Cognex: Vision Quest - [Business Breakdowns, REPLAY]

0:00This episode is brought to you by Portrait. It's the AI research system that I used to prepare for today's episode and for all Business Breakdowns episodes. Portrait was built by former buy-side investors, and they understand great investing isn't just about having more information from low-quality sources. It's about having the right information organized the right way. And if you listen to the show, you appreciate diligence consists of many things. Diving into the history of a business, framing the nuanced competitive dynamics, tracking key signposts around your thesis. And historically, that would take up material time that you do not have.

0:41But Portrait is basically like adding an army of analysts to your team. It's powered by an AI system specifically designed for investment research workflows. So you get nuanced idea generation. Portrait assesses the same types of qualitative attributes that we discuss on this show, and that can help identify businesses which fit your frameworks. Portrait also customizes research report generation. And I use Portrait to generate a primer and lay out bold bear cases ahead of today's episode to help frame the conversation. And third, there's intelligent thesis monitoring. And that's where Portrait assesses thousands of data points across value chains each day, extracting the insights, driving the business.

1:32Again, all this work would typically take hours and hours and hours. It's at your fingertips now. Visit PortraitResearch.com to start your free trial today. This is Matt Russell. We have another replay this week before we get back to regularly scheduled programming. And this one is on Cognex. And I remember doing this episode with Brett Larson and leaving thinking to myself, this is the AI and robotics story that I want to monitor. Because Cognex has such an interesting position with machine vision, all of the algorithmic tools associated with machine learning. Since this episode's been released, the company has talked in several different events about some of the progress that they've seen using AI to expand their customer base, introduce new use cases, particularly with physical AI.

2:27So it's only one piece of the broader Cognex story. Obviously, there's much more that goes into this business with all of the different end markets and who their natural customer base is. But wanted to replay this one given where we are in the market, particularly in 2026, and so much focus on applying AI into the robotic space. I hope you enjoy.

2:53This is Business Breakdowns. Business Breakdowns is a series of conversations with investors and operators diving deep into a single business. For each business, we explore its history, its business model, its competitive advantages, and what makes it tick. We believe every business has lessons and secrets that investors and operators can learn from, and we are here to bring them to you. To find more episodes of Breakdowns, check out Colossus.com. All opinions expressed by hosts and podcast guests are solely their own opinions. Hosts, podcast guests, their employers or affiliates may maintain positions in the securities discussed in this podcast.

3:35This podcast is for informational purposes only and should not be relied upon as a basis for investment decisions. Today, we are breaking down machine vision leader Cognex. Our guest today is Brett Larson from NZS Capital, and Brett returns to business breakdowns. We covered trained technologies last year, and Brett gives us a different angle on the industrial economy with Cognex today. We get into the history of machine vision and how Cognex is a factory player. But I particularly enjoyed the strategic angles of this discussion. Cognex is not your typical reoccurring revenue story. They are a self-proclaimed cyclical that has tended to focus on a specific customer segment over time, looking for S-curves that might trigger their next growth era.

4:24This is evolving. And we get into a lot of that with Brett in this discussion. I'd mentioned the big name businesses tend to attract the widest audience. but these are my favorite episodes. So please enjoy this business breakdown on Cognex. All right, Brett, it is great to have you back. It was about a year ago that we covered Train, maybe a little bit more than a year ago, but it's an episode that I referenced quite a bit over the past few months and one that I consistently go back to. I'm hoping that we get the same thing today with Cognex. It was a name that I wasn't familiar with, And it's a very interesting name that seems to not get a ton of attention or conversation going as I was doing my research.

5:09Easiest place to start is just with an introduction to who Cognex is and what they do. Thanks for having me, Matt. It was fun. So I was excited to be able to join you again. Cognex stands for Cognition Experts, and they are leaders in machine vision. What that means specifically is they sell ruggedized cameras with embedded processing and then software, which is the real value add, which captures images and analyzes them in order to automate decisions at high speeds and manufacturing and logistics environments. So the product might look like two cell phones stacked together with the Cognex yellow that sits right out there on the factory floor.

