Cadence Design Systems, Inc. (CDNS) Earnings Call Transcript & Summary

Great. Good morning, everyone. Thank you so much for joining us at the NASDAQ conference. My name is Elizabeth Porter. I'm an equity analyst on the software team over at Morgan Stanley. I'm very, very pleased to have with us today the President and CEO of Cadence Design Systems, Anirudh Devgan. So keep in mind, we are taking audience Q&A, so the mics are going to go around at the end of our session. And just before we begin, the company asked me to make you aware of their safe harbor statement. The discussion today will contain forward-looking statements and will make use of certain non-GAAP financial measures. Please make -- please see the company's most recent 10-K, 10-Q and website for a discussion of risk factors and their use of non-GAAP financial measures. From my side, for important disclosures, please see the Morgan Stanley disclosure website at www.morganstanley.com/researchdisclosures. And with that, Anirudh, thank you so much for joining us today.

So just for those that are in the audience that might not be as aware of the story, can you just give us a brief overview of the company and kind of where you fit in within the broader electronics ecosystem?

Yes. Hello, everyone. Great to be here. Just a brief overview of Cadence. So we are a software company for design of semiconductors and electronic systems. So almost all semiconductors in the world will require some form of our software to do the design and same thing, increasingly, for electronic systems. So this is all geographies, all verticals. And we are mostly ratable. 85% to 90% of our revenue is ratable with about average duration of 3 years. And this year, we are -- our revenue is about $3.5 billion, roughly. So this year, we are growing about 19%. And our margin -- operating margin is about 40%. EPS growth is about 29% versus last year. So over the last few years, our EPS growth CAGR is in the slightly over 20%. So company is in great shape. And about 40% of our customers now are system companies, so -- or close to 45%. It used to be 40%. Now close to 45% of our customers are system companies. And so we are also increasing more and more towards system design. So we have a new initiative, not just to do EDA, which is semiconductor design, but also system design. So that's one key focus for the company going forward. And second focus is using intelligence or AI to revamp our products. So that's a quick overview.

Awesome. Yes, thanks for that overview. And obviously, with 40% plus of the -- 45% of the system design companies, that has been a really kind of big focus. So can you just walk us through what are some of the big, more salient points of that Intelligent Design System technology?

Yes, absolutely. So like I said, 45% of our customers are now system companies. So there are 3 big growth drivers for us going forward. So one is semi design itself. I know there is a lot of variability in revenue right now in semi. But in terms of R&D spend, there is still very strong design activity. But the second thing is these system companies doing silicon, it started with these phone companies. But now, if you look at Tesla or you look at all these data center companies, Google, Microsoft, Amazon, you name it, they're all doing silicon design. So that's a new TAM for us. That's a new market for us. And I think we are still in the early innings of that expansion into the system company. So that's the second growth driver for Cadence. And then the third thing is we are also expanding our TAM into SD&A, system design and analysis. So right now, about 12%, roughly $400 million of our revenue is coming from system design products, not semi design products. So these are things like thermal analysis, simulation, PCB and board, package design, this big view of 3D-IC and all that. So -- and that part of our business, the company overall this year is growing, like I said, about 19%, that part is growing in the mid-20% range. So that's a pretty successful expansion of our TAM. So those are the 3 big drivers for us. One is the system -- the semi company is still doing a lot of design. Second, the system company is doing silicon. And third is our expansion into SD&A, not just EDA.

Got it. And then in addition to system design, you also mentioned AI. And I wanted to just get a better view on how AI is impacting just the capabilities of your product offering today. And not just today, but also how you expect that to change in the future. And for Cadence specifically, is AI just more about enhancing kind of your capabilities of design software? Or could you also see the company just penetrating more into your company offering kind of AI products?

