequently lengthened to 2026 as well. In August, the French government proposed one additional year: 2027. While the latter is a proposal, it seems like this is the start of a new tax for Hermes. The bad news for Hermes is that the company has its roots in France and going elsewhere seems unlikely. There’s a potential and non-trivial implication for investors from all of this: the current P/E multiple might be closer to normalized than it looks.

We could definitely strip out the special tax or normalize it, but what if it becomes permanent? You convert something in a source of downside rather than in a source of upside. This also means (and here comes the hard part) that all else equal, the EBIT (or any pre-tax multiple) that investors should be willing to pay for Hermes should now be lower. The reason is pretty straightforward: the French government is keeping a larger piece of the pie. Hermes is still an outstanding business that (imho) deserves a high optical multiple due to its high terminal value, but this is something worth discussing.

Let’s take a look at the model.

Is Hermes’ “there” yet?

I’ve divided the valuation into three scenarios: a bear, a base, and a bull. This is what I have decided to assume for each in terms of constant currency revenue growth:

For the bear case I assume revenue growth considerably below history, which I deem pretty unlikely. For the base case I assume that revenue growth remains as is, despite Hermes (and the industry) suffering considerable headwinds currently. For the bull case, which I also don’t deem extremely likely, I model revenue acceleration to 9%.

For operating margins I assume that they barely expand 100 bps over 5 years in the base case. I believe this is realistic as Hermes is currently investing significantly and margins seem to have stabilized around those levels:

Things get interesting in cash conversion. Hermes has historically enjoyed an OCF/EBIT cash conversion of around 85%. I decided to drop this to 80% in the base case due to higher cash taxes paid (which don’t impact EBIT but do impact OCF). This might prove to be conservative, but I believe it makes sense to do it this way if we assume that taxes will stay higher for longer:

Finally we have to think about exit multiples. I do believe that Hermes deserves a relatively high multiple due to its terminal value (especially when looked over a 5Y period). If we combine this with the fact that the EBIT multiple should (all else equal) be lower than history due to the higher taxes that I’m assuming, I decided to give the company an EV/EBIT exit multiple of 20x in the base case. The average EV/EBIT multiple has proven to be significantly higher in the past, but I believe these levels make sense in a conservative model:

Add the current dividend yield on top of this and we can finally calculate the total 5Y expected return CAGR for the three scenarios:

• Bear: 0%

• Base: 11.5%

• Bull: 19.7%

I believe the base case scenario is the most likely although I wouldn’t entirely rule out the bull case. I believe my scenarios are pretty conservative and that a 12% expected CAGR for a company like Hermes and its risk profile is pretty enticing. As I have historically had a small position (currently around 120 bps), I have decided to increase my position.

Have a great day,

Leandro

You can read this article (almost) entirely for free. If you like what you read, consider becoming a paid member to get access to…

• All in-depth reports (16 companies profiled thus far, and growing)

• Earnings follow-ups

• Other investment related content

• A community of like-minded investors

• Complete access to my portfolio and transactions

• Occasional webinars

You’ll also get access to the 30+ page detailed report on the robotics industry.

The in-depth reports of Stevanato and Deere are also free to read to gauge the quality of the research. Other in-depth reports have been written about…

• Deere (DE)

• FRP Holdings (FRPH)

• Five Below (FIVE)

• Diageo (DEO)

• Hermès (HESAY)

• Atlas Copco (ATCO.B)

• Stevanato (STVN)

• Keysight (KEYS)

• Zoetis (ZTS)

• Judges Scientific (JDG.L)

• Medpace (MEDP)

• Trupanion (TRUP)

• AAON (AAON)

• BBB Foods (TBBB)

• Eurofins (EPA.ERF)

• Sabre (SABR)

• Rosebank Industries (ROSE.L)

• Undisclosed position

• Undisclosed position

Join today:

Subscribe now

Markets were up during a week very few expected it to be up (surprise?). There were plenty of what many would consider relevant or significant news for financial markets, but markets seemed to climb a wall of worry once again.