5:47And the applications generally fall into four categories. So there's guide, gauge, inspect, and ID. ID is the Data Man family of products. And that's using vision for reading barcodes and for optical character recognition, which is reading letters and numbers that are printed or etched on something. So the easiest example to picture for somebody would be an Amazon sorting facility that's moving tens or hundreds of thousands of packages per day at extremely high speeds. and Cognex's vision products, scan all those barcodes and then perform the automation then performs the actions that are required.

6:23Insight is the other main product category and that covers guide, gauge and inspect. Inspect would be quality inspection. So 2D and 3D vision systems that are used to inspect something. A common example would be a printed circuit board. The system would make sure that it has the correct number, size and placement of the different components and quality soldering. Or another classic example would be making sure bottles in a bottling machine have the lids on and labels. Guide would be vision systems, which are used to guide robotic arms. And then gauge would be measuring the dimensions of something.

6:56And ultimately, you can think of the value prop that they're trying to deliver as improving quality, throughput, reducing costs and waste. And then a more and more common one is just addressing labor constraints. And the payback period is usually quite quick for these types of products. You mentioned there's the hardware, which I can imagine. you have these cameras on a conveyor belt, the software aspect to them, they have to link into whatever you are working on. In some ways, that software is machine to machine execution. Is there a lot that's also going back to humans or human interaction with the software?

7:30Ideally, the only time the humans really interacting with the software primarily would be upfront when you're programming the application. And then after that, the camera's really sitting out on their factory floor, capturing the data, analyzing it. And then that data will be communicated to more like the programmable logic controller that Rockwell might sell, which is then performing whatever action needs to be done. Ideally, once it's out on the factory floor, the human's kind of out of the loop other than overseeing it. I think you brought up a great visual in terms of an Amazon sorting facility and how much of that is now automated.

8:04You don't have the humans involved moving the big packages here, the smaller packages there. So it's gotten more advanced. But how big of a market is this machine vision world today? And I'd also just be curious how much that has grown recently, if there's been a step change in terms of the growth of the market. There's a lot of different TAM estimates out there for machine vision, but Cognex doesn't play in all the different niches within it that get captured usually. Cognex last estimated their serviceable addressable market at 6.5 billion, which was up from 2.9 billion in 2017. And they'll probably update their SAM again at their investor day.

8:42And it'll probably be something like 8 to 9 billion based on new products categories they've entered. Industry has probably grown at a 10 % type of CAGR the past decade. And obviously there's cycles in there. And then there's dispersion between some that are growing above that range, like Cognex and some that have been below. I like the Sam terminology. I might have to mix that in with Tam in the future. Where does Cognex rank in terms of market share within the market? What are some of the characteristics about their positioning and then just the market overall? Cognex is number two behind Keyence, which is a company based in Japan.

9:18The two companies do obviously compete, but they've also historically focused in slightly different areas, which is interesting. So Cognex, they historically have focused on the top of the pyramid. When you think about the sophistication of the customer, typically they're hiring very trained engineers who are working with customers to spec systems for specific tasks. So it's a more technical sale with very sophisticated customers that are automating very complex tasks, basically. They're not winning on price. They're a more expensive vendor of machine vision, but they have a reputation for having really good application engineers, the best tech that they can differentiate at the spec level.

9:57So for example, that logistics facility where let's say you're scanning 100 ,000 packages a day, Cognex, usually they can deliver read rates that are 100 basis points or even 200 or 300 basis points better than a peer. So over even just a day, that's thousands of packages that don't have to have a human there to look at something. That's usually how they go about competing. And there's a really strong brand reputation as well. So like if a new COO of a company walks into a factory for the first time and sees the yellow Cognex cameras, that says something to them. The installed base is very sticky as well for all the vendors.

