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

Hi, everyone. Welcome to our next presentation. I'm Jason Celino, the analyst covering vertical software in Cadence. I want to welcome our Cadence CFO, John Wall. John?

Jason, thanks for having us at the conference. Before we begin, we need to mention the safe harbor statements. Today's discussion will contain forward-looking statements and we'll make use of certain non-GAAP financial measures. Please see our most recent 10-K, 10-Q and website for a discussion of risk factors and our use of non-GAAP financial measures. And if I may, I'd like to introduce you to our President, Anirudh Devgan.

Great. Thanks. What a surprise to have you, Anirudh. And welcome. For everyone joining the webcast, we do want to keep this rather interactive. [Operator Instructions]

With that, I think. I'll start with you, John. Since -- a little bit about your comments in the past about your 50% incremental margin. What are some examples or process improvements, technology you used to drive greater efficiency in some of those cost savings?

Yes. Jason, it's just a very disciplined approach. And I mean, since -- back in 2015, 2016, we took a very detailed data-driven analysis of our pricing with our customers to work out average selling price by product, by product groups for each customer and basically came up with our own algorithm for no matter what the configuration was, that you'll kind of get your manufacturers' recommended retail price, if you like. But -- and then we measure deal quality based on what percentage of the annual value that we'd expect for that configuration are we getting on each account. And that's helped us improve pricing on the -- so we're more aggressive in improving pricing on accounts where the deal quality metric is lower. But -- and then we play more defense where the deal quality metric is higher. But -- and that approach, that disciplined approach, has served us well. I mean since we started that, kind of you've seen 2017, 2018, 2019 and, again, this year, you see we're consistently driving high single digits to low double-digit revenue growth across all our portfolio of businesses. And then the -- and that's throwing off -- is profitable and it's sustainable revenue growth because you're seeing incremental margins coming through at about 50% -- and slightly higher than 50%. And that's part of the model. That's what we target. We target profitable revenue growth. Generally, in the plan that we do, we target double-digit growth. And then I reserve the right to pull back from some deals where the pricing or the profitability doesn't look right. And that kind of -- maybe I'm dragging us back into high single digits from double digits. Good, it's because it's better for investors if it's more profitability and more sustainable profitability that's sustainable.

It's interesting that you've kind of mentioned your pricing discipline. We have a company later presenting, PROS Holdings. They do pricing optimization. Do you use some sort of pricing software? Or is it more of an internal focus that you're doing?

It's very internal. That's -- we've done all the analysis internally ourselves. We've hired some -- I mean we've got some great engineers on the staff at Cadence. But in the early days, we started -- we were using spreadsheets in Excel, that's typically how you start out. Then we were using Access and we broke Access because it was too big. And then I think we got up to something like 300 million lines of data, but just trying to do our analysis, we had -- so we developed our own -- we hired our own engineers to do the data analysis for us.

Okay. And you note that...

Yes, I'd be interested to meet that company.

Okay. Maybe we can introduce you later.

Please do.

Last question on margins, and we'll move on to the good stuff. 50% incremental margins, if you keep this up, what would prevent the company from doing 35% or 40% operating margin someday?

Right. Jason, I mean the whole point of the model is that when you look at -- if we can consistently drive 50% incremental margins and you can consistently drive kind of high single-digit to low double-digit revenue growth, well clearly, there's a path to higher and higher profitability. That's why we always say that there's no near-term ceiling. Now we only guide traditionally for the current year. But like you say, that -- I think in terms of pricing discipline that we've established, I think we're still in very early innings there. But in some cases, pricing might have been -- like, let's say, if we were pricing contracts in some cases with customers at 50% to 60% of the average selling price, it's tough to fix that in 1 contract cycle. And our customers have 2- to 3-year contract cycles. Even if you even if you make a big inroad in improving that over the next contract cycle, over 3-year period, by the time you get to the renewal, that average selling prices have improved across the board, and it's a constant battle to just try and keep up with average selling prices. So I do think it's a decade-long journey. But in terms of before we get to the level of pricing discipline that I would like us to have and more consistency of pricing across the board. But -- so I do think it's -- like I say, it's early innings from a price improvement standpoint. And from that perspective, that's certainly what we're targeting. Targeting high single digits to low double-digit revenue growth and 50% incremental margin.