The new in-depth report

This week I published another in-depth report. This time, I travelled to Japan to discuss a company…

• Growing at a 40% clip and expects to do so in the coming years

• That has a minor market share in a large and growing market

• Insiders own 50%+ of the outstanding shares

• Is trading at 11x EV/EBIT

I know it sounds too good to be true, but the reason why I started a position is that I don’t agree with the market’s perception on this one and therefore I believe the risk-reward is heavily skewed in an investor’s favor.

Second Japanese company to make it into the portfolio.

10,000!

This week Best Anchor Stocks reached 10,000 subscribers on Substack!

It has been a long and entertaining journey to get here, but thank you all for trusting this project. It will only get better from here!

As a thank you token for the 10,000 subscribers, I have decided to give out 10 annual subscriptions at a 30% discount. The discount will end as soon as the 10 spots get filled.

The subscription will give you access to all past in-depth reports (including the most recent one) and all of what’s coming (which is plenty!).

I believe the companies outlined in the last 5 in-depth reports remain attractive despite some having run up significantly since I published the report:

• Tiendas 3B (TBBB)

• Eurofins (EPA:ERF)

• Rosebank Industries (LON:ROSE)

• Undisclosed (a Canadian company)

• Undisclosed (a Japanese company)

You can sign using the link below:

Claim my spot

Without further ado, let’s see what the markets did this week.

Market Overview

It was a pretty interesting week in financial markets because we had a bit of “everything” and plenty of what many would consider critical news for financial markets. Just like last week, both indices finished in positive territory, but not without their fair share of volatility:

And like pretty much always, financial markets demonstrated that they are pretty much impossible to forecast. The Federal Reserve raised rates for the first time since 2023 (expected) and the commentary was pretty hawkish (maybe not as expected): 16 out of 18 members believe that there’s space to do one more interest rate hike this year.

Treasury yields climbed above 5% throughout the week, which together with oil above $100, formed a “perfect cocktail” for an end-of-day selloff on Wednesday. Many claimed that the selloff was logical, but in an expected (not a typo) turn of events, markets ended up in very green territory on Thursday and ended up on a positive note on Friday! So, is/was everything “priced in”? The “priced in” question is always an interesting one because despite many claiming that “x”, “y”, or “z” is priced in, it’s ultimately impossible to really know what’s in the numbers (directionally one can more or less have an idea). The reason is pretty straightforward: there are too many variables playing a role in financial markets. Of course people will tell you they knew what was coming after the fact while not promoting their mistakes, but if the inflation era portrayed something is that most “macro experts” don’t really have a more informed opinion than the average market participant.

We had a rumor this week that, in my opinion, is “terrible” for the well-being of long-term investors: US markets might expand trading hours to 23 hours a day:

Let me share some of my thoughts around this. The first and quickest one is:

The reality is that long term investors shouldn’t really care, but then at the same time it’s easier to avoid getting carried away by financial asset prices when the market is closed (this is a fact). No matter how long-term oriented you are, asset prices and volatility can influence your decision making (and being aware of this is important). All this said, I do believe it shouldn’t change one thing if you are long-term oriented: you’ll likely get more “bumps” along the way (volatility), but the destination shouldn’t change.

The second thought that comes to my mind is that this may be massively bullish for companies like Interactive Brokers (IBKR)? IBKR is (uncoincidentally) one of my biggest mistakes of omission. Incentives are very powerful and there are plenty of market participants with an incentive to lengthen market hours. All the intermediaries benefit and everyone is happy because it’s done in the “name of liquidity!” The market is simply transforming itself into a casino (not that it wasn’t before), which is interesting considering that people used to believe that markets would get more efficient over time. My contrarian thought is that markets are becoming less efficient over time (for long term investors that is) because they are becoming flooded with information and “quick” and violent trading. This doesn’t mean that they are always inefficient, but it does seem like stock prices move faster every year whereas businesses do not to the same extent. I recommend reading the intro to my recent in-depth report where I discuss this topic because I consider the company to be one of said mispricings.