Yes, it's a big topic, especially these days. And everybody calls everything AI, that's the other problem. So there's this famous professor from Berkeley, UC Berkeley, Michael Jordan, not the basketball player. But -- so he says, stop calling everything AI, and he's one of the top AI guys. So the joke I have these days is that, I don't know if you know programming, but even if you have an if statement, it's AI now, it's artificial intelligence. And you have L statement, it's reinforcement learning because you learn from the if statement. So -- but that having being said, AI can be transformational if applied properly. So what we say, what we are really good at is computational software. If you look at EDA, right, the history of EDA is computational software, which means CS plus math. So this is very numerical mathematical software over the last 40 years. And AI is also computational software, and a lot of the basics of AI are like training and inference. These are basic matrix multiply or conjugate gradient. So we have a lot of expertise that we can apply to AI from our main R&D team. Now the value of AI to us or in general, I think -- I mean there's a lot of talk of AI. But in my opinion, AI is like an outside-in science. When you say data science, that means you are observing something from the outside. So to give you a simple example, right, if you remember your basic math, you say, Y is equal to F(X). F is the function, X is the input, Y is the output. When you're doing AI, you're observing from outside in, you're observing inputs and outputs. So we like that, but you also have to have the inside-out view, which is based on the physics or the chemistry or the biology. And so that's simulation or inherent properties of the system. So if you combine both inside-out and outside-in, I think there can be a lot of value. So to give an example in the design process, so we have a tool called Cerebrus, okay, which is outside-in. And it can really improve design productivity because in the past, right, the industry is focused on like more inside-out. Like we do a good job of optimizing a design. But when the user is doing a design, they're running a lot of options and a lot of changes. So it's a very iterative process, and that can be optimized by AI. It's almost like self-driving or like a driver assistance for a chip design. So with Cerebrus, which is one of our AI tools, it's not improving the basics of like how we do placement or routing. I mean that, we have done for like 30 years, but it's more how the user interacts with the tool. And using AI, we can improve productivity by 5 to 10x. We can also improve the PPA, power performance in area, by 10% to 20%. So to give you an example, when you go from one node to another node, let's say 7-nanometer to 5-nanometer or 5-nanometer to 3-nanometer, the PPA gain typically is 15% to 20%. So you are getting that just by better AI tools. So the demand is really high in all the top customers to use AI in chip design. And same thing applies to system design because system design even has less automation than chip design in the past. So I'm very optimistic about AI, but it has to be applied in the right context, yes.

Great. And then I wanted to just switch gears a little bit for kind of your positioning within chiplet and kind of the 3D-IC segment. And will this segment kind of be enough to be a significant driver for the company?

Yes, that's another big thing, 3D-ICs, which is basically a system in a package. Typically, in the old days, as you know, you would have one chip in a package, now you can have multiple chips in a package. And they could be next to each other or they could be on top of each other. So they're all called 3D-IC. Some people call it 2.5D, some people call it 3D. So the key thing becomes is package design becomes super critical. Among -- so one thing to remember is Cadence has about 80% share in advanced packaging historically. So we have a tool called Allegro, which is the market leader in package design already. So we are in a unique position to have -- of course, we have a massive analog design platform, a digital design platform at the chip level, Allegro at the package level and all the system tools I talked about. So we are in a unique position to offer a comprehensive 3D-IC platform, and we call that Integrity. So recently, if you saw in October, TSMC, one of our big partners, announced 3D blocks. That's their 3D-IC platform. So a lot of it is based on Cadence flow. So we are, at this point, working with all the major foundries on 3D-IC, and I think it will be significant. And it will start -- it's starting anyway, mostly on high-performance computing, HPC, but it will happen to other segments. So I'm pretty optimistic. Actually, there's another research from Stanford this time, not Berkeley. I have to keep both of them happy to some extent. So that talks about that 3D-IC can extend more slower by 7 generations. So even if you discount that by half, okay, so that's 3 or 4 generations. So I think 3D-IC can provide a big boost for next 10, 20 years to the whole design process. And the thing to remember, I believe Cadence is in the lead position in that. And almost all the flows right now are using our software.

Great. And then China has obviously been a big topic and would love for you to help us just understand the mechanics and just the implications around the latest round of American export controls for semi-related products. So I guess a couple of questions in here. Like, one, how do you evaluate just the apparent increased risk for operating in China? And then second, does China still present a good business opportunity for American-based companies?

Yes. That's a great question. It's a very topical top subject. So I mean, in a nutshell, if you look at it last 5 years, our China business is growing like 20%-plus CAGR, which is higher than our -- Cadence is growing around 15% last few years CAGR-wise. I mean, this year, we are higher. And China is slightly more than 20%. So I expect China to be a good business long term. Okay. And there's a lot of recent activity. There are 2 kinds of activities or 2 kind of questions we get. One is some local development in China. There are some companies trying to write EDA tools. And second is U.S. regulation. So I mean the U.S. regulation is a long topic. But in summary, the impact to Cadence is pretty limited. I mean those regulations are targeted mostly towards manufacturing in China at this point. And even if you go to China, all the big companies are not -- they're doing 3-nanometer design, they are not manufacturing in China. They are manufacturing in TSMC or Samsung. So we are more connected to the design activity. And of course, China has its design activity all the way from 3-nanometer to mainstream node. So we have some effect, but it's very, very limited effect to our business. And then the other question I get is, will the U.S. government do more, okay? Is this only the beginning? So normally, we are in discussions with them because nobody can predict what the U.S. government will do, and I don't want to do that. But typically, we are -- our discussions with them are focused on what they just announced, and I think there will be a time of digestion of that. So we are not expecting any new regulations at this point because the ones that were just announced we knew for about 1 or 2 years. So now the second part is, will there be local competition in China? Because at this point, like I said, we are in a very strong position. So these things take a long time. So -- and there are some very small companies doing point tools. But to really do a full flow at 3-nanometer with analog, digital packaging, I think this -- so we are -- we, of course, watch it very carefully, but I'm not concerned that this is a short to midterm issue with local competition in China. So if you back that out, I think the effect is pretty limited. And there is still a lot of design activity in China. So overall, I'm pretty optimistic about the China market, yes.

