Good morning, everyone. I'm Tommy Moll, an analyst here at Stephens. We appreciate you joining us in Nashville this week for the Annual Investor Conference. To my left, I'm joined by the CFO of Cognex, Dennis Fehr. Dennis, thank you for your time and insight today.
Yeah, thanks, Tommy, for having us. Thank you, everyone, for being here and your interest in Cognex.
This conference is intended for both specialist and generalist investors. With that in mind, I'll start with some introductory questions for the generalists in the audience who maybe aren't yet familiar with Cognex. Then we'll move to some of the more specific questions. If at any point you want to ask a question directly, by all means, shoot up a hand and ask one or more than one question that might be on your mind. Just to kick us off, Dennis, Cognex is a machine vision competitor. Tell us what that is for those who may not have dug into your market before.
Yeah, no, absolutely. Think of us as a software company, software on device. That means we are working in factory automation and warehouse automation. Maybe to give you a few examples of what we do, I would reference a customer visit I did earlier this year. I was in France and visited a famous kind of cosmetic manufacturing brand. They are manufacturing only in France. We went into that factory and walked the line and saw a couple of applications. Basically, what is this company doing? Obviously, they create, they would call it magic, probably, in kind of a cream form. At the end, it goes into glass bottles. These glass bottles at the end get packaged into kind of some carton packaging and being sold on shelves, right? Think about that.
In that regard, applications could be, for example, that they need to feed in these glass bottles into the production line. They have some robotic arms, basically. They have pallets of these little glass bottles coming up. They have robotic arms coming and picking these glass bottles up and placing them on the production line. You have the guidance of that robotic arm that would have a machine vision system to really identify where the bottles are and where to place them. That's one of our applications, guiding. From there, it goes further that these glass bottles, going further down the production line, eventually, they may fill the cream into it. You will want to know, like, oh, maybe along this way, was there any damage to the glass bottle?
One of the big issues would be if little glass splinters maybe even drop into the cream or some outside defects. That means we do such kind of inspections. That means we have the cameras placed there. The cameras would look at these glass bottles and would see are there damages in the glass. There is an inspection application as well. As this is also a kind of highly quality environment, they would like to have some serialization of their products. They have little numbers printed on these glass bottles, very hard to see. We would basically, in each step of the production, read out these serial numbers so that they basically can tie back the step of the production to that particular glass bottle. That means there is some kind of identification application with our cameras as well.
Maybe the last application, as an example, would be a measurement application, a gauging. In this case, for example, before they pack the glass bottle, the filled glass bottle into kind of that carton wrapping, they want to make sure that the carton is really cut out in the way they want it. Because if they feed in the carton into the machine and it's not cut in the way, then they get a jam. That means they want to inspect and measure, basically, is the dimensioning of this carton in the right way. There is some kind of this gauging application as well. Think about it that along this production process, there are different areas and applications which we can solve. That is kind of what we bring to these types of customers.
We are active in different verticals, maybe to wrap that up, right? This was an example. We call this market really packaging because it is all about packaging. It is food and beverage, health care, cosmetics. We are active in the logistics vertical. Think about warehouse automation. A lot is about code reading there. Automotive is a big market of us, consumer electronics, kind of device manufacturing, and then semiconductor manufacturing. These are kind of our top five verticals, so a diverse set of markets we sell into.
Let's think about how artificial intelligence factors in here, Dennis. First, on the opportunity side, this is something that Cognex has talked about for a while now. Give us a sense of how AI can improve the ease of use and expand the use cases.
Right. Yeah, so we think AI is one of a great opportunity for Cognex, right? So we are a 40-year-old company. And we have been basically the technology leader in that time. That was basically when we talked about what we would, from today's perspective, call rule-based software architecture. That means you have very smart people sitting there writing codes, saying, what if this, then that, right? Back to that example, maybe with the glass bottle inspection, we would try to write a code and to say, if you see this, then it's good or bad. You can do a lot of things with that. It can become very complicated to write such a specialized code. The question is, is it repeatable? That is here where really AI models. Basically, we created our proprietary vision model, basically, which then is loaded on devices.