Okay. Okay. Great. Well, I'd like to ask some strategy, product questions to Anirudh. It's becoming more clear that the progression to the next process now is getting harder and harder. Does this mean that EDA tools have to get incrementally better and better? And then where are we seeing this innovation coming for Cadence?

All right, Jason. That's a good question, Jason. So definitely, the nodes are getting harder and harder. So to give you example, I think our latest -- currently, our -- most of our R&D efforts are on like 3- and 2-nanometer. Most of our customers are at 7- and 5-nanometer. But 3-nanometer was considerably harder than 5-nanometer, okay? And I think one thing that you will see as these things get announced by our partners, it's the foundries that will announce their process nodes. That 3-nanometer has a lot more EDA innovation in it because of the pure scaling of transistors actually is slowing down. So 3-nanometer especially has a lot of EDA-specific optimizations to get performance. And so I do believe as you go lower in the nodes, the importance of -- I think EDA has always been important but the importance of EDA goes up because you need more design methodology and design tools and IP improvements to get the PPA and performance benefits. And so of course, the question is how long does it continue. But right now, we are active at 3 and 2 and see how well it goes, so...

Okay. And maybe if we can talk about simulation a little bit because I know you guys have been very active there. Some notable EDA blogs have talked about the better performance for some of Cadence solvers. What is the difference in architecture, deployment or what makes it faster and more accurate?

That's a good question. So first thing, before I go to the newest products like system analysis, Clarity, I just want to emphasize one thing. So we -- like this year, we gave guidance of roughly $2.6 billion in revenue. Now we don't break up by individual products, but a good portion of that is already simulation business, okay? This is like circuit simulation with Spectre or IR drop with Voltus or logic simulation with Xcelium. So what I'm trying to say is that Cadence already has a lot of expertise in simulation over the years. Now this simulation has been applied to more IC kind of problems, not on the system problems. So what we're trying to do now is extend our expertise and simulation to the system level, which is with Clarity and Xcelium. So the question naturally is, okay, what will make us successful there because they are entrenched players in that space. So one thing you have to remember is because we are applying to the IC level, the number of -- the complexity is much, much higher, okay? So if you have to simulate -- so we have solvers that can simulate like 40 billion, 50 billion node matrices, okay? These are huge, billion with a b. And when you go to the system level, 30 million, 40 million is considered a big matrix or a big system. So we are taking our algorithms, which are optimized for much, much larger problems because these are 3-nanometer, 7-nanometer kind of chips, so they have a lot of elements. So you're taking the same algorithms and applying it to the system level, and so now the question is -- so about speed and accuracy, okay? So I was in CMU in the early '90s. And I was doing circuit simulation. And there were a bunch of other people doing like electromagnetic simulation, which is finite element. And if you look at the finite element products, they are from that generation. The initial companies are early '90s. But the math of finite element is from 1940s. So nothing has really changed even though we claim there's a lot of new stuff, but there isn't. So what we do is when we apply finite element in Clarity, we don't make any approximation to the math of -- or math or physics or finite element, which is from 1940s. So therefore, it is completely accurate. What we change is how we solve the system. So that's more numerical computation and speed of numerical computation based on our expertise. So Clarity by its very nature is fully accurate because we're not changing any meshing or approximations. And just we solve it in a much more efficient way. And that's why we can do bigger things and much faster.

Okay. Well, for the sake of discussion, let's agree that Cadence solvers like Clarity are more accurate and faster. How do you compete against the competition, which is considered the gold standard and can bundle several physics? Is there any benefit to using Cadence EDA flow and Cadence simulation tools versus some of these other competitors?

Yes, that's a good question. So there is -- so first of all, I think -- I firmly believe that we are, of course, in the technology business and the best product always wins. And we have seen this over and over again, sometimes it takes a little longer sometimes takes less longer. Now in terms of simulation, my experience in the past, I used to do -- I mean, I've done several of these simulation products, that simulation, typically is faster to replace than implementation products. This is the typical history of -- because it's a kind of a batch tool, right? You run simulation, you can compare it, put an output. So this is the traditional history, and we'll see how well it holds in this new market. But traditionally, simulation is less difficult to replace than implementation. And that's why we are not -- see, we have implementation platforms like Virtuoso and Innovus. And when we go to the system space, we naturally targeted simulation things. Because, one, that is our expertise. But also implementation part -- platforms at the system level, let's say, CATIA or Autodesk, I think, are much more difficult to replace, this is my opinion in general. The second thing is, if you look at our customers, designing these systems or cars, there is a natural synergy between the IC and the system level. So the customers are asking us not only to evaluate the chip, but where it sits in, the thermal properties and all that. So the customers are also asking us to combine our EDA solutions with the system simulation solutions. So that's the second point. And the third point I want to highlight is that we also had some implementation platforms. Remember, I was saying implementation is more sticky. So implementation platform we have is Allegro in PCB, right, which is a chip -- which is a package and board solution. So there's natural synergy between Allegro and Clarity. And then we recently acquired AWR, which is another implementation platform for RF. So that's also a natural synergy. So to answer your question, there are 3 reasons. One, I believe analysis traditionally has been less difficult to replace than implementation. So that's why we do analysis. Second, there is natural customer synergy and channel synergy. And third, we do have some implementation, Allegro, especially at the system-level that helps us with our customers.