The industry map was mixed this week, with semis doing pretty well despite the higher rate environment:

The fear and greed index worsened a bit:

Company-specific news

Read more

Financial markets are typically known to be fairly efficient machines when it comes to pricing assets (this could be considered their main job). The famous term known as “consensus” establishes that stock prices reflect a consolidated opinion from thousands of market participants about the public information available for any given asset. This is the theory, but reality is a bit more nuanced. While financial markets are indeed “almost perfect” in pricing assets over relatively long periods, misspricings can occur over shorter periods. These mispricings typically arise from diverging investment horizons and ultimately allow long-term investors to capture the spread between the current market perception and the true value of a given asset.

Now, the fact that the market CAN misprice assets at any given point in time doesn’t mean that it always DOES (even though some people seem to think that the market is “retarded”). This ultimately means that if we come across an asset that seems heavily mispriced we must think long and hard about the market’s variant perception (in some occasions there might not be one and the asset might simply be ignored for a host of reasons like liquidity or positioning). The above perfectly describes the company that you are going to read about today. Just to provide some context, today’s business has grown its revenue and operating profit at a 40% CAGR over the last 5 years, intends to do the same over the next 5, and yet trades at a 2026 EV/EBIT multiple of 11x. While it sounds too good to be true, management has already proven skeptics wrong once and benefited from it by remaining the largest shareholders in the business. While the market remains fixated on one side of the “equation,” management has clearly laid out that they moved to the other side of the equation long ago.

So, without further ado, let’s jump right into the business.

Read more

You can read this article (almost) entirely for free. If you like what you read, consider becoming a paid member to get access to…

• All in-depth reports (16 companies profiled thus far, and growing)

• Earnings follow-ups

• Other investment related content

• A community of like-minded investors

• Complete access to my portfolio and transactions

• Occasional webinars

You’ll also get access to the 30+ page detailed report on the robotics industry.

The in-depth reports of Stevanato and Deere are also free to read to gauge the quality of the research. Other in-depth reports have been written about…

• Deere (DE)

• FRP Holdings (FRPH)

• Five Below (FIVE)

• Diageo (DEO)

• Hermès (HESAY)

• Atlas Copco (ATCO.B)

• Stevanato (STVN)

• Keysight (KEYS)

• Zoetis (ZTS)

• Judges Scientific (JDG.L)

• Medpace (MEDP)

• Trupanion (TRUP)

• AAON (AAON)

• BBB Foods (TBBB)

• Eurofins (EPA.ERF)

• Sabre (SABR)

• Rosebank Industries (ROSE.L)

• Undisclosed position

Join today:

Subscribe now

Both indices were slightly down this week with significant volatility. We had two interesting earnings reports as well as an interesting appearance by Druckenmiller who shared some thoughts around the AI trade. There were also several interesting company-specific news.

Without further ado, let’s get on with it.

Back in the Chit Chat Stocks podcast!

This week I was lucky enough to be able to attend the Chit Chat Stocks podcast with Ryan and Brett. We had a great time and discussed BBB Foods (TBBB) (in depth report available here) and my robotics report.

You can listen to it on Spotify:

Or YouTube:

What you can expect over the next few weeks

There were no articles this week as I continued advancing several companies through the pipeline. In the following week you can expect…

1. An in-depth article on a very interesting (and unknown) Japanese company growing at a 40% CAGR while trading at an 11x EV/EBIT multiple

2. An in-depth article on one of the most controversial companies one can currently find in financial markets. Yep, it’s Oracle (ORCL).

3. An in-depth article on an aerospace company that I believe is not yet well understood

So, plenty of things are coming to Best Anchor Stocks over the next few weeks.

Without further ado, let’s see what the markets did this week.

Market Overview

Both indices were down slightly this week, not without their fair share of volatility:

In yet another example of how unpredictable markets can be, both indices were significantly up on Friday despite inflation coming in hotter than expected. A lot of people thought indices would be considerably down because hotter-than-expected inflation gave the Fed more reasons to raise rates, but the markets did the exact opposite (I kid you not!).