The customer only needs to train a few of their own examples. In this case, like, for example, some of these glass bottle inspections, you would need to show a couple of pictures, good, no good. The model would learn that. With that, you basically achieve higher accuracy. That means you can solve more use cases for the customers, maybe, which you could not do in the past. Like in the past, without AI, it would have been very hard to inspect like the top edge of the bottle for glass damage inspection because they are so small. In rule space, you could just not do it. With AI- based, you can. That means out of a sudden, you have an additional use case.
It means the customer, instead of having a person looking at it, all of a sudden, basically can put one of our cameras there. That means we are creating basically market with that. That is the great opportunity which we see as bringing AI-based technology into the machine vision market.
Now, on the potential threats from AI, in particular, open source substitutes, how do you protect against what I would call a good enough solution that may cost a lot less than Cognex?
Right. I mean, see, I think most would feel like that's probably one of the biggest misunderstood parts in some of the Cognex story over the last 12 months or so. A lot of people thought like, oh, are you being disrupted by AI? There are all these hyperscalers creating these large models. Will these large models all be able to do what you can do very easily? I think we showed at our investor day in June this year that really four distinct topics which you need to have to be successful in this market. First, you need to have high precision, right? That means you can't miss a damage on that bottle. Glass has fallen and you miss that. That means precision, very high accuracy you need to have.
For that, you really need to drain it on creating this proprietary vision model, which we have. It really is very highly specialized to the applications which we do in our markets. Kind of generic language models can't solve this. They might be trained on a variety from pictures of horses to mountain landscapes. Our models are really trained on very concrete and detailed factory automation examples. Precision matters. We can prove we are better there. Second, speed matters a lot, right? In this example where I visit these French customers, the bottles move so fast you can't see them with your eye, right? It looks like maybe they move very slowly. They move so fast your eye can't see the individual glass bottles anymore.
In that regard, you can't use a very large language model because the processing time would be much too high. You need to have a very specialized model like ours. You need to bring it on device. You can't do it with the cloud because your connection time, your latency is too long. Therefore, you need to have that capability to match the hardware with the software, so to say. You want to have scalability. That means, right, you have a manufacturing line. Today, you produce product A. Tomorrow, you may want to shift to product B, right? The bottle might look different. The colors are different. The shapes are different. You don't want to recreate everything. You want to have scalable models. That's where our models help a lot.
The last piece is ease of use. You do not want to have very complex. You want to have as much as you can plug and play solutions. That is, I think, where we especially here we invested a lot. We are also AI helps kind of self-tuning, self-setup, where we created these types of features. I think that is how we differentiate. That is why we are not worried about these types of larger generic types of models or good enough types of approaches.
Thank you, Dennis. I want to move on to discuss some of the elements from your recent investor day, just starting at the top and working our way down the P&L. In terms of the multi-year sales outlook, you framed it in the low double-digit range in terms of the compound rate. One of the key levers to enable that was doubling the customer count over roughly five years. What are the elements of the strategy that you're pulling together to achieve that?
Right. Here, I would say the growth algorithm was which we introduced at the investor day. Really looked at the vertical markets, which I mentioned before, which we're serving. We then basically looked at and said, like, what's the underlying growth rates in these vertical markets? They're clearly not in this low double-digit range. They're maybe in a low single-digit range. We see a strong opportunity by increasing market penetration. Market penetration can come from two areas: the AI side, which we just talked about, like that means creating more applications and with that driving more penetration. The second piece is what Tommy just referred to. That means it's like bringing machine vision to more customers. We traditionally, as Cognex, have been serving kind of large-scale sophisticated customers. We saw that there's a long tail of the market as well.
That almost brings me to, I would say, the second point, which probably has been most misunderstood about Cognex in the last 12 months or so, right? I talked about the AI and how we think that's an opportunity. The second piece is about our transition from emerging customers to Salesforce transformation, right? That means maybe in 2023, we launched an initiative called Emerging Customers. That means, like we said, like, let's put boots on the ground and go to more customers. It was more like expanding Salesforce to serve more customers and give them easy-to-use products. It was a good strategy to say, like, let's tackle that long tail of the market with really the focus to create more customers. Since then, we have really evolved what started as an emerging customer initiative to what we now call Salesforce transformation.