Okay. That's a good background. Maybe kind of leading my next question. We spend a lot of time comparing against the gold standard in simulation, which is ANSYS. But we forget that there are other smaller simulation players that could be lower hanging fruit. Who would those be? And why not focus on that segment?

No, absolutely. I mean, that's a good point. No, we focus on our customers and what they want. And the good thing about this space, the simulation space, is that it is kind of well distributed. Now there are some big players, like you mentioned, and there are some other -- but there are a lot of small players, and there's a lot of niche markets and niche applications. And so we wanted to make sure we have a good -- we always start with the core engine, okay? We want to make sure we have the best core engine because we are in this for the long haul. So we want to have fundamental advantages. Then you can always customize it for different verticals and domains, and we will look at that. So I think we just follow what our customers want from us. We have a good team. We have a great product. And then the good thing is this market is not as concentrated as the chip market. In the EDA market, it has become much more concentrated. I think the system simulation market is still wide open and also growing faster. So we completely agree with you on that.

Yes. Okay. And I've got questions, I can go all day [Operator Instructions] One more on simulation. Most of the customer wins that you've announced have been on the semiconductor side or the electronics side, how much of the product would need to improve? Or what needs to change in the go-to-market to address some of the other core simulation workload, solvers in the automotive and aerospace customers use?

Okay. That's a good one. So I think we started with finite element, right? Finite element is a core engine. If you look at simulation, there is another core engine, which is fluid dynamic, CFD, so leave that aside. But finite element has a lot of applications. Naturally, we start with electromechanics because that's where our customer and channel synergy is, okay? And also a lot of the big customers are naturally electronics customers. So even if you look at simulation in general, there is some big customers like in any market, and then there is a long tail. The big ones are, by nature, these household names, which are these big worldwide semiconductor electronic system companies. So we naturally start with that because that's where the money is and that's where our channel synergy is. Now we are looking at expanding into the tail and other domains, especially looking at things like a cloud model or a distributor model because it is -- inherently in EDA, we always had a more of a direct channel. But in the system simulation, there is a good use of indirect channel, okay? So what we are looking at is with our direct channel to focus on the electronics and the big kind of household names in terms of electronics. And then use cloud and the distributed channel to go after the rest of the market. So that's starting now.

Okay. On the earnings call, you guys have mentioned 100 or something opportunities that you're pursuing or something like that. How much of the pipeline consists of these customers in automotive and aerospace or even some of the hyperscalers?

Well, hyperscalers, we mentioned, right, in the last earnings call because hyperscaler are naturally we have our channel synergy with our rest of our chip business. I don't know the exact breakdown, but I think right now it's still mostly electronics, like you mentioned. But we're starting into the other 2 big verticals that you have highlighted. So I think we are ready in terms -- hopefully, we are getting ready in terms of the core product. We just have to make sure that these channel partners, and we've targeted those other markets as well.

Okay. And I don't know if this next question would be for you or for John. But because of COVID, we've seen several hyperscale customers really accelerate some of their data center investments and some of their initiatives. Could this drive an air pocket in demand for EDA in 2021 or 2022 just from these companies digesting some of the investments that they've made?

