Cognex Corporation (CGNX) Earnings Call Transcript & Summary

Good morning. My name is Brendan Luecke, and I'm the multi-industrials analyst at Bernstein. We're joined today by Rob Willett, CEO of Cognex, a role he's held for over a decade, overseeing explosive growth and enviable to many of your peers, I'm sure. Rob, thanks so much for joining us today at SDC. I understand you have some opening remarks.

I do. Great. Okay. Just a few kind of remarks to get grounded. Thank you, Brendan. Thanks for inviting us here today. Great. I always enjoy this conference. So let's start by talking a little bit about Cognex overall, just to get grounded, right? Obligatory statement about -- I'll make some statements that may have forward-looking implications, and that's basically true based on what I know today, but look at our SEC disclosures for risks. The founder of Cognex used to just like to say, don't trust anything I say, basically that's what it says, but we've evolved a little since then. All right. So here's to get grounded in the numbers of Cognex. So we're about $1 billion business overall to give you some sense of that when I joined the company, our first year, 15 years ago, we were $175 million. We've seen organic growth, really organic through that period. We have $10 billion of cumulative revenue overall. And profitable, we're primarily a software business. Gross margins are normally in the mid 70s, they've been a little bit lower in the last 12 months due to broker buys that we've been making on the chips that have been short in supply. So what do we do? We're a growth and technology company. And everything we do is focused on this field known as machine vision. And machine vision technology is the technology that makes computers see, right? And I think the analogy of how we see as human beings is kind of useful, right? We have an eye and we have a brain. We have an eye that captures information and it sends it to the brain that makes sense of it. In machine vision, there's a camera, which has an imager and lights and lenses. It captures images and then it focuses those onto pixels and the value of -- light hits those pixels, it creates value, which is sent then to a processor, which is like the brain and Cognex makes machine vision tools or algorithms that run on that processor and make sense, make sense of those images, right? And that's -- that's what a machine vision is. It's a form of artificial intelligence. And we take that technology and we -- it's everything we do is machine vision, and we apply it to factory automation, right? And there are basically 4 things that we do. We guide, we would use vision to guide a robot to put a windshield on a car or a chip on a board. We inspect. We look at, are all pixels on an imager, are all the chips on a board, right? We gauge, which means we measure. We might look at a brake pad on a car brake -- is the groove in the brake pad that correct width and depth. And we identify, which means we read letters or numbers or barcodes on products moving down a production line. Machine vision is a difficult technology. It might be analogous to voice recognition in some respects. It's very, very big and confusing data sets that can change very much based on things like the light in an image or how products are moving or color or distortion, right? And it's the thing that a lot of PhD students learn in a lab, but to apply it in the real world to factory automation takes huge amounts of know-how that we've developed over our 42 years of focusing on this market. It's a great market, machine vision, and it's a market that has huge growth potential and has had huge growth. One way of measuring the potential for machine vision is just to look at humans in manufacturing today. There are about 360 million people go to work every day in manufacturing. And about 10% of them or 35 million really are purely using their eyes to inspect product. Is that -- are there defects in the products? Is -- things correctly situated, well-lid, attached. And it's a very difficult job to do. Humans are trying to concentrate for 8 hours at a time to do that. It's not something they do very efficiently overall. And machine vision is getting to the point where it's outperforming humans in many of these functions. Particularly it is the result of new technology called deep learning, which is coming to the field of machine vision, which we might talk about more. We're the technology leader -- I think pretty much undisputed technology leader in the field of industrial machine vision. We're very committed to R&D. We came out of MIT, and we've focused a lot on being the technology leader. We spend about 14% of R&D -- of revenue on R&D. We have lots of know-how. And it's a complex field, but in areas of optics, in terms of software, in terms of applications, expertise, where we have a huge domain of knowledge that we apply in order to help our customers solve some of the most challenging problems. In September, we came out with this updated version of our served market, and I stress this is our served market, not our addressable market. If we could win every dollar of business that our products can perform in, we would be