We think that's not just a rebranding. That's really kind of a next step in the evolution along that line. If you think back, emerging customers, boots on the ground, versus if you think about Salesforce transformation, first of all, it's not standalone. It's really creating a combined structured Salesforce with very dedicated sales profiles. That means we thought a lot about, like, what type of sales profiles do we need selling to what type of customers? We structure that in three groups, basically newer, let's say, smaller customers, then kind of the larger existing customers, and then machine builders, which are very kind of spec-in type of sales approach. On top of here, and here's really a very distinct view, we are focusing a lot about sales efficiencies and data-driven sales analytics.
That means bringing CRM tools and then also the way how we manage our sales organization in terms of KPIs, leaderboards, and how they perform on a day-to-day basis, on a weekly basis, on a monthly basis. That means that we move basically from more boots on the ground towards a focused and sales efficiency-driven kind of Salesforce transformation. That means with Salesforce transformation, we not only want to reach more customers, but we also want to increase the efficiency of our Salesforce and basically to optimize the dollar spent per dollar booked. That is really what is key difference between where emerging customer initiatives started. That is why we think it's really part of an evolution. I think a lot of people maybe thought it's just a rebranding when it's not.
Moving down the P&L here, I'll hit a couple of the elements in your expense lines. Then you can answer it in whichever order makes the most sense. On the gross margin line, you moved away from an explicit target there. On the R&D line, you're emphasizing much more than previous the efficiency of that R&D spend. More holistically, at the OpEx line, you've talked about more dynamically managing those expenses through a cycle. I've thrown three different themes at you to hit on them in whatever order you choose.
Right. Yeah, maybe start on that gross margin piece first. First of all, I think, right, we are, say, attractive gross margin company, traditionally slightly above 70% right now, maybe in the high 60% gross margin range. It's an important lever for us, right? I think in the past, Cognex, I would say, almost over-indexed on gross margin. I would say almost from also internal decision-making to say, like, are we focusing on gross margin? Or are we looking at shareholder value creation, which we think is much more in terms of adjusted EPS growth? In that regard, I think there's a bit of a shift in terms of financial management of the company happening, which, however, does not mean like we're not caring about gross margin, right? We just think like it's one lever.
It is also very clear that with that high gross margin we have, our COGS are smaller in absolute dollars than our OpEx. That means there is much more we can work on in OpEx than on COGS. That is why I think OpEx has become a much higher focus and why we talk much more about it in the last, especially since investor day, where we clearly also outlined that our path to a higher profitability is through OpEx efficiency, right? In that regard, it has been clearly a focus area here. I think over the last three quarters, we made great progress here with revenue up in each of these quarters and adjusted OpEx in absolute terms down. Really driving nice efficiencies and flow through to the bottom line. Yes, it has different components when we think about the OpEx efficiency.
There's clearly the R&D piece to it, where we have two things. First of all, thinking about how we do capital allocations in R&D, that means bringing more financial metrics when we make decisions about into which R&D projects we allocate funds and where not. I think that's a shift, something which you can't see from the outside, but it's happening on the inside. There's clearly efficiencies which we drive, right? First of all, over the last couple of years, we migrated from a product line software stack to more a platform architecture. That means we are creating synergy effects across product lines and software coding. Then adopting AI-assisted coding is a big productivity driver in our engineering function as well.
That allows us basically that we think we can achieve the similar level of innovation going forward or as a midterm target, kind of rather a low teens in % of revenue R&D spend versus a mid-teens in the past. Yeah, we already talked a bit about the Salesforce efficiency, where we really transitioned from that more boots on the ground towards, like, let's drive really Salesforce efficiency, optimize that dollar spend per booking. That means we think we can also drive an optimized SG&A in % of revenue over time. That kind of is very much part of driving profit, bottom line growth, both in adjusted EBITDA margin % as well as in terms of adjusted EPS.
Dennis, to compare what you've described in terms of margins to others in the market is sometimes difficult because there aren't a whole lot of great direct comparisons. KEYENCE is one that often comes up, in particular in terms of the gross and operating margins there. Why would you say that your aspirations are perhaps more modest by comparison?
Right. I think there are two things, right? First of all, where are we against our own historic comparison, right? Historic average has been 28% adjusted EBITDA margin. Last year in 2024, we have been only at 17%. We first have worked to get back to where we have been historically. I think with our implied guide for Q4, we're kind of starting to see maybe a 2 in front of that number. Certainly, our next milestone would be to get to, like, a 25% number where we said, like, that's the number which we want to see as kind of the minimum even also in a down cycle, right? That means avoid that we are getting compressed on adjusted EBITDA margin. Certainly, the question then comes, like, where could you go from there?