We had a couple of interesting earnings this week. Adobe and Oracle both reported on Thursday, and once again, the first algo-driven move proved to be the “wrong” one. Adobe was initially down on earnings and Oracle was up, but the movements flipped throughout the day and Adobe ended up with Oracle down. I have been eyeing Oracle lately, and the main problem I see with the Oracle investment thesis is that I don’t think that bear concerns can be alleviated for a while:

Oracle claims that the unit economics on the Capex that is currently coming online are great, and while this is evidently true, the reality is that we are not currently in a normalized supply-demand environment. One is likely to see incredible economics when demand far exceeds supply, but this doesn’t mean that they are sustainable economics. The key here is to understand whether the different players (Oracle, the hyperscalers, or the neoclouds) have any competitive advantage to be able to charge a premium price for each GW of capacity once the supply normalizes. Some people argue that the normalization will take a long time to come (fair), but it nonetheless becomes a question of “when” not “if.” We’ll see.

Frankly, I believe the same is true for Adobe in terms of the market’s visibility into the bear cases. Despite many Adobe bulls claiming that the quarter was spectacular, RPO and ARR growth continue to decelerate. Management claims this is a self-inflicted wound as they push the freemium model through, but evidently a long-term concern has become whether they’ll be able to convert these users to long term Adobe paid customers. So, something that can’t be denied is that Adobe is growing faster today than it was a couple of quarters ago while enjoying a compressed reported valuation (both things are facts), but one can’t also not ignore that the leading indicators are not painting a rosy picture. Is this worrying? Well that’s where one needs to do more work, and frankly I don’t believe that a lot of people are focusing here.

In other (albeit related) news, Stanley Druckenmiller participated in a private conference this week. As one of the greatest investors of all time, I believe that paying attention to whatever he has to say is a must. Druckenmiller shared some thoughts around the AI trade and claimed that “there’s a good chance we’re in an earnings bubble because this AI build-out is going to end at some point, and let’s face it, banks are also on the AI trade. I mean, these guys are making hundreds of millions of dollars when they bring these companies public.”

I echo his thoughts in the sense that I believe there’s both a bull and a bear case to be made about the AI trade (there always is). The bull case is two-fold:

1. The technology is a revolution. I believe few people can claim the opposite at this point.

2. It remains far from being penetrated.

As for #2, I can only share my anecdotal evidence. It seems pretty obvious that X is a bubble in which people use AI much more than people do in the “real” world. I have coded agents to help me do things faster and improve my productivity, but in my close circle of people there’s not one single one of them who has ever coded (or uses) an agent. According to Jensen Huang (and this is something I went over in my robotics report), agentic AI requires around 10x the computing power of generative AI, with the robotics opportunity being yet another 10x of top on that. We seem to be extremely early in agentic, which means that the runway seems pretty long.

Now, the bear case to be made is that plenty of people believe there’s no limit to AI. I would be very careful thinking this because everything has a limit and we might encounter bottlenecks along the way that pretty much nobody expected. So I must say I’m torn between both, I do believe AI still has plenty of legs, but I also don’t think that a lot of the companies believed to be AI winners will be winners in the end and that there’s obviously a limit to the technology, probably driven by adoption/bottlenecks (pick your fighter). I mean it’s fine to make larger, larger, and better models, but maybe there will be a time when each incremental intelligence unit can’t be monetized economically because what we have is good enough for most tasks (just thinking out loud here).

The industry map was mixed this week, showing some very green pockets in semis, hardware, and communications equipment:

Source: Finviz

The fear and greed index retraced further into fear territory:

Source: CNN

Company-specific news

Read more

You can read this article (almost) entirely for free. If you like what you read, consider becoming a paid member to get access to…

• All in-depth reports (16 companies profiled thus far, and growing)

• Earnings follow-ups

• Other investment related content

• A community of like-minded investors

• Complete access to my portfolio and transactions

• Occasional webinars

You’ll also get access to the 30+ page detailed report on the robotics industry.