10:31And I'll foreshadow a bit, but they're focused on the top of the pyramid. Currently, they're also broadening out a little bit lower as well, which is interesting. And then Keyence, as I mentioned, is number one. They're based in Japan. It's been a really successful company. And it's also very much an enigma, just in terms of it's a public company, but there's just not a lot of intel out there on it. My best guess, what I've gathered is maybe 20 % or so of their sales are comparable to Cognex. They also do things like scientific microscopes and PLCs and direct part marketing and stuff like that.

11:04But what's also quite interesting is they spend 2 % of sales or low single digits on R &D and have mid 80s gross margins. And that compares to Cognex, which spends more like mid-teens on R &D and has more like 70 % gross margins. And the other machine vendor peers spend a little bit less and have even a little bit lower gross margins than Cognex. And it really comes down to the way Keyence goes to market and how they focus. They generally focus just on the middle to lower tiers of customers. And then they're really trying to develop more standardized products that are going after the very high frequency applications.

11:38And it's a very process-oriented sale. So they'll hire more college graduates compared to Cognex. And then they'll train them with the products, put them out in the field. And it's more of a scripted sales process where they're keeping track of more of those activity like KPIs and the CRM. So it's how many calls are you making? How many shop visits? How many demos? It's really good coverage and really relentless. If you get on the rabbit hole, it's become a meme in the community. There's some pretty funny memes out there. If you're going to download a product spec sheet up the website, you should use your buddy's email and phone number, not your own.

12:12You'll never hear the end of it. That's the two main players. Maybe I'll briefly mention a couple other competitors while we're here. The other bucket I would say is China. And these are primarily for manufacturers in China. Hike Robotics, which is a division of Hike Vision, is probably the third largest player in the industry overall. And they've grown quite quickly, driven by that domestic market with the domestic manufacturers. And they're probably about half the size of Cognex in terms of sales. There's a players as well. And then outside of there, there's more legacy players. There's with IntelliDyne, there's Dalsa and Point Grey.

12:48There's SICK and Basler in Germany, DataLogic. Matrox was a company acquired by Zebra. And there's even companies like Inditech that just sell the software and then you can go get your own hardware. So that's kind of a lay of the land in terms of the industry structure. I did read quite a bit about Keyence and their sales process and how effective of a sales organization they built. But in the sale for someone like Cognex, where it sounds like it's much more technical, built to spec, perhaps advanced and not so much off the shelf. Are they selling to an Amazon who then works to integrate it into the other hardware and equipment that they're using?

13:23Or are they selling into the equipment manufacturers, whoever builds the conveyor belt, who is the customer for Cognex? The answer is yes. If you think about how they're going to market, 70 % of sales are direct and that would be direct either to the factory floor where Cognex and the automation engineers at the customer are working to implement the system. But direct also includes selling to a machine builder or an OEM who would then integrate it into the machine. And then they take that machine to the end factory floor. And then the remaining 30%, about half of that is going to be through systems integrators.

13:58And that's primarily logistics at this point. And then the other half of that 30 % would be distribution. And that's primarily just for in markets like Cambodia or something where they just don't have any presence. That's kind of how I would separate the customer. That makes sense. It's an interesting dynamic within the overall industry structure conversation. Before we move on too far, I do want to get into a bit of the history and what came across through the work that I was doing is that there is technical expertise here and there's a focus on that technical expertise. So can you bring us back to the beginning and the origin story here when they got into the market and some of the evolution over time?

14:37It's really important to understand for Cognex the DNA of them stacking S-curves essentially over 40 to 50 years. The company was founded in 1981, and they came to market with what was the world's first industrial optical character recognition system. So again, reading numbers and letters. And at the time, the system looked more like a big video camera with external processing. And the first application was actually to read serial numbers on semiconductor wafers for IBM. That was the DataMan product family that still exists. From there, they got a call from Johnson & Johnson to do optical character recognition for labels.