People pointed at KEYENCE and said, like, hey, how can they achieve even higher gross and operating margins? I would say on the gross margin side, it's a simple statement. They work in different markets than we do. I mean, right, I would say our factory automation gross margins are very comparable to what they have. And they are factory automation only. They're not in warehouse automation. It means there's just a mixed effect. I think where they are very well set up is in their SG&A efficiency, right? I think they, A, first have a very good playbook, which is kind of what we are driving also a bit into our Salesforce with the Salesforce transformation. Then, B, they certainly have also a clear advantage. We are a pure-play machine vision company. They are a more diversified factory automation player.
That means with that, they can achieve higher Salesforce efficiency than we think we can do as a standalone machine vision company. That is kind of a bit where our M&A thinking comes into play, right? As part of investor day, we said, like, part of our growth algorithm is also inorganic growth. Clearly, adding additional products, maybe in adjacencies to machine vision, could be part of that. That would really then drive further Salesforce efficiency and certainly would, in the long term, enable us to go beyond the adjusted EBITDA margin target range, which we have announced so far.
Last question for you, Dennis, on the Investor Day Themes. If we pull all of this together, is it fair to say that if you achieve what you set out, that the P&L will be less cyclical and more boring than in the past? And I do want to get you on record here. Can we still expect profligate spend on Halloween and annual reports?
You know, we are having a lot of fun on Halloween and with annual reports. We definitely don't want to give up on that fun.
They're the sacred cows.
You know, as CFO, you want to say, like, there are no sacred cows. You know, I would say, like, there are just things. They are fun. You know, I said, so we will not give up on those.
OK.
On the cyclicality side, I mean, clearly, we're going for a broader base of the market. That's clearly part of the strategy. Are we there today? No, we're not, right? It's really a multi-year journey. In that regard, I would say we are definitely still a more cyclical and volatile company, both in the up and the down, as many other companies. I would expect that to change over time.
Last question for you on the P&L, and then we'll pivot to Q&A. Early 2026 Commentary you offered on your most recent earnings call, you basically said, if we look at today's PMIs, the rough math would suggest mid-single-digit sales growth next year, but that with continued discipline on the OpEx line, you could feasibly achieve, call it, 20% EPS growth. How do these comments tie back to some of the themes you laid out at investor day in terms of managing through a cycle?
Right. I mean, yes, I think my investor day, so I just concluded a prior remark where a cyclical company, so that means we have basically looked at typically like your five- to seven-year cycles, typically in three phases, maybe two years or so, kind of initial phase with moderate growth, and then two years or so with high growth, outsized growth, and then kind of two years to tail end of the cycle down to flat. Basically, if you think back, the last peak, the peak of the last cycle was in 2021. You had then kind of flat and down in 2022 and 2023. Probably 2024 felt like the end, the tail end of the prior cycle. At the moment, you could think that 2025 is the initial stage of that cycle where you see that type of moderate growth.
That is where we really said, like, in this phase, especially compared to where we are in profitability, that is where we really want to drive outsized bottom line growth. Some of the OpEx efficiency measures, which we talked about, play a key role in that. At the same time, I would really like to remind everyone we are a short cycle business. That means, A, we have limited visibility from a funnel perspective, right? In factory automation, it might be only three months. We use this PMI basically to give us a bit of a view, like, maybe how could the next six months look like? Maybe, you know, PMI can move fast.
I would say, like, I would in that regard say, like, our remarks were probably like to say, like, hey, based on funnel, based on PMI, where it is, maybe the next couple of months or a couple of quarters, we would expect moderate growth. That's a time where we can still drive very attractive adjusted EPS growth. What's coming beyond that, that's very hard to say. That's really we'll need to watch PMIs further. Certainly, you know, I would say maybe give it a positive spin in that regard. The moment where we see higher growth, the flow through to the bottom line could be even larger, right? I mean, we are typically seeing 50%-60% flow through to the bottom line of each dollar, which we grow on the top line.
In that regard, think about if you can grow mid-single digits, top line, and 20% or so adjusted EPS, that could become even more attractive if growth rates go further up.