a $6.5 billion business. As it is, we have about a 15% share. You can see the main markets that we serve on this slide, automotive, electronics and logistics. And you can see how we think those markets will grow over the long term, with logistics being the biggest and fastest-growing market that we're in today, electronics and automotive being others. And then there are many other industries that we serve that are smaller and newer to adopting machine vision as it becomes easier to use and more powerful. If I talk about some of those markets specifically, automotive, obviously going through a transition to EV batteries, a big growth area for Cognex as it's replacing internal combustion engines. Electronics, we do a lot of business with smartphone manufacturers, but also components in the electronics industry and waves of technology and investment have helped drive the growth of that business significantly over our life. And then logistics has been a huge growth driver for Cognex. About 10 years ago, we had almost no revenue in logistics. And then we saw 5 years of successive 50% CAGR growth in that market through to 2021, where we were about $300 million of our business was in logistics. We've seen a downturn in that in the last 18 months as the big e-commerce spend and the build-out around COVID ended up having more capacity than the industry needed. But in the long run, we see huge growth potential furthermore in logistics. Automation is relatively unpenetrated in logistics and online shopping is relatively unpenetrated. I think it's something like 15% of transactions still in America are going online less in the rest of the world. There's very limited automation still in warehouses. And we do primarily barcode reading in that market today where we can read a barcode moving through a distribution center as many as -- well, 1 to 3 points -- percentage points better than our competitors. So if a big e-commerce player is shipping 1 million units a day out of a distribution warehouse that's between 10,000 and 30,000 products today that has to be reprocessed or go late to a customer, gives you an idea of kind of the power of our technology and how it's applied. This is our revenue history. You can see that we're a growth company, obviously. Our growth isn't always linear. We've had various events that have happened that have really driven step changes in revenue, and they tend to occur on a relatively regular basis. You can see 2014, we saw the beginning of smartphones. 2017, we saw the OLED screen changes in some of the major changes to smartphones that occurred during that time. And in 2021, we saw really e-commerce really kicking in heavily during that time. Okay. Last thing I'd like to say is about Cognex is we're a company which has a very strong culture. We invest and we focus very much on our culture as a business. We've been very successful over our life in attracting really phenomenal engineers, and we create a place where they love what they work on. Machine vision is a very challenging and interesting problem that people love making their career in. And then we create a culture where it's enjoyable to work. So we describe our culture as work hard, and we have a lot of very smart people who certainly are doing their best work at Cognex, but we think it ought to be a lot of fun, right? So we invest a lot in playing hard. If you see our annual report, it will be a manifestation -- you can see a picture of Halloween at Cognex, whereas the senior team dressed up as Willy Wonka, at the Chocolate Factory. It's pretty legendary. You come and visit us as I hope you will. You can join an ultimate frisbee game every lunch time at 12:30. The final thing I'll say is move fast is the third part. And what does that really mean? It's like we really are very successful by hiring very great -- very smart people and empowering them to take risks and do great work, right? And they love that. And I view one of my jobs is to sweep the way of things that are going to stop them doing that. One example would be we have something at Cognex called C3, Cognoids Cutting Crap is what it stands for. And we invite Cognoids to say, what can we stop doing, right? Other companies have Kaizen. We have crap cutting that goes on. Engineers don't want to go through a 3-day Kaizen program. They just wanted us to get out of the way. So over the years, they've submitted ideas. We award $1,000 to $3,000 for ideas we implement. Rob, I write that report for you, every Friday, I don't think you read it anymore." You're absolutely right, stop writing it. We hate the reply all button on e-mail. Everyone's hitting replay all, it's going everywhere. We eliminated it. There's no reply all e-mail anymore at Cognex. We used to count inventory every month, why were we doing it? We count it every 3 months now. Stuff like that really kind of is getting out of the way of moving fast and getting stuff done. So we're a technology company with a commitment to R&D in a difficult field that has a long way still to run. We're in exciting areas in terms of growth and automation. We're a software company with high margins and a lot of growth potential with a great culture. Okay.