Good. And then kind of more near term is obviously macro, lots of debate going into recession, kind of what it looks like. And we've already seen a number of semiconductor companies cut estimates just as the business environment slows, even from some of the hyperscalers who do some of their own chip design, we've seen layoffs from them as well. So the question is, one, historically, EDA has been more resilient in downturns. So kind of first question is help us understand like why that is. And then second, in the case of another downturn, would you expect something similar or anything different to play out within EDA?

Yes, that's a great question. Thank you. So overall, the reason we are more resilient -- of course, we are not immune. So if there is a massive downturn, then nobody is immune, right? Nobody in this room is immune. But if there is a -- and I don't know if there will be a massive recession. I mean you guys are experts at that. I think if there's a slight downturn, I think we should be pretty resilient in that. And the reason why we are resilient is 3 main reasons. So first is we are ratable. So like I said, 90% of the revenue is ratable. We have average of 3-year deals. So when we go into any new year, most of our revenue is already in backlog. It's coming from backlog. And you can look at our backlog numbers is growing throughout this year, okay? So overall, our backlog is pretty healthy. So that's number one. Number two is that we are very diversified. So almost all regions -- and we talked about China. China is still one market, right? So almost all regions in the world, almost all market segments, they use -- they need to use our software. So I know some parts of the market may be strong, some may be weak. But overall, it's like an index fund for semiconductors and electronic systems. So we are very, very diversified. Okay. And then the third thing, which is probably the most important reason, is that we are mission-critical. So we are -- our spending is coming from R&D spend. So this is engineering teams, right, in semiconductors or system companies. So I know there is some fluctuation in revenue, of course, in semi companies and system companies may have some layoffs. But typically, if there is a mild recession, it doesn't affect R&D spending that much. And design activity is still very strong in all these companies because they are designing for when they will come out of the downturn. So I think the line you have to watch is not the revenue of R&D companies, but the R&D spend of these companies. And I talk to all the big CEOs all the time. So, so far, still, we see a lot of design activity. So for all these 3 reasons, we are pretty resilient. Now of course, like I said, nobody is immune and nobody can fully predict the future. But at this point, we feel pretty strong about how things are going, yes.

Okay. Then I'll ask another question, and then we'll open it up for the audience if you have any questions. So you talked about system analytics kind of moving into that market. You highlighted products like Clarity and Celsius. And so why does expansion into this area actually make sense for Cadence? And what are you bringing to the table that is so different than an incremental to what some of the current suppliers are already doing? And then if you can kind of expand even more beyond that, but just understanding the recent moves in the computational fluid dynamics and kind of digital design, and now you've been doing things in more of the molecular simulation.

Yes, yes. The key thing is everybody wants to expand, right? Why not grow? But the question is, where do you grow, right? All companies want to grow. And a lot of them -- and I was part of like big companies and small companies. A lot of time, they sit around the table, see what company to buy or something like that. Okay, that's not how we do it. I think to grow, first, you have to figure out what is your expertise, right, what is your core strength and grow in the direction of your core strength because there's always competition in any area you want to enter. So what is Cadence's core strength? So like I mentioned before, what I believe our core strength is, is computational software, the CS plus math. So we are best in the world for this compared to all the hyperscalers or any other software company. When it comes to this CS plus math, this numerical kind of software, EDA has a long rich history of that, and I get into a lot of details of why that is. Okay, so that's the main strength we have. And then if you look at the world, we look at it in terms of 3 main spheres, okay? So the innermost sphere or circle is silicon, right? So when you -- and then the next one is systems and the next one is data. So the perfect example is like electric car. You have all the navigation data, you have the actual physical car, which is hardware, software, mechanical, electrical, and then the silicon that drives the car. So that's what's happening in all verticals. So when you overlay that with computational software, so if you apply computational software to silicon because that's EDA and IP, that's our core business. If you apply computational software to system, that's simulation. So system software is about a $50 billion market. Simulation is about $8 billion to $10 billion, all this thermal simulation, CFD, things like that. So that almost doubles our TAM. EDA traditionally is around $10 billion, so it doubles our TAM if you go into system analysis. Okay. And then if you apply computational software to data, that's AI, okay? So that's the 3 kind of core business plus systems plus AI. And the reason we are successful in that, and as you can see our growth rate is very high in systems, is because the math we can apply, the algorithms we can apply are much more complicated. Because silicon design, Moore's law, to give you example, when you simulate inside a chip, like some 5-nanometer chip and we do some design we will simulate, there are billions and billions of variables, right, because these things have billions of transistors. So we'll simulate 50 billion variable things on the chip. And already 1/3 -- roughly 1/3 of our EDA business is simulation. Now when you apply that to the system level, you do a thermal simulation of a phone, they would consider 40 million variables to be a big system, but we can do like 50 billion variables. So we can apply a lot of expertise from the IC level and go to the system space, which is almost impossible for the system guys, the traditional system simulation players, to come to the EDA space, and they have tried for all these years. So that's why I'm very optimistic that there is a lot of R&D synergy. Okay, so that's number one. Number two, there is customer synergy, right? 45% of the customers are system companies, so there is customer synergy. And number three, the margin profile, even though EDA is high margin, we are 40% operating margin, the system simulation margin profile is even better than EDA. So financially, there is a lot of reasons to do that. Because you want to expand, you want to grow revenue, but you also want to grow margin, so you don't want to enter areas which are lower margin than your business. So for these 3 reasons, EDA plus SDA, to me, is a marriage made in heaven. Now I'm surprised nobody did it before, but that's good for me and us at this point. This combination of chip design and system design is going to happen, is happening and I think we are leading that combination.