Thank you, Dennis. I want to pause here to take any questions members of the audience may have. Just raise your hand and fire away.
If you think about the competitive environment in machine vision, are you seeing new competitors emerge beyond the existing competitive size? Are there a couple of Chinese players that continue to get more and more share? Why is it growing significantly compared to the market?
Right. See, I think if you zoom out a little bit and look first over a bit longer period, you would say that competitive landscape is almost unchanged in that sense that, like, the biggest competitor in factory automation always has been KEYENCE. The largest competitor in logistics where also automation was always SICK, a privately held German company, spelled S-I-C-K. That has almost not changed for, whatever, two decades or so. You saw maybe name changes below that. Maybe some competitors entered, some others exited, and so on. You saw things like that happening. I would say a similar theme I would almost at the moment think, like, is happening with Chinese competitors. There are clearly some Chinese competitors in factory automation rising, like a Hikvision, for example.
At the same time, other competitors like OMRON, for example, have been rather struggling in that space. That means for us, it feels more like names are changing, but not necessarily the entire landscape, as it's also clearly, it's still a very fragmented market, right, with maybe the two largest players with KEYENCE and Cognex together by far not even holding 50% of the market. That means smaller players falling out, others joining. In our mind, the competitive landscape is not really changing that much, even though you see movement.
Any follow-ups or anyone else want to jump in? Sure. We'll pivot to an end market discussion. For those in the audience, at any point, just raise a hand and we'll work in here. Dennis, starting on automotive, during earnings, you mentioned that you were nearing a bottom there. Let's break up the discussion into two pieces. First, on the U.S., when would the recent announced big investment announcements from the OEMs theoretically be actionable for Cognex? Then we'll move to Europe second.
Yeah. Yeah. Automotive has for us really been the most painful market over the last almost two years, right? In that regard, was the market contracted the most? Yeah, in the last earnings call, we said, like, hey, we think, like, that we are nearing the bottom. I hope that in one of the next calls, I can call the bottom. The interesting thing here is that we see to start to see kind of a geographic diversification in that market. That means, like, North America or America specifically is actually right now the best performing regional market within auto. Europe is the worst performing. Somewhere Asia, somewhere in the middle. Within Asia also some differences there. In that regard, I think we start to see that the America's market is coming back faster.
At the moment, I would not know yet if we could call the bottom, for example, in Europe, right? The bottom we could call on the market as a total. Especially in Europe, if you see some of the announcement, if you see how profits have compressed in the last earnings call of some of the major European car manufacturers, especially some of the German manufacturers, that still looks like a market still pretty much in distress. In that regard, I'm a bit more cautious about a Europe statement. Yeah, America's looks more and more positive.
When is the pain going to end in Europe? For those of us not as close to the market as you, what's going on there that's driving all of this?
I mean, I think, see, maybe taking a step back, right? Automotive, first of all, had the view of a large technology transition. A lot of companies invested. Now we are seeing that also that transition is not happening as fast. You have different speeds in different geographies, right? Like China leading the pack and the U.S. probably rather going away from this transition. That caused a lot of pain and capital requirements for companies and made it hard for them to make investment decisions. I would say, especially for the European manufacturers, the tariff situation in the US hit some of them pretty hard. In that regard, I think there is pretty much uncertainty.
I would think, like, as at least some of the tariff deals with the EU have now been negotiated, I would think that some of that certainty should come back. We haven't seen it yet. I wouldn't be able to give you a timeline on it, Tommy.
Moving on to consumer electronics, trends improved year to date in 2025. I think you've mentioned it was the first growth year in three years. At the same time, investors have taken note of some recent form factor innovations. Expectations might have gotten a little bit elevated in terms of what that might mean for revenue opportunity to Cognex. Just situate us between those two data points, please.
Right. I mean, first of all, change is good for Cognex, right? That means change in terms of form factors, change in terms of production locations. These are things which typically drive revenue for Cognex. If you think back about the last two peaks of the cycles in consumer electronics, one was in 2017, was very much driven by a display technology changeover. Then 2021 was COVID, really. It means different working patterns and different behaviors by end users. In that regard, certainly whenever you think about technology changes that can drive business for Cognex, also changes to supply chains moving from China to India or to Vietnam, that drives. We are seeing some of that, which is why it is the first time in many years we are now seeing growth returning in consumer electronics.