The in-depth reports of Stevanato and Deere are also free to read to gauge the quality of the research. Other in-depth reports have been written about…

• Deere (DE)

• FRP Holdings (FRPH)

• Five Below (FIVE)

• Diageo (DEO)

• Hermès (HESAY)

• Atlas Copco (ATCO.B)

• Stevanato (STVN)

• Keysight (KEYS)

• Zoetis (ZTS)

• Judges Scientific (JDG.L)

• Medpace (MEDP)

• Trupanion (TRUP)

• AAON (AAON)

• BBB Foods (TBBB)

• Eurofins (EPA.ERF)

• Sabre (SABR)

• Rosebank Industries (ROSE.L)

• Undisclosed position

Join today:

Subscribe now

Both indices were slightly up this week and I found two concepts that I wanted to discuss in the brief market commentary. I also share the articles of the week and the news of the week, including some highlights from Rosebank’s (you can read an article about the company here) most recent earnings.

Without further ado, let’s get on with it.

I’m back and articles of the week

After a week of resting and recharging the batteries…I’m back! I published one article this week: the seventh edition of On The Radar.

I profile three very interesting businesses that I am currently looking further into:

1. An unknown engineering business with growth capped by supply rather than demand

2. An aerospace aftermarket business with a very interesting and controversial model

3. A very interesting set-up in a business with a balance sheet “problem”

If you are looking for some fresh ideas, you can check out all the seven editions of on the radar (22 companies profiled) here.

Without further ado, let’s see what the markets did this week.

Market Overview

Both indices were slightly up this week, with the SPY outpacing the gains of the QQQ:

I would like to briefly discuss two topics in today’s NOTW. The first one is the concept of the random walk. Something that has become abundantly clear to me over the past few years is that valuation doesn’t matter one bit over the short to medium term. When I looked at my portfolio at the beginning of 2026, I made an effort to try to guess which stocks would go on to perform better. I naturally went for those that enjoyed the most appealing valuation multiples, but here’s what happened.

My best performing stocks YTD have been (I’m only counting those that I have owned for since 01/01/2026)…

• Keysight Technologies +58%

• Deere & Company +49%

• ASML Holding +49%

• Texas Instruments +46%

• BBB Foods +45% (counting this one since I owned it since January)

Maybe with the exception of ASML, I wouldn’t in my wildest dreams would’ve thought these would be the companies carrying my portfolio. I know there’s a lot more to valuation than the reported multiple, but here’s the EV/EBITDA multiples these companies were trading at to start the year:

• Keysight Technologies 28x

• Deere & Company 16x

• ASML Holding 25x

• Texas Instruments 21x

• BBB Foods 65x

I could go on to explain why some of these multiples are not normalized, but the conclusion would hold: valuation multiples are a terrible leading indicator of performance over the next 12-24 months. My good friend WTCM shared the following on X this week:

I agree with it entirely and my view has been shaped by my experiences, which might be anecdotal, but they are experiences nonetheless. I would argue that valuation always matters over the long term, it’s just that one never knows when it’ll start to matter! If anything, momentum seems to be a much better leading indicator than valuation in the current market.

The other topic I wanted to talk about is a concept that I’ve decided to coin as “holding bias” (it 100% has another name, but I decided to coin it with that one). This week I published the following about Adobe’s CEO succession:

The post got many replies claiming that “I wouldn’t be saying this should the stock be at ATHs.” While I definitely agree with the notion that we are most likely inclined to overthink the bad news when a stock is doing poorly, I believe that these comments portray that some people get too emotionally attached to their holdings. It’s fine to hold something while not agreeing with everything management does or not believing that everything is perfect.

Adobe may well play out brilliantly as an investment over the long term, but the great performance will not be due to the great CEO succession the company has done, but rather IN SPITE of it. If one actually believes that Adobe has handled the CEO transition well…they should try to ask themselves whether maybe they are too emotionally attached to their position.

On the other side of the spectrum we have Constellation Software whose founder and president (Mark Leonard) had to retire due to health reasons. The company quickly appointed Mark Miller to the position, a 30-year CSI veteran who came on board with the company’s first acquisition and decided to take a $1 salary (just like Mark Leonard’s). The succession events of Adobe and Constellation are night and day, which doesn’t mean that it’ll be a relevant thing when looking at these businesses in 5-10 years. Now, we should call things by their name.