Excellent. Thank you so much.

Thank you.

With that overview, why don't we dive into Q&A. Just so folks know, we are accepting audience questions on pigeonhole, and we'll try and work those into the conversation as we go here. So I'd love to start by looking back. We try to take a long view at these conferences. To put it mildly, we've seen a challenging macro environment over the past couple of years. How has this shifted your long-term thinking about the business? And what have your key learnings been?

Well, yes, I think we -- as often happens, there's disruption that occurs and changes that happen in industries with technology. And obviously, a big thing that happened over in the 2021 time frame was with COVID every -- or so much shopping went online, right? And we certainly saw a huge adoption of automation, really that I think accelerated that adoption. It proved out many of the things that we knew were going to happen. I think it just accelerated them. And what we saw was really the technology leaders in that space who tend to be Cognex customers, we tend to have very tight engineering relationships with big innovators, the big tech companies. We saw them pile in and other retailers kind of stepped back. So that was one phenomenon we saw. I think another -- there are a lot of macro trends, I think, that I've seen occur over the last few years that have played out more excessively perhaps than I thought they were going to -- labor shortages. Obviously, we will see that, the cost of labor increasing hugely. And I think that's validated a lot of things that we know are going to happen as companies look to implement more automation to improve productivity or improve scalability in light of shortages of labor. And difficult environments where I'd say, in the United States, more unionization or more concern about workplace injuries or other things that are going to drive growth overall. I think we've seen other trends, obviously, that we might talk about. One is kind of supply chain changes, desire to be more robust to more reshore production, right? So I think we've -- we're seeing that coming through. And then environmental concerns play out. Certainly, we see EV battery manufacturing as a very strong and long-term big growth driver for Cognex. The difficulty of manufacturing EV batteries, their dangerous nature in terms of their ability to catch fire and they need to inspect them the huge investment in automation to scale up what we're going to need in terms of batteries over the long term. So those are a few things that come to mind.

Fantastic. So I guess turning to the business. I'd love to chat a little bit on the emerging customer strategy. So how does this tie into product strategy? And how should investors be thinking about the incremental economics as you invest here?

Yes. So to take a step back, so I think of our markets sometimes like a pyramid, and Cognex has very strong share at the top of the pyramid, the most sophisticated customers with the most engineering acumen, love to work with our engineers on their automation plans, sometimes 3 to 5 years out. And that's -- Cognex, the name stands for cognition experts for kind of those people that those customers love to work with. And that's served us very well. Our technology traditionally, machine vision, has been very complex technology and hard to apply. But some things are changing, right? One thing that's changing is we've been leading in the development of technology known as deep learning. We can talk more about that. Deep learning is complex and powerful, but it's -- and then we've developed a version of it now, which we call edge learning, which is a pretrained model that we can deploy into an embedded camera that's very easy to apply. So when we think -- I don't know how deep the audience's knowledge is. But generally, deep learning tends to train on huge data sets, right? Think of server farms dissipating in a massive huge Nvidia chips or whatever pounding away at complex problems. The large language models we're all starting to learn about. Edge learning takes pretrained models, and it puts them into embedded hardware and it can train on as few as 10 samples of good and bad, right? So this has really changed our technology to the point where now it's going to be -- it is easier to sell, very easy to sell. And as a result, we no longer need to focus on those very sophisticated customers, but we can start to move down the pyramid. We think of our -- we do today serve about 30,000 customers but there is 200,000 customers or so that exists more in the base of the pyramid. It's been uneconomical for us to serve them due to the sophistication of our technology. No longer the case. So what are we doing? We're building a sales force now to go and sell to those customers further down the pyramid. And we're recruiting people, and we're training them to sell products that they can demonstrate and sell in a 20-minute session in front of customers. And there are a lot of customers and target customers in that part of the pyramid who, some of whom have barely used machine vision at all or have very basic needs in that area. Some of whom are already exposed to machine vision through our competitors. So it's an opportunity for us to move down and share our superior technology with those customers. You asked about what it's going to do economic-wise. Generally, those customers can be very profitable at a gross margin level. These embedded products we sell to them, there's a lot of technology and a lot of R&D went into the building materials is relatively small, right? And so I think it can be gross margin accretive for us. And then we're going to spend money over many years, I think to build out the sales channel. We've pegged that this year at about $25 million to $30 million. We'll be investing to get a sales force trained and up and running in that space. We'll probably exit at about $10 million a quarter incremental from where we were, and I would expect us to go on building that out. To give you some context too, about 70% of Cognex' sales are direct today, up from 50% 10 years ago, right? And we would see more and more of that being directors, the scale and the ease of our products means we can reach more of those smaller customers.