This year, it's a bit more driven by the change in the supply chains, a little bit less by form factors. As you know, it's a very different thing if you change in your entire product lineup a certain component to a newer technology versus you maybe add one more production line in terms of the magnitude, what that does to a production line refitting and build-out. In general, I think we think positively about consumer electronics for the first time for a long time. We think about 2026, there can be more form factor changes, further supply chain diversifications, component manufacturing moving into the U.S., final assembly probably still staying in Asia. Eventually, people are starting to think seriously about the time after the smartphones, right? You see the Meta glasses starting to being sold in millions, not in very small quantities.
It is not as a smartphone is selling today. It is clearly that there are companies, OpenAI announced some partnerships there. In that sense, people are starting to think, like, hey, there could be coming the next wave. In that regard, we think positive about consumer electronics. It is just from today's perspective, for us, hard to call when and how strong that will happen.
Moving on to logistics. This is your second year of solid double-digit growth, albeit the base in 2023 was rather depressed because a lot of the key players there were digesting excess capacity. Walk us through the different customer types here within logistics and give us a sense of where they are in their adoption cycles and build cycles.
Yeah. If you think about the logistics market, you could probably divide that in further subverticals, so to say, right? We have e-commerce customers, think about large online retailers, distribution centers. We certainly also have what we call more parcels. Think about more like FedEx, UPS, this type of the world. Maybe you could even add airport baggage handling. In that regard, there are different segments. We traditionally think our home turf is really e-commerce. On one side, there's for us an opportunity to go into more subverticals over time. Really, the growth has been coming and is with the e-commerce side. That's driven by a very low penetration of machine vision in today's kind of e-commerce, I would say, networks. Even sophisticated players, sorry, in the space have a low penetration rate.
We sometimes would say for sophisticated players, they're maybe at a 25% penetration. Maybe the broader market has a 15% penetration. In that regard, there's clearly a penetration opportunity. Some of this is related to what we announced in the last earnings call. We introduced a new product lineup called the SLX. Think about it, like, in the past, the logistics market was very much code reading, high speed, high precision, high accuracy. That's still a great market. There's still so much more to optimize in this distribution network. We can grow just with code reading. Customers also see other problems happening than just code reading.
That means jams on conveyor belts, side by sides, two packages side by side, and then only one code is being read, damaged packages, which you want to not only find out at the end, but maybe right when they enter your network so that you're not running through basically waste through your network. That you can't solve with code reading. You need machine vision. That's where we're basically teleporting, if you want to say it, existing technology, which we use in factory automation and bringing that dedicated to logistics. That's the SLX lineup. That basically is what gives us that positive view that logistics can be a multi-year growth driver for us. However, growth is not linear. After two very strong years of growth in logistics, now especially in the second half of this year, that growth was much more focused on large-scale customers.
We're a little bit more cautious about 2026. Therefore, while other markets are getting better, we would definitely consider that maybe growth rates would also naturally get a little bit lower in 2026. That does not mean that we are thinking differently about this market from a long term. It's just about growth is not linear.
Any questions from the audience before we wrap it up? I have just one more prepared. Jump in if you'd like to ask anything. All right. I have one to wrap it up here, Dennis. I want you to gaze deeply into the crystal ball here. Historically, Cognex has grown top line well above the market average. If you look at the source of that growth, it's evolved over time. If you go back far enough, it was semiconductors. Pre-pandemic, it was consumer electronics. More recently, it's been logistics, like we were just discussing. If you gaze into the crystal ball and think about what are industries that could, I'm talking five, 10 years out, pop up as significant sources of growth, what comes to mind there?
Right. First of all, I would think, like, in the existing verticals, there's already a great growth opportunity, right? I just talked about the low automation penetration level in logistics and talked about consumer electronics with that new form factors time beyond smartphones. There are great opportunities there. I think a market still nascent today is, and which we're otherwise not talking a lot about, is aerospace and defense. I think that's definitely a market where there are great machine vision applications. It's a small market today, but it's definitely an interesting market. We'll see where that goes over the years.
Thank you very much, Dennis, for all the insight. We appreciate your being here this week.
Thanks a lot, Tommy. Thanks, everyone.