The industry map was mixed this week:

The fear and greed index retraced considerably and entered fear territory:

Company-specific news

Read more

If you want to know what this series is about, I recommend reading the first issue. Let me save you the click with a short summary: the goal is to share interesting companies that I’ve looked at but have not researched in-depth yet (at least not to my standards). Therefore, the “On The Radar” series should be understood as a well-curated list of companies that fit my investment philosophy and that might (in the future) make it into the portfolio.

So far 2 out of the 19 companies I have profiled in this series have made it into the portfolio, with good (but still early) results: +17% and +5% (in not much more than two months). The average return of the companies profiled thus far is around 10%, albeit this series started not long ago.

Today’s issue brings yet another three companies (one for free and the remaining reserved for paid subscribers) that I’ve been looking into or that I plan to look into in the coming weeks/months. Some of these will most likely end up as in-depth reports:

1. An unknown engineering business with growth capped by supply rather than demand

2. An aerospace aftermarket business with a very interesting and controversial model

3. A very interesting set-up in a business with a balance sheet “problem”

Note that all of these businesses get included in the “On The Radar” tab of the Portfolio Management Tool with their respective price evolution since the issue.

As this is the seventh issue, free subscribers have had access to 7 companies (one for free per issue) whereas paid subscribers have had access to 22!

So, without further ado, let’s jump into today’s issue.

Friedrich Vorwerk (ETR:VH2)

Brief description of the business

In all honesty, I’d be very surprised if you’ve heard about Friedrich Vorwerk. This is not in any way criticising your knowledge about businesses because, for what it’s worth, Vorwerk is a small cap ($1.39 billion) German company that operated in a pretty boring industry. At its core, Vorwerk is an engineering services company specialized on energy infrastructure projects in Germany.

There are several things that make the company rather special/interesting. For example, Vorwerk is vertically integrated across specialized equipment and labor, two tasks that its peers generally choose to outsource. This helps the company earn higher margins than peers (albeit carrying higher capital requirements) while being able to manage projects in a more organized manner.

Even though the company specializes in services, we can also find several technologies at the core of its competitive advantages:

1. 5C Tech’s (a subsidiary) automatic welding system. This technology not only lowers the need for specialized welding labor but also carries lower defect rates

2. A specialization in Horizontal Directional Drilling, a key technology for laying underground pipes/cables

The company was historically exposed mainly to natural gas projects, but these have increasingly been “substituted” by “new energy” such as electricity, CO2 transportation, and (in the future) hydrogen. The way Vorwerk makes money is pretty straightforward: a new energy infrastructure project gets tendered, Vorwerk bids (and hopefully wins it) and gets paid throughout the project as they execute it. This revenue cycle must be repeated over and over again (i.e., there’s very little recurring revenue related to infrastructure maintenance, although there is some).

Why it caught my attention

The company caught my attention after looking into MBB SE, a German serial acquirer of industrial companies. MBB still holds a significant position in Vorwerk and the latter’s stock price evolution seemed pretty intriguing.

The stock got clobbered (down 77%) after IPOing at a decent valuation of 9x EBITDA, but recovered in spectacular fashion from the lows to deliver an almost 10-bagger at the peak:

What was the reason behind such obnoxious mispricing? This is what ultimately got me interested in Vorwerk. After a bit of reading, I found out that the “mispricing” was caused by inflation. 2022 and 2023 (as you know) were years of considerable inflation after the post-pandemic spending surge, and Vorwerk had not included inflation-escalation clauses in its contracts. This evidently led to significant margin contraction. However, things slowly but steadily turned into Vorwerk’s favor: these legacy contracts eventually ended and the company included inflation-linked clauses in new contracts. The war in Ukraine also made the German government worry deeply about energy independence, leading to a considerable pipeline of projects. So not only did margins rise considerably, but Vorwerk also saw a significant influx of new business. EBITDA inflected hard, rising from €32 million to more than €200 million only a couple of years later (so much for efficient markets!):

At €200 million in EBITDA and an EV of €995 million (the company has a considerable cash position), Vorwerk is currently valued at an EV/EBITDA of 5x. This seems low on three dimensions:

