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

Great. Hello, everyone. My name is Jason Celino, and I'm the vertical software analyst here at KeyBanc covering the EDAs. I want to welcome President and recently announced CEO, Anirudh Devgan. We also have Alan Lindstrom here, IR. And then hopefully, CFO, John Wall, will be joining soon as well. Alan, did you want to read the safe harbor?

Yes. Thank you, Jason. So before we begin today, we'll mention the safe harbor statement. Today's discussion will contain forward-looking statements and will make use of certain non-GAAP financial measures. Please see our most recent -- our most recent 10-K, 10-Q and website for a discussion of risk factors and our use of non-GAAP financial measures. So back to you, Jason.

Great. And it looks like we have John as well, so perfect. And for the audience, we do have this awesome feature at the bottom. It's a box where you can submit questions. Happy to weave those in as we get them.

Maybe first, as a general, maybe warm up, Anirudh, we heard the CEO transition announcement last week. It's been in the work for some time. What should investors and customers expect from Cadence and their, sort of, new CEO?

Well, thank you, Jason. Great to talk to all of you. I think I'm truly excited, really humbled, to be appointed CEO of Cadence. And like you said in our announcement, I think we are in the strongest position we have been in the history of the company, especially over the last few years. And I have worked closely with Lip-Bu and John over the last several years since I became President 4 years ago. And of course, I'm a big believer in the strategy, intelligent system design strategy of doing well in the core EDA and expanding into system analysis and AI and data analytics. So I think in summary, you should expect continuity and focus on keep building the momentum for growth in the strategy for increasing margin. So full speed forward. I'm really excited about where we are and where we can go, so.

Okay. Excellent. And maybe just a quick earnings recap question for John. On the call, you mentioned higher ACVs. That's a good sign for customer budgets. But it is typically happening -- but is it typically happening with the type of customer or industry? And how sustainable is this?

That's a great question, Jason. Yes. I mean we're always trying to improve our annual contract values with customers. And I think our TAM expansion strategy naturally helps improve annual contract values over time. But the strength is pretty broad-based and not specific to any customer or industry. We're highly focused on growing our relationships with customers. And I would say it can be sustainable as long as we continue to focus on things like executing our strategy, and as long as we continue to execute on technology differentiation and keep up the pace of innovation and partnering with customers and ecosystem partners. But yes, it feels like it can be sustainable.

Okay. And it probably helps with new products and maybe one of them I wanted to talk about was this Cerebrus Intelligent Chip design, Explorer. Can you maybe just give us a brief, what it is and why is it significant?

Yes, Jason, it's pretty exciting. We have been working on it for like 2, 3 years on Cerebrus, and we just announced it in July. And then even working in terms of R&D for 2, 3 years and with customers for the last 6 to 12 months, and we want to make sure it is rock solid when we announce it. And basically, what it does is that it uses AI and ML to drive the tools. So typically, like digital flow is a critical flow for our customers. So let's say, Genus, Innovus, it would run like in 1 or 2 days, typically, right? You take a block part of the chip, you run it through implementation, you get great PPA. But the chip is designed in like 12 months or 9 months or something like that, typically 6 to 12 months. Some are longer, some are shorter. So what the user or the designer is doing is they're running these tools over and over again, trying different parameters, searching the design space, floor planning options. So it's basically the tool by itself is doing a good job, but the designer is using their infusion to search the space. And traditionally, this has been a manual process going back decades, right? This is -- it's a user kind of intuition. So we still want to keep that, but we want to add this more mathematical way to do that search using ML. So these Cerebrus will sit on top of the implementation tool, and it does the full flow. And it's almost like ADAS, like advanced driver assistance to take analogy from self-driving cars. And of course, you could have Level 1 ADAS or Level 5 ADAS. So we believe that it is -- you still need the designer, it's critical. But one designer can do a lot more blocks than they used to do in the past. And the other benefit -- so that's the productivity benefit, because you're not running -- you're not running it yourself all the time. You're using machines to do it, so you can do more blocks, right? So the other benefit is that the PPA can be better. So PPA, I mean like power performance in the area. Because you're mathematically running these tools, not only you can get to point A to point B in a faster amount of time. The point B, which is the end design, can be more efficient than done manually. So some examples are 2 of our public endorsements are from Samsung and Renesas. So Renesas mentioned that the frequency of the chip was improved by 10%, that's a huge improvement in frequency. And Samsung mentioned that the power using Cerebrus on top of our tool is reduced by 8% more than before. So this 8% power reduction, 10% frequency, these are huge improvements. As you know, typically, you have to go from 1 node to another node to get this kind of improvement, but we are getting it through better algorithms. So we are excited. It's still kind of early days, and it depends on -- it depends how good a driver the customer is already. So how much improvement you can get on top of that. There is some other benefits we have seen, which is like typically the customers are global companies. So they have groups all over the world, right? So you'll have some team in Silicon Valley, maybe some in India, some in China, some -- so -- and typically, the talent of the team can vary by location or by engineer. But this Cerebrus provides some normalization to that because, again, the driver is getting assistance from these tools. So it's still early. And we say, of course, it is packaged on top of our digital flow and we see how it goes, but it's pretty exciting case in there.

