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

Okay. Welcome, everybody, to D.A. Davidson fireside chat. Today, we have the management team from Cadence Design Systems. Before we get started, I'll hand it off to Alan Lindstrom to do his business.

Yes, here's the safe harbor statement. Before we begin, I need to mention this. Today's discussion will contain forward-looking statements and will 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. Back to you, Tom.

Okay. Well, thank you. So with management today, we're pleased to have Anirudh Devgan, who is the President; and John Wall, the CFO. So with that, I think we'll just start with a few of the kind of the current topics here, if that's all right. A lot of questions I still get from investors is, all right, are we through the COVID issue yet? Is COVID still an issue? So maybe just spend a couple of minutes talking about the impact both on operations as well as the supply chain, what you're seeing today? And how long do you think this will last?

Yes, Tom, great question, and thanks for the opportunity to update everybody. Yes. I mean, unfortunately, the COVID-19 pandemic continues to evolve in an unpredictable manner, I guess. But we're pleased, though, with -- I'm very grateful for the response of our leadership team out in India, who went above and beyond last quarter. I think Q2 was a really, really difficult quarter for us in India. I think we're through the worst of that. And certainly, in the last couple of months, I don't think we've had any cases, and the vaccination rate keep climbing. So that's a positive. I mean generally, I think we've been in this pandemic environment, what, over 550 days or something now. But -- so I think it's become a new normal for us to a certain extent, and we've operated pretty well. I've been encouraged by, like I said, the recent reduction in cases in India, we're naturally modeling the impact of Delta variants and everything. But I've been pleased with how we've been able to adapt, and I think the company has been very resilient. I think, like I said, the issues that we raised. I think probably at the end of Q1, we saw that the spikes in India, I think we're through that now.

Great. And then what about just on supply chain with some of your hardware businesses?

Yes. On supply chain, that's, I guess, a few years ago, when I go back to 2019, that we made a conscious decision that from an inventory perspective that we wanted to move from a just-in-time basis to just-in-case. And we did a very diligent exercise and looking at our supply chain to make sure that we had second source suppliers for any major components. And to the extent that we felt we were exposed with anything. We did larger purchases. So we retain a higher inventory for those. So I think we were more prepared than most. We started that back in like say, back in 2019 when we noticed that there were increasing kina trade tensions, and we thought it was a good idea back then to kind of review the entire supply chain. So I think we were more prepared for the pandemic as a result and for chip capacity constraints.

Yes. Okay. I mean what's your sense of just the duration of the current chip shortages that we're seeing in the marketplace that may impact some of your customers?

Yes. Hard to tell, Tom. I mean, when I talk to other CFOs, everybody is reacting the same way. It's kind of double ordering is the solution double-ordering kind of makes the problem worse than it is. So it could solve itself pretty quickly. It's hard to tell, though.

Yes. Okay. No, it sounds good. Another recent topic really is just kind of your involvement with a few acquisitions, NUMECA and Pointwise, in the fluid dynamic world. Maybe talk a little bit about these acquisitions, how they fit into your long-term plan? And then how big this emerging business is or could be for you?

Tom, so I will answer that one. So we are pretty excited about CFD, computational fluid dynamics. As you know, we are expanding into system analysis because we believe that, that is very synergistic with EDA for a variety of reasons. There's a lot of customer synergy, there's a lot of technology synergy. That's what the customers -- that's where we think the customers are going, the blend of a chip and system design. And no system analysis effort is complete without CFD. So when we started, it was more on electromagnetics and then thermal, which is about $800 million market. But CFD, we estimate about $1.6 billion. And so I think it's critical to our long-term plan. Because for system simulation, we need to simulate solid, okay, that's like finite element. So clarity is a final element solver. And then you need to simulate fluids, which is CFD, and therefore, the acquisition of NUMECA and Pointwise. So we are pretty happy with the customer reach and the technology we got. And on top of that, we can keep improving that. So we have a lot of expertise that we will also bring to CFD as we go forward.