Excellent. That's very useful. And as you sort of approach that long tail, go from 30,000 to 200,000 eventually, I suppose, maybe. Is there price pressure? I mean, I would assume that simpler use cases would translate to lower price points?

Yes. I think we think they're going to be higher margin because they're purchasing small quantities. I've been on some sales calls with some of our initial sales pilots in this area. And I find a lot of the customers, they have certain needs. I visited one, a pretzel manufacturing company recently and sort of customer we would never have focused on recently -- until recently when we had this edge learning technology. They wanted to see are the flaps on their box closed. And they were looking at, do I buy a $9,000 taping machine, a new tape machine to make sure the box is taped. They weren't sure that was going to fix the problem or do I buy a $3,000 machine vision system that can look at the entire box and inspect it for various things, but are the flaps closed correctly. 20-minute demo, a short trial, a purchase, right? No negotiation over price. So a different kind of dynamic.

Interesting. And I mean, just by extension, I mean, sometimes when it's a long tail approach, I worry about lower price point products, diluting or competing with the tip of the pyramid. It sounds like the use cases in your 30,000 base today are significantly differentiated.

Yes. Although I think an average sale of Cognex today might be more on the range of, say, $8,000 but it may take longer to close. These will be smaller, maybe on the range of 1 to 2 units at $2,000 to $3,000 each, but they'll close much more quickly, right? So I think -- and then the gross margin will be higher, right, on those. But we're going to invest in the sales channel to build it out. So I think operating margin, it will be diluted this year, obviously. And as we start to scale and get the sales force up, I think it should be operating margin accretive in the long run.

Fantastic. Very good. So I'd love to touch on the cycle briefly. It's a very confusing cycle. I would imagine that Cognex has a pretty good line of sight in your global business. Reshoring is really the topic because you are right now in U.S. industrials. How do you see this showing up in your business today? And can you tell us, are the investments you're seeing broader than sort of the stimulus-driven Ships Act, IRA, EVs? Or is it perhaps an industry-specific motion?

I think there's a lot going on, right, okay? I think certainly, the Inflation Reduction Act is causing a lot of our customers, particularly in environmental businesses like EV batteries, particularly solar, to rethink their investment plans and to bring them back onshore. And it's kind of stopped some investment while we reconfigured their plans. I think of a big customer of ours who was going to build out something in Germany and decided to build it out in Texas, right? Another one in Korea, moving it to Atlanta, right? That's slowed down their investment plans, but in the long run, will be good. We have very strong share in the United States. So I think it's good for us in the end, more production coming onshore. Other factors would be definitely, we see among our customers more concern about supply chain robustness, and I would call it near shoring, I think, is the term that I hear used a lot around in automotive, and I see that driving a lot of investment in Mexico, not so much in the United States, but in Mexico coming back from other markets aboard. And then the third big dynamic we have is we do a lot of business with big smartphone manufacturers, right? And certainly, they've been concerned for a while but are concerned much more now on their dependence on China, right, particularly. So you see -- we see action where they're looking to move production not to the United States really but more to India, Vietnam, Thailand, Philippines, but particularly those first 2, India and Vietnam.