1. The path the company is in and the secular growth drivers it is exposed to

2. What the business has historically traded at (an average of 8x)

3. What peers trade at

To this we must add that it seems pretty obvious that the current limitation for growth is not demand but supply (mostly labor). My research into Vorwerk is a quest to answer the following question:

What I like

There are several things I like about Vorwerk:

• The company is exposed to significant long-term tailwinds as Germany rebuilds its energy infrastructure after the Russia “scare”. There doesn’t seem to be a shortage of energy project currently and many of these are long-term in nature

• There’s considerable skin in the game. The current CEO owns more than 18% of the shares outstanding and MBB owns around 40%

• The business is currently supply-capped, meaning that its fate is currently dependent on execution. Management has been executing well with labor supply growing at a 10%+ rate over the past few years

• The labor constraint discussed above acts as a significant entry barrier because potential disruptors can’t source labor. The largest players are increasingly making acquisitions to source labor and consolidating the market

• The company is relatively unknown and trades at an (apparently) low valuation multiple

• Management is not afraid to follow their own path and are not driven by industry standards (proprietary tech, vertical integration…)

• The cash on the balance sheet gives management considerable investment optionality, both organic and inorganic

• There’s also optionality in the company’s automatic welding technology, which has opened up international opportunities (mainly in Asia)

• I also believe there’s a miss-understanding about the company’s non-IFRS metrics such as backlog and orders. The large projects are mostly managed through JVs (Joint Ventures), but Vorwerk only includes in the backlog and orders the equipment and labor they “sell” to the JV. This is not necessarily representative of the money they’ll make through the JV, meaning that these metrics are not great leading indicators (very similar case to Medpace)

Now, despite Vorwerk having a lot of things to like, there are also several things that give me pause.

What gives me pause

The main thing that gives me pause is the overexposure to the following…

1. Germany

2. Underground projects

3. Natural gas, electricity, and hydrogen

#1 is straightforward to understand, but the second and third maybe less so. Even though laying underground infrastructure is 100% the safest and most efficient long-term bet, it’s not the cheapest. The German government has hinted at above ground infrastructure becoming a possibility to save on costs. If underground projects shift to above-ground projects, Vorwerk’s involvement in the infrastructure build-up will most likely decrease. Management believes this is a longer-term problem (if it ends up playing out) as the projects that are already in the pipeline last at least until 2035 and those will certainly be underground projects. The third point there is also interesting, because any kind of a return to nuclear energy (probability seems low), would also reduce the demand for Vorwerk’s expertise.

Lastly, another thing that somewhat worries me is a revenue cliff. Vorwerk has executed over the past few years a very sizable project called A-Nord. This project lasts until 2027 and will need to be replaced with new business, with management claiming that it will. The reality, though, is that revenue growth has decelerated significantly once the project has started to have a less significant participation in the company’s revenue.

What I am watching

I have been trying to speak with IR/management to clarify some doubts, with no success thus far. I will keep trying, but I don’t anticipate owning Vorwerk unless I’ve had a chance to speak with management and settle some doubts regarding the accounting of the JV pipeline.

Let’s now jump into the next company.

Read more

You can read this article (almost) entirely for free. If you like what you read, consider becoming a paid member to get access to…

• All in-depth reports (16 companies profiled thus far, and growing)

• Earnings follow-ups

• Other investment related content

• A community of like-minded investors

• Complete access to my portfolio and transactions

• Occasional webinars

You’ll also get access to the 30+ page detailed report on the robotics industry.

The in-depth reports of Stevanato and Deere are also free to read to gauge the quality of the research. Other in-depth reports have been written about…

• Deere

• FRP Holdings

• Five Below

• Diageo

• Hermès

• Atlas Copco

• Stevanato

• Keysight

• Zoetis

• Judges Scientific

• Medpace

• Trupanion

• AAON

• BBB Foods

• Eurofins

• Sabre

• Rosebank Industries

• Undisclosed position

Join today:

Subscribe now

Both indices were down this week, although the Best Anchor Stocks portfolio performed pretty well. The reason was something that happened this week related to the healthcare sector, which comes with several important lessons for investors (discussed in the market commentary).