Okay. So who are the big competitors in that space? Who do you see most of the time in the marketplace?

Well, CFD, I think there are 2 or 3 bigger competitors. The good thing with CFD is that there is a lot of applications for it, all the way from medical, to aerospace, to automotive. So there's a lot of applications. So there's a whole bunch of small companies. But among the big companies, there are people like Siemens and Dassault and ANSYS, I think, are -- with varying degree, are the 2 or 3 of the bigger ones, and then a few small ones. But the market is big and evolving. So we'll keep that close eye on.

Okay. So there's a couple of the traditional EDA guys that are playing in that space as well in addition to you?

Yes. So I think Siemens acquired Mentor, but they do have a CFD business. And then ANSYS, of course, has -- also has CFD. And I think this trend, and you must have seen the recent news of Autodesk trying to acquire a PCB company. So we are also leaders in PCB and packaging with Allegro. So I think this trend of system and chip coming together, I think we've known this for a while because we have a good kind of front row seat to what is happening in the industry. So I think it's still -- this is being seen by other companies, but I think we are in a good position to do implementation and analysis with this chip and system coming together. And the other thing that accelerates that is 3D-IC. I mean, 3D-IC is going to be huge. So that's a natural intersection of chip and system. And one of the key issues in 3D-IC apart from the design part is the analysis. So one of the big foundries told me that one of the biggest things they see is thermal issues in 3D-IC. And that actually requires a combination of finite element and CFP. Because in thermal, you need to more air flow and then you model heat generation from the chips. So Celsius do that, we have embedded CFD and finite element in it. So I think this trend is going to continue and accelerate. And I think the other thing I want to highlight apart from analysis is important of 3D-IC in the future.

Okay. That makes sense. I think that was part of the thinking too, and Siemens actually who came in and bought an EDA company to get some of these tools to link it from start to finish across the whole supply chain. Interesting. Okay. The other big topic today is kind of the emergence of artificial intelligence or machine learning in the design process. I know you've recently made a little bit of noise with you. I think it's Cerebrus? Talk a little bit more about that and more maybe more specifically, where you see machine learning impacting EDA over the next few years?