Got it. Yes. And then if we think about some of the key verticals for Cognex, where do you think we are in the cycle? I mean, logistics feels like maybe we're approaching a trial. Automotive, probably still growing. Consumer electronics, not so sure. What are your thoughts?

Well, first thing is we tend to think long term, right? We're thinking about long-term drivers. We have long-term product road map discussions with customers. And on that score, I'm very optimistic. There will be -- there's a lot of new technology coming in electronics. There's a lot of transition to EV happening in automotive that I think will drive big growth a little bit further out here in the medium term. And then logistics, there's a lot of growth to be had with e-commerce being built out on a more global basis and not just with the technology leaders, but everybody else. But that said, I'm not optimistic about the short term. I think the PMI is a good measure of kind of current sentiment among our broad customer base, and that looks pretty universally weak everywhere. And how that manifests itself is in customers delaying orders, right, and postponing plans and being -- having extra rounds of reviews on projects that are going on. And so I think we're going to see a pickup in business when logistics absorbs is excess capacity, which could be sometime next year. I don't think it will be this year when we see some of those big EV battery investments kind of break loose. And that's coming. You can definitely see that coming. We said we expect electronics to be slightly down this year. I think companies with COVID built out a lot of capacity that they're going to absorb. It's hard to know the cycle of introduction of new technology. We see with our customers new technology they want to add, and some of it makes it into the product or not as we move through. We can -- so we can sometimes find we underestimate or overestimate the cycle there a little bit directionally. But I think we saw a big growth year last year. We'll see less next year. It can be on a tick-tock kind of cadence. So we'll see what happens. But generally, I'm not that optimistic about this year. I'm very optimistic about the following years in terms of the cycles kicking in.

Excellent. Thank you. And I mean the flip side of the cycle, of course, we're always looking for recurring revenues, be it break fix, replacement, upgrade cycles that might take the edge off some of those down years. Is there a potential for that? How do you see that in Cognex? Or is this really sort of a capital projects-driven business at the end of the day?

I mean I think automation is a capital business overall. So kind of with that caveat, I think if you look at other companies like Rockwell or those that play in the space, that's generally how the industry has been formed, Mitsubishi, et cetera. We're -- for Cognex, we can see big growth where there's big incremental plans, technologies coming in like flat panel displays, smartphones, et cetera, sort of waves of technology coming through. And then we can see in periods where spending is tight and capital is tight, more of a focus on retrofits, productivity gains, focused on short-term payback type projects, which in the end, when we're in that cycle, it's not big growth for us, but this kind of business that we can pull back on to help our customers and maintain their plans overall. We have our own plans to develop more recurring revenue, right? It's particularly I think something important to understand is we have a huge installed base of machine vision systems that operate. And generally, they're producing just huge amounts of data. Just think we're involved in making about 2 billion products a day. Each of them is probably 3 megapixels of data. It's more data than one could ever possibly consume in the current environment. But we're getting more successful and through better processors and better data transmission with getting that data off machine vision systems and then into sort of middleware that we provide, we have a product called Edge Intelligence. And what it allows us to do is remotely tap into our vision systems and help our customers maintain them and optimize them, train their data sets, right, and help improve their production and upgrade their systems. And we're starting for some of our bigger customers to offer that on a license basis. So it is monthly recurring licenses. But I don't want to -- it's a long journey we're on, but I think it's one we see value for more stable and recurring revenue.

It certainly feels like a sustainable advantage.

Yes. Yes. It certainly is. And I think when we talk about emerging customers too, obviously, there, we're kind of -- we're having a much broader customer base, which is less susceptible to volatility as we've had with, say, some of the huge tech leaders that have traditionally been our customers.

Excellent. So I guess coming off your data comment, I mean, how do you think about in industrial, there's cost, there is spread of commoditization -- cross automation, actually, and businesses cycled through it. How do you think about the threat of commoditization? With 70%-plus gross margins, Cognex has a pretty prevalent position.