15:13They did that application, but then J &J asked them, hey, can you do some of these other, at the time, what would have been novel applications, like verifying the caps around the bottles and the labels were present and the like. So that's when Cognix added more of these inspection type of applications. The first 20 years or so of the business was pioneering machine vision for all these different applications, but it was very much for semiconductors and the electronics capital equipment industries primarily. It was as high as 80 % of sales going into the dot-com bubble were to those industries.

15:45And that went to 54 % following the hard way. Around 2000 was when we saw smart cameras really come onto the scene. So where you embedded the processing directly into the camera, smaller footprint, more ruggedized. Now that enabled the product to be ready for the factory floor as we more often think about it. So we saw machine vision get adopted in more heavy manufacturing industries like automotive, consumer electronics, food and beverage, packaging. So by the end of 2010, that semiconductor and electronics capital equipment was down to 15 % of sales as they really brought in the end markets that they could serve all those new S-curves.

16:23Around 2010 was when Cognex came out with their first ID product for reading barcodes. It was primarily aimed at displacing laser-based scanners in logistics facilities because I could do it faster and then with much higher read rates. If you go back in 2010, they were talking about they'd be happy if one day long-term target$75 million of sales from this line of business. And then they worked very closely with Amazon to develop the technology. And in 2021, the peak, it was about$300 million in sales and 30 % of Cognex's overall business. So it turned out to be just an enormous new S-curve for them.

17:02At present, we're really in the early innings of the next tech evolution for the industry. And it began in 2017 when Cognex acquired Viti. And then in 2019, they acquired a company called SuaLab. And those were both essentially IP and acquihires in the field of deep learning or the application of AI into machine vision. Viti's co-founder and CTO is Cognex's current VP of AI technology. So after those acquisitions, there was a big development effort and they came to market with these new deep learning and edge learning products. And the way I conceptualize this is moving from rules-based programming of vision systems to teaching by example, which has large implications.

17:43And I think we're probably in the first inning of this next chapter behind the scenes of a long cyclical down cycle. Could you paint that picture in terms of an example? Rules-based, let's say, a package is over 100 pounds with these dimensions, send it in the left lane. If it's under that, send it through the middle. If it's super small, send it to the right. And it's a pretty standard tree of logic and decisioning. Where would this new example or this new training come into play? Do you have any real world applications that would paint the picture? There's deep learning and edge learning, as I mentioned.

18:20Deep learning first. So traditional rules-based vision is great for a lot of things, but not for things very subtle or nuanced tasks with a great deal of variation, those sorts of tasks, you still usually have a human performing them because frankly, it's just hard to quantify them to then program them. Maybe an example of that would be inspecting a phone case for very subtle defects. There's obviously a wide variety of colors and there's just an endless list of potential defects. So scratches, dents, blemishes in the paint, and they're very small, very nuanced. And then there's also usually a scale of what's acceptable and what's not, where a human would be able to look at it and pretty quickly discern that.

19:00But it's very hard to program a machine to be able to do that, especially at obviously high speeds. But with deep learning, instead of programming it, you teach it with very large data sets and very large image sets. These are good phone images. These are defects. These are acceptable defects. And the machine learns what is acceptable and what is not and is able to accomplish that task. Just to quantify this a little bit. So there's currently still 30 million people in the world doing visual inspection. And that's something that humans actually aren't great at. They can do it quickly, but they get fatigued and they miss things.

19:36That's a big opportunity. Another example would be something like deboning chickens, where every chicken's different. It comes down the line in different orientation. The wings are somewhere different. It'd be impossible to quantify and program a robot to be able to grab the right places. But with Deep learning, that's an application that they can now do. Deep learning, the implication would be potentially new applications for machine vision. Edge learning is the other end of the spectrum. And it's probably more financially tangible for the investor community right now. These products, they come pre-programmed for specific applications, but then they're trained with just as few as five to 10 images.