Well, I'll let -- John can comment also. I think the good thing that we see about EDA and IP is our market is expanding. Of course, one, we're trying to expand beyond EDA and IP. But in our core business of EDA and IP, the market is expanding to more of these, let's say, system or system semi companies rather than traditional semi companies. And these system companies are, of course, have different scale. So -- and I think it will continue for some time. I mean we'll see, I don't think it's a 1- or 2-year thing because this thing, the migration to the cloud and the big hyperscalers, I think this has a good runway to go. And I think these companies are trying to do more and more of their designs in-house, as you have seen, then talk about it. You're probably well familiar with these companies. So this is great for the industry at large, I mean not just Cadence that these companies are getting into this. And if you look at even from a foundry utilization and you can look at these big foundries, a lot of their business is coming from nontraditional semi companies, which, of course, helps Cadence as well.

Okay. And staying on this COVID impact topic a little bit. From an execution standpoint, has there been anything difficult for you in closing -- because across software, we've heard of close rates being more difficult, large deals being harder to sign. I mean have you seen anything like this in your business? And how have you kind of adjusted to it? Because it seems like Cadence is, at least from the results perspective, executing better than some of other software companies?

Yes, I can take that one. I mean, we're blessed with the recurring revenue model and a very mature customer base. So many of our contracts are renewals and many of the larger contracts that will be renewals with some of the biggest technology companies in the world. So -- but -- and those are relationships that are already firmly established where we partner with them over a long time. So no real impact in -- on the sales channel side. But maybe it's a bit slower in terms of getting new customers, but that seems to be working quite well as well through Cadence channel partners that. But yes, I think that the Cadence team has adapted really quickly to the challenges of doing business in a pandemic environment. And I think everybody's adapted really well to working from home. And that appears to be the new normal for at least the foreseeable future. And we're trying to help others to adapt to that as well. I think if you go on to the Cadence website, you'll see that our learning and support portal, we're providing a huge amount of free training to anybody that's at home, that has an interest in Cadence tools or Cadence products that -- and wants to learn more. You can get access to a whole bunch of free trainings on our website. But -- and again, that's just helping people to be more efficient in terms of the use of our tools as well. But -- and I think that all plays into, like I say, it's a -- we obviously take a very, very long-term view. And we're basically generating and cultivating the engineers of the future here, trying to take the opportunity to do that. Yes, so I think we fit well in this environment. We operate well in this environment. There are some start-up customers that are having a challenging time for liquidity, and we're trying to help many of those out. But in the main, that's more than offset by increased investment activity from some of our larger customers.

Okay. And we do have some good questions coming into the queue. I think maybe we can hit some of them. One really quick one. This is a China question. How big of a risk is it that you -- U.S.-China trade war in particular, considering certain customers, how big of a risk is this? And what would this change to your business if the U.S. government said you couldn't sell into China?

Yes. So I don't really want to get into all these hypotheticals. I mean, we're a U.S.-based company. So we will continue to comply with all government regulations, while at the same time, supporting all of our global customers as best we can. But Asia Pacific and China is a very strong growth region for us. And China is committed to growing out its semiconductor ecosystem. And we're doing all we can to support customers, including all those customers in China, while still ensuring we comply with U.S. government export regulations. But I should point out that -- I mean, 55% to 60% of our revenue comes from our top 40 customers. But -- and we have a very diversified global customer base with no heavy concentration in any 1 customer. But while China continues to invest heavily in the semi industry, there's many exciting start-ups and opportunities out there in China. But -- and we're benefiting from that right now in terms of you see the growth rates that I think China was about 13% of our revenue in Q1. I think it was 12% of our revenue in Q2. I think it would be premature to try and put a figure on any potential impact for all hypothetical situations. What we choose to do in terms of providing guidance is, we let people know that, look, if restrictions remain in place, exactly as they are today for the remainder of the year, that's where our guidance is based on. And if you think it gets more restricted than that, well, then that's a headwind. If it's less restrictive, then that's a tailwind. But -- and we let people draw their own kind of judgments and conclusions.

Okay. And maybe one quick question, changing subjects a little bit. If we think about the intensity and usage of EDA tools and IP, whether a customer chooses to do manufacturing in-house or outsourced, is there any difference in the way that they'll use their EDA tools or IP?

Well, as said -- I mean, normally, the -- it shouldn't be that much different, okay, because depending on what kind of product it is, if it's a similar product like a digital product. Now one thing that does help is if they're using an external foundry, we will normally work with the foundry. So for example, our big partners, these big foundries, and we optimize our solutions. For example, TSMC, we have talked. We work very closely with TSMC and other foundries. So to the extent that a customer typically goes from internal fab to external fab, they will buy more IP because it's more readily available. And also the tools are more optimized. So typically, when they go to something like external foundry, we do better because we have already optimized our tools working with other big customers. So we can provide a good solution to that customer. So that's what we have seen.