Yes. It definitely sounds like just a natural adjacency for you guys to move into. I wanted to turn it over to audience Q&A. We have a mic going around.

How do you price a tool like Cerebrus? So if I'm a design team, we're using Cerebrus or not using Cerebrus, how much more is it to use Cerebrus if I'm getting 5 to 10x productivity and 20% more PPA? And then besides Cerebrus, what other tools are you most excited about for using AI in the -- infusing AI in the tool?

That's a great question. And there's natural like price discovery in something like Cerebrus because the way it is priced, it is priced well, okay, let me say that. So the way it is -- for the -- so the other thing with these AI-based tools is that when Cerebrus runs on top of Innovus, which is our implementation tool, it will run like 10 copies of Innovus. So these are naturally more parallel because you are exploring the design space. So the way we price it, you have to buy Cerebrus and you have to buy like 10 copies of Innovus. So it is much more expensive to use Cerebrus than just the old way of doing that. So -- because we think there's a lot of value there, like you said. So there is some price discovery that will happen. And so we are happy with that to happen in the next few years. Now the question is -- there's a lot of interest in the top 20 companies. Almost all of them are engaged with Cerebrus. And some of them are using it a lot. Some of them are only starting to use it. And the price may come down a little bit as the volume goes up, as a percentage of design moves from Innovus to Cerebrus plus Innovus, but it still should provide a lot of growth, I think, to the company. So it's well priced and deservingly so. Now in terms of other parts of AI, like I said, AI is like this outside-in science. So the other big area is verification. So if you know about the chip design process, it's never complete in terms of verification. All these -- you can never fully verify the chip because of exponential problem. So we launched a new platform called Verisium, which is AI applied to verification with support from like Samsung, ST, several other, Renesas, several other companies. So I'm very optimistic about the use of AI and verification, not just the design process.

We have time for one more.

On AI, again, do you see this as a -- the chips used in AI, do you see this as a very fragmented market, so you will need a different chip for each application of AI? Or could there be a more standard processor?

In terms of -- okay, there are 2 parts. AI is -- like I said, there are a lot of things. There's one is like these chips that accelerate AI, like NVIDIA and this and that. That's -- and they are -- all of our -- all those are our customers and they use our software. I think that market has different dynamics. Okay, I think we'll see how much of the new platforms really emerge for hardware assistance of AI. But that's -- when I talk about AI, I'm not talking about that. Of course, we help our customers to design different AI chips. What I'm talking about is AI applied to our own products. So that will be pretty universal because we are, anyway, one of the leading platforms for chip design. And the way I look at it is -- and talking to a lot of customers. So if you look at our customers, right, about 10% of the -- like I said earlier, we are part of R&D spend, right? So -- but 10% of the R&D spend is going to automation, 90% is going to people. Roughly that's the high-level breakdown. So what AI can let you do with these things like Cerebrus and Verisium, because there is more automation, anyway, they are having difficulty finding all the engineers. And over time, there are not enough engineers in the world to design all these things. So I expect, over time, bigger percentage of our R&D spend will go to automation. So that 10% should go up, and we should be deflationary. EDA by nature, automation is deflationary. But with AI, because if you can do 10x better or 10% to 20% better PPA, and that's what we are seeing, that more of the spend should go to AI or automation in general. Now that's very different than the chips used to accelerate AI. I mean that's a whole different market. And we are glad to participate. Almost all those companies use our products to design the chips. But I'm more optimistic AI on the software side of how to make it more and more productive to design these things.

Great. I think that takes us to our time. Anirudh, thank you so much for joining us today.

Thank you. Yes.