20:14So it's essentially very easy to deploy relative to traditional rules-based vision. you can usually get an application up and running in a few hours. So the major implication here is that Cognex can now sell to less sophisticated customers. Again, thinking about where they've historically focused on the top of the pyramid. And they can do it with a much less technical sales force. So broadening their customer base below where they've typically served. So now they have the product and Cognex is currently making a sales force investment called the Emerging Customer Initiative. And it's really taking the key in its playbook.

20:48These emerging customer salesnoids, they call their employees cognoids, their salespeople salesnoids. They're hiring young salespeople instead of technical applications and engineers. They're arming them with these new edge learning products that are really easy to implement. And they're going after these less sophisticated customers, more SMB types, and even less sophisticated tasks within existing customers where those exist. And then it's a much more formalized selling playbook that tracks the KPIs, just like Keyence. they're getting compensated on both their selling activity as well as their commission.

21:20Last year, cohort one of these sales noids hit the field and they did 80 ,000 customer visits, added 3 ,000 new customers to a base of what was previously 30 ,000 at accretive gross margins. And they exited the year at about a million dollars a week in sales from that first cohort. The second cohort entered this year in 2025, and hopefully we'll do a bit better just from lessons learned. Obviously, they're now competing in Keyence's bread and butter target customer set. So they are running into them sometimes with these customers being vended. But the majority of the time, these customers have never used machine vision at all.

21:59I think it's more like 60 % to 70 % of their sales have been people that have never had a camera on their factory. Obviously, the financial implications here are they're trying to grow their customer base from 30 ,000 to hundreds of thousands over time, which could be obviously revenue opportunity and then at accretive gross margins and reduce the customer concentration as well over time. The customer base, I think you painted a picture just in terms of the various applications and who might use them, but what does it look like in terms of a customer, whether it's a contract, whether it's a sale, the ramping of that contractor sale, and then what the stickiness is once you build something out?

22:38The stickiness of a customer is quite high. Once you're on the factory floor and the people that are managing the floor are familiar with the software used to program the cameras. And it's usually more or less standardized, especially for some customers where they, for the sake of simplification, they don't want a bunch of different machine vision vendors out there and the associated software. It's a CapEx sale. And they have, as I mentioned, historically been focused on the top of the pyramid customers. So they've been tied to the large CapEx buildouts of companies like Apple and Amazon. And that obviously comes in a wave.

23:11And then it's more of a question of stacking new S-curves and when the next waves in CapEx is going to come. But the customers themselves are generally quite sticky. Maybe this is a good segue into the different end markets. And we can start with consumer electronics. It's interesting because that end market's tied to the CapEx spending around new consumer electronics or features. There was a tailwind from the adoption of cell phones. And then there was the tailwind from Apple building out the iPhone manufacturing. At one point, Apple was as high as 20 % of Cognex's sales, which is pretty incredible.

23:43It's now probably down to more mid to high single digits range. There was another big tailwind when OLED came on the scene and that was with Samsung. So looking forward, it's interesting to think about if there will be new form factors or changes to the existing ones for consumer electrotronics tied to things like LLMs, or if there's AR or VR is going to take off ever, or if humanoid robots are going to take off. Basically, if any of those things were to be manufactured at high volumes, that would fall right into Cognex's sweet spot. Sophisticated customers, big manufacturing capability. In consumer electronics in 2024, 17 % of sales.

24:21And longer term, they expect that market to grow mid-teens is the target. Logistics is the largest end market. It's 23 % of sales. At peak, it was 30 % of sales and Amazon was up to 17 % of Cognex's sales. Obviously, there was the well-publicized down cycle in that end market. They did return to growth in 2024. They grew 20%. And that's the long-term target for that end market is that they can grow 20%. And Amazon probably is back around that high single digits to 10 % range in terms of a customer. In Cognex fashion, they're stacking new S-curves in logistics, which is really interesting. So they're taking the vision tunnels, which they've really honed with Amazon to new customers and new geographies.