Okay. Interesting. Now like the ADAS and Genus stuff it's -- I guess we'll see. Maybe staying with this digital theme, your digital segment has been growing 10% plus for the last 3 quarters. Now how much is this from maybe the easier year-over-year comps versus other strength? And then with the new products like Cerebrus, it seems like Cadence has the right products to support it, but how sustainable is this 10% plus growth in digital, specifically in the second half?

So Jason, we don't want to look at quarter-by-quarter, though we are happy with quarter-by-quarter growth. We want to look at year-by-year. And then I think in our CFO commentary, we talked about 3-year CAGRs because as our typical deal length is 2.5 years, so 3 years CAGR is a good metric for what's happening in the marketplace. So we do believe our digital is growing faster than market and has been for several years. And right now, I believe that will continue, okay? And there are maybe 3 main things to drive that. One is, of course, these lower nodes require more and more investment and tools like Cerebrus that can add to the mix. So that's advanced node. Most of our work right now is at M2 and M3 and there are a lot more foundries. And then we have these new tools like Cerebrus. So from a technology standpoint, I believe we are in a very good position. Second thing is at lower node, there is full flow. We talked about it for a while. We are doing very well in place and route, but now at lower nodes is natural coupling with synthesis and sign-off. So that also provides some momentum. And then the third thing is we want to keep doing well in customers, which are traditionally not Cadence customers. So there is still opportunities for us in some big customers to do more deployment and things like that. So overall, we feel good about the segment and the segment itself is also growing, so.

Okay. Great. Well, I guess we can move on then. Moving to maybe system analysis. Maybe can you talk about what drove Cadence into the simulation landscape? And then remind us what the TAM potential is there?

Yes. Thank you, Jason. So I'll start and then John can add on to it. So we always believe that EDA is a great business, okay? And it's becoming more and more critical as silicon design becomes more and more critical. I don't need to tell you that. But the key was that can we grow beyond that using the strengths of EDA? This is -- I think this has been EDA's issue going back like 20 years or something. And the question is, what is the right area to go beyond EDA, okay. So I believe that the right area always comes from 2 things. One is what we are really good at, what is your core competency; and where the market is growing. So if you can combine where the market is going, where the customers are going and what you're really good at, then you have the right to go into that area. So what we believe we are really good at, first let me talk about that, is what I would call computational software. So what EDA has done for decades. This is my PhD from Carnegie-Mellon is numerical analysis, okay? Is this kind of mathematical software. EDA is math plus CS, right? All kinds of numerical mathematical software, whether it's real numbers or it is like boolean analysis. So we call that computational software. So if you look at that area, we have like 6,000 people in R&D, most of them with advanced college degrees, we are best in the world at that. Okay, I want to say it in the most humble way. But in this kind of numerical mathematical software, EDA have done the most complicated algorithms for decades, okay? So that is our core strength. So -- so that -- we want to -- anything we build beyond EDA has to be built on the strength of this computational software. Now if you look at where the customer is going, that's the second angle, right? So we have the silicon as the core circle of our strategy. And then there are 2 circles around that circle. So the next circle around silicon is of course system. And then the circle beyond system is data, and AI. And the perfect example is sticking with ADAS, right? A perfect example is electric car, is a perfect example of what is happening in the marketplace. So you have -- the silicon that drives the car, and some of the car companies are doing their own silicon. And then you have the actual physical car and the design of it, which is the software and the hardware, and all the self-driving data around it, direct captures and all that. And this is happening in cell phones, it's happening in data centers, it's happening even in defense. So if you look at those 3 circles, we have 4, silicon, then system and then data and you overlay the strength in computational software. So what naturally falls out if you -- when we go from the silicon circle to the system circle, so the system market, I think, software-wise, is about $50 billion. There's all kinds of software. But the computational part is system simulation. Simulation is inherently computational. It'll always need for more simulation. And that's about $7 billion to $8 billion and growing. So that's why a few years ago, we made a significant investment in system analysis -- system design and analysis. And I think right now, we have maybe 800, 900 people in R&D in that space. So that's a big investment for us, and we're doing pretty well. We are happy with the progress. There are a lot of other benefits of that, I will talk about like margin expansion and things like that. And then the third circle is this AI and ML, which by the way, is also computational. If you look at inference as a matrix multiplier. Training is like conjugate gradient. In all these techniques we have done in EDA for a long time, like the placement algorithm in Innovus is the same algorithm used in ML training. So we can take a lot of people who are traditional EDA experts and apply to machine learning, so like things like Cerebrus we just talked about. So we are -- of course, we hire more people also, but we are confident in our strength to do these 2 areas and go from there. So this is the key part of our expansion. And I believe it is very synergistic in terms of technology. It is very synergistic in terms of what the customers are doing. So the response has been very positive. And the third part, which John loves, is that I believe that it is a higher-margin business, because EDA, we want to do better and better, so simulation is a higher-margin business than EDA. Because even within EDA, about 1/4 of our business is simulation, and that's a higher margin business. Because one user can run multiple simulations. So simulation by -- so it has customer synergy, technology synergy, revenue growth and higher margin. So actually, I'm surprised why we didn't do it before. We didn't do it sooner.

Okay. Okay. So I'll ask one more question here, and then we'll move on because I know you're quite excited about this segment. But the one thing you didn't talk about was go-to-market. So the heaviest simulation users are in aerospace, they're in automotive, they're in industrial sectors. So I guess my question is, what does Cadence do in go-to-market for these industries today? And then, I guess, how does the go-to-market need to change to realize the full opportunity?