Without further ado, let’s get on with it.

Out of office next week

I’m taking a holiday next week so I’ll be away from my computer. There will be no articles and no NOTW, but I expect to bring an in-depth report during the first half of September and will continue to read about several companies I have in the pipeline.

Articles of the week

I published two articles this week. The first one was (as promised) an update on Tiendas 3B (TBBB).

The company reported spectacular earnings and continues to demonstrate it is taking the Mexican grocery industry by storm. You can also read my in-depth report here.

The second article of the week was Deere’s earnings digest.

The results were better than expected, but I believe there was one specific quote that got the stock moving. I explain what this is and why it might be relevant for the investment thesis.

Without further ado, let’s see what the markets did this week.

Market Overview

Both indices were down this week, with the Nasdaq falling more than 2%:

I read Crossroads Capital’s Q2 2026 letter this week (written by my good friend Ryan O’Connor) and liked the following quote (among other things):

If you’ve followed my work for a while you’ll know that my investment philosophy has been shaped through the years (and I don’t anticipate it’ll be static going forward). The most significant change has been that I used to despise cyclicality, but now I actually strive to find it. Everything began when my good friend John Rotonti interviewed Jerome Dodson (founder of Parnassus Investments). The following quote from that interview stuck with me and shaped my investment philosophy:

I believe that the two quotes above are intrinsically related. Ryan claims that clarity and the rally are the same event, and I agree with him because the good returns are found when there are clouds, not when the sky is clear. I believe this type of scenario is much easier to find in the type of situations that Jerome Dodson describes in his quote: cyclical but secular industries where many investors remain on the sidelines waiting for an inflection. In many cases, people will be simply waiting on the sidelines for things to start working, hoping to hop on to the momentum trade (a phenomenon that has been shaping markets for a while).

We got a pretty good example of the dynamics described above this week. If you are not living under a rock, you might have seen that Moderna (MRNA) and Merck (MRK) released some pretty relevant news this week. Both companies announced positive results from their INTerpath-001 Phase III trial (albeit they did not share the data as such, which is an important caveat). This was relevant because it’s as close as humanity has gotten to a “ personalized cancer vaccine.” Now, there are certain caveats worth pointing out here because it’s more of an adjuvant treatment than a cure, but it seems to stop (in some cases) the cancer from coming back once the tumor has been removed. However you want to spin it, this is great news, especially since it somewhat demonstrated that there’s a potential mRNA platform to achieve personalized cancer treatments. Moderna’s shares rose considerably (+177%) on the headline, but what’s most interesting about this event is how it shifted the sentiment in the healthcare sector.

I’ve been writing for a while that sentiment across the healthcare sector had not been the best, driven by a violent post-pandemic boom and bust that had left the top lines of many companies starved for growth. Many investors had been waiting on the sidelines for “healthcare to start working” but little did they know that what got the healthcare sentiment shift started (still too soon to know if it’ll be sustainable) was an “unexpected” event that doesn’t even impact the top lines of these businesses (at least not immediately). The XLV ended this week up 4.5%, a significant contrast against down markets:

So, what is the bottom line of all of this? My reading is that timing inflections is impossible, but they eventually do happen for those willing to wait. There are very few people willing to hold something that’s “not working” (especially since there will be a lot of things that are working at any given point in time) and herein lies one’s opportunity. Now, granted, I don’t know if this will sustain or not, but the lesson doesn’t change: inflections eventually happen, we just don’t know when. Seems like a good idea to buy those companies in which the inflection is not priced in even though it’s a sure thing that it’ll happen. You might not get the highest IRRs in these situations, but you sure will get attractive risk-adjusted returns (which is what any investor should strive to find).

Something interesting that typically happens in financial markets is that a small inflection shifts sentiment a bit, then sentiment drives price, and price continues to drive sentiment in a self-reinforcing loop. This has happened time and time again (both to the upside and downside) and I don’t expect it to change anytime soon.

The industry map portrayed what I discussed above: a somewhat red market but with healthcare pretty green:

The fear and greed index retraced a bit and is now in neutral territory:

Company-specific news

Read more