Okay. And would you say that, that applies to just the broader EDA market and that maybe it benefits kind of the ecosystem?

I think it definitely benefits Cadence because we are well optimized with some of these big foundries that we have worked for years on. Now the big foundries will work with all the EDA players. But I think it depends on how well -- how good the PPA is. And so I think with the Cadence solution, especially these big foundries, our PPA is very good. So typically, if there is a competitive situation, we do well if they're using an external foundry.

Okay. And maybe we'll change subjects a little bit because we have a pretty interesting question here in the queue. You talked about TAM expansion before. Can you break down how much of your future growth is coming from TAM expansion versus core segment growth?

John, you want to take that?

Okay. So in terms of the TAM expansion strategy, it's basically, our current TAM is probably about $10 billion, that's in terms of core EDA and IP. And then our plan is to get that to $30 billion by -- I think, within 5 years or by 2025. But the thought process there is that we'll go through, this is our intelligent system design strategy, system analysis and our initial foray in to that part of the market. That's probably a big, that may be $5 billion to $6 billion of TAM right there. Our early efforts maybe address $600 million to $700 million of that TAM, but -- in terms of Clarity and Celsius. But eventually, we'll build out a broader portfolio there and be able to address. That's why we've taken a 5-year view. By the end of the 5 years, we'd expect to have a much more broad portfolio to address more of that TAM. But -- and then in the intelligence part, the AI part of our strategy, there's probably another $1 billion or $2 billion of TAM expansion opportunity there today. And we think over the course of -- so maybe that, what, $17 billion, $18 billion today, we think that will grow to about $30 billion by 2025, those markets. And generally, we'd expect system analysis to grow faster than core EDA and IP and we'd expect the AI portion to grow faster than system analysis. And I think you can basically work it out from there.

Okay. And maybe one final question. And we think we have one left, but one question that came through e-mail. With the industry pushing towards advanced packaging design and chiplets, what does that mean for simulation and system analysis intensity in semis?

That's a fabulous question. Sure. We had had our Cadence Live a couple of weeks ago. And so I do want to talk -- we talked a lot about that. So what people call, let's call it, chiplets or 3D-IC. I mean is 2.5D-IC, but we can round it up with the 3D-IC. So there are these chips on the interposer. I think it's going to be a huge trend. So in the past, what happened is people would use these things to merge 2 technologies like RF and regular CMOS. So more than more, this is probably a few years ago. But what I see now is that it is used for [indiscernible]. So what you're seeing now is these big companies, they're putting multiple like 10, 5 and 6 and 7 chips. I mean they don't have to be on the same process node. But all of them are like these advanced node chips to basically allow much bigger things to be build. So instead of system on a chip, it's like system on a package. And the beautiful thing there from a Cadence standpoint, that we have been waiting for this for a while. I had a whole effort on 3D-IC, like 5, 6 years ago. And it didn't go anywhere because we were kind of too early, and we worked with our foundry partner. But at this point -- because 1 thing we have to remember is we also have a very strong position in packaging. So Allegro is the most heavily used advanced packaging tool, which is our packaging solution. And then we have Innovus, which is for digital and Virtuoso, which is for analog. So we are the only company that can put all these things together from an implementation standpoint. So Allegro, Innovus plus Virtuoso. So that is very good from a implementation standpoint. And then on top of that, we have to run these analysis tools, okay? And so the key analysis tools that we are -- that our customers are asking for and the big foundries are thermal analysis. So the key thing that happened in 3D-IC is thermal becomes critical. And so this is very timely for Celsius because Celsius is the truly integrated thermal tool good for the chip and the system level. And also, Extraction, which is Clarity and also EMX, the latest company we bought is very good for a certain kind of interposer extraction. So long story short, I think, first of all, this is going to be a huge trend for next 5 years, all the talk is going to be about 3D-IC or these chiplets. And I believe that Cadence is in the strongest position, and it naturally plays into a traditional strength with Allegro and the new strength with system analysis.

Okay. Well, unfortunately, we're out of time. I have like 20 more questions and there's a host of other Q&A questions, but I think we have to stop here. But John, Anirudh, thank you again for joining us today, and if anyone has any follow-up, you know where to find us. Thank you.

Thank you.

Thanks, Jason. Take care. Thanks, Anirudh.