Well, like we're students of the innovators to [indiscernible] at Cognex. We're always thinking about that, how will our products be? How will there be new entrants to the market? Who will offer a more commoditized solution at a lower price point? We see some of that more coming out of China at the moment, specifically. So we're very mindful of that. Our technology in terms of software and performances, according to those who study the industry, 3 to 5 years ahead of our competitors based on what we spend and our technology and our intellectual property. So we tend to sell older versions of our products at lower price points in those in those markets. So we watch that very carefully. And then yes, and then I think the history of Cognex is we invent kind of the cutting-edge technology, it kind of moves up the kind of S curve, and as it starts to mature, those are products that we move away from and back to the sort of innovative growth, higher gross margin products. And that's kind of linked with our culture. We really like a high-value stuff that we do. To give you -- I've been at the company 15 years now, so have a little historical context. But when I joined the company, people said, "You'll never sustain those gross margins, Rob, I don't know why you went there. It's sort of like in the end, everything is a toaster. And you'll never sustain those 67% growth margins," right? So we made it up to 77%. We've dealt with some chip shortages. Guidance is mid-70s for the second quarter. So certainly, we've been able to do it now, partly by innovating, partly by being paranoid and partly by making sure we're on the cutting edge, and we're focused on technology so...

Excellent.

That's how we think of it.

And I've actually got a couple of inbounds from the audience. You're mostly in competitive dynamics. So with 15% share in your served markets, why do customers choose Cognex? And what levers do you have to increase that?

Yes. So they choose Cognex because we have the best technology, right? We have a fantastic reputation for quality and customer support, right, overall. And we have a reputation of working with the most sophisticated customers and the best engineers in the world in manufacturing are working at those customers. And it's a high-octane environment, a lot of them leave and go to other customers and they take Cognex and our processes with them. So that is sort of how we've tended to grow overall serving those customers and those needs. I think -- what else would I say about that? I think how we grow overall is we maintain share in that area, and we make sure those customers are well provided for, and we understand their road maps and our road maps, and we marry them together. But then we also make sure that we're selling our products to more and more customers who now can absorb them because they're easier to use and easier to sell. I mean we're a technology company, right? So our products are getting less expensive, easier to use, more powerful and we're riding that curve, and we need to make sure we just continue to do that.

Excellent. Right place in the S curve.

Yes, yes.

Fantastic. Let's shift to product. So AI, I mean, obviously, you've been investing here through your edge offerings. How do you think about AI? Is this a threat? Is it an opportunity? Is it potential that it could allow competitors to catch up more quickly or maybe provide you with a deeper moat?

Right. Well, first of all, like what is AI, right? Like I see companies that I don't think of as AI companies saying we're an AI company. That reminds me of 1999 where everyone was an Internet company, right? But I think we have a good claim saying that Cognex is at its heart an AI company. In the 1980s, there was a textbook that a lot of Cognoids study that MIT and it was called artificial intelligence. And there was a chapter on machine vision, right? It's a core of how computers have become intelligent, right? So I think -- at the core of what we do, we're an AI company. We're benefiting from the GPUs and the treadmill that the whole world is on in terms of more capacity and capability for that. I think when I hear most people today talk about AI, I think what they're talking about is data science, right? And it's how we're applying data in big data models to make our lives easier, right? So of course, Cognex will benefit from many of those things to be more efficient on how we operate. But I also think, for us, certainly how our customer experience is developed and how they interface with our technology will benefit from AI in the long run. So I think machine vision can be difficult to use and difficult to train. A lot of -- it has been quite programming-intensive, right? But our newer products and particularly our edge learning products are not at all program-intensive. They're doing a lot of that work in the background, they're training, they're kind of wizards, if you like, that allow us to configure our technology to make it easy to use and easy to sell. So that's a big way in which I think data sets and training are going to allow our products to be more widely adopted.

Excellent. And as you mentioned, I mean, it seems like an evolution, maybe more than a revolution, you had a textbook in the '80s.