Yes, Tom. That's a great question. So -- and I'm pretty excited about Cerebrus and what it could do for us. But before I describe Cerebrus, I think one thing I want to highlight is -- and if you look at our strategy, Intelligent System Design, we talk about like 3 circles, 3 concentric circles. So the inner circle is silicon design, okay, which is, of course, for us, ED and IP. And then the circle around that is system design. And then the circle around that is data analytics and the whole thing we take. And what we think our core expertise is and we are best in the world that this is computational software. It's kind of numerical mathematical software. This is what EDA has done for 30, 40 years, long before even I got involved in it. But we are -- my background is a numerical analysis. So if you take that core expertise and extend that to system and data, that is what we are trying to do. So the computational part of system and data. So the computational part of system is system analysis, simulation, and we just talked about that. And the computational part of data is naturally machine learning and AI. So the machine learning is inherently computation, the basic inference function is like a matrix multiply, trading is some kind of a conjugate trading. So we have a lot of expertise in these methods, which are now emerging in AI and ML. So of course, we hire people from outside. But internally, we have 6,000 people in R&D. A lot of them have advanced degrees. A lot of them can do these larger simulation about AI and ML. So we have about 20 -- more than 20 projects in AI and ML. And of course, the definition of AI has become very broad now, right? But this I'm talking about like real, like neural network base or deep learning or this computational kind of methods. So one of them is, of course, Cerebrus, which we are pretty excited about. Last year, we did talk about verification and ML. We have Xcelium ML product, which is giving a lot of benefit for simulation, for logic simulation and verification. Now coming specifically to Cerebrus, okay? The best analogy I can think of is ADAS. It's a driver assistance for chip design, okay? And I don't think we are at like "full self-driving," and I don't think we will get there. And I don't think that's the right goal. But the goal should be that one design instead of doing one block or driving one design can drive like 4 or 5 designs in parallel because of the productivity improvement. So that's -- we have seen that. And the other benefit is that you can get a better quality of reserves because this is done mathematically. So to give you example, that typically each block takes 1 or 2 days to run in our tool, like Genus, Inovus will take 1 or 2 days to do the block. And the chip may have 100 blocks, so you run 100 of them in parallel, you get some results. But the chip design takes about 6 months, right, or 9 months or 1 year depending on the chip. So what the designer is doing is running this over and over again with different parameters. So historically, that is done by the infusion of the designer. But now we can do it mathematically with Cerebrus, and we have seen like 10% to 20% better PPA. So that is significant, PPA, I mean, power performance in the area. The reason that is significant, for example, Renesas got a 10% improvement in frequency in one of our public endorsements. I mean that's almost equal to one technology generation. And if you go from 7 to 5 or 5 to 3, you could get 10% to 20% performance. So you're getting that just by better algorithms and software. So therefore, the response is quite good. And the other benefit we are also trying to use Cerebrus internally with our AE team to improve our own productivity and cost of supporting customers. So I think this is an important tool box that -- and we believe that we have all the state-of-the-art technology, and we will be a leader in this space. Now it still sits on top of the design system. So you have Innovus -- Genus, Innovus, and this will sit on top of that. So it's like you need to have a Mercedes and then -- or Tesla or whatever and then self-driving on top of that. So I think the AI part is critical, but the base car is critical, too. So we want to make sure that the base tools give best quality of results, and then you have ML on top of that. Just having ML with a not-so-good base card is not go to the help either. So we need to do both.

Okay. Well, it's not too long ago. We used to talk a lot about the number of engineers, the seats, if you will, and how that drove EDA. It sounds like with the new tool set, you get kind of an acceleration of what a single seed or employee can do and obviously, great for your business long term to be able to accelerate productivity and the usage of your tools.

Absolutely. So I mean typically, what we have found is the more productivity we give, the more experiments the designer would do. And Cerebrus, it runs in parallel, a lot of times, it will run 10 -- each block, people run 10 machines in parallel and do all these experiments mathematically. So then they need to have Cerebrus and then multiple copies of Genus and Innovus. So it drives more usage. And the other thing that I want to highlight is part of our business and even in EDA is simulation, about 1/4 of our business is simulation. Now the implementation tools traditionally are tied to a number of engineers, and things like Cerebrus make them a little more untied. Even 1 designer can run more and more jobs. But in simulation, that has always been true. So because stimulation is not interactive tool. So 1 designer could always run a lot of simulation. And therefore, simulation is always the most profitable part of EDA. And therefore, when we go to system analysis, that is more profitable than EDA. So historically, simulation is not coupled with the number of designers and it is exploring because you need to analyze more and more. And then implementation, you're right, traditionally has been tethered to the number of designers. But with things like Cerebrus, it becomes more untethered and 1 designer can do a lot more with experiment.

That's great. Okay. Another kind of big theme in the world over the last couple of quarters has been this emerging foundry war between TSMC, Intel Now and Samsung. So John, I'm wondering, have you seen the real impact of that yet with your fabless customers? Or do you think that's to come over the next few years as they spend tens of billions of dollars to build out these new fabs?

Yes, Tom, I mean, generally, in these cases, we're Switzerland, I mean neutral that we're happy to help all of our customers and everyone you're talking about, there is our customer is, too. So yes, yes, we generally remain neutral. Just try to help everyone through it that even when you talk about chip capacity constraints. We're not at that end of the design cycle. We're in the production cycle, we're in the design cycle. So typically, those constraints don't impact us as much because we probably correlate more to design activity. The more design activity there is then it's good for cadence. But at the production end, that's all the people's business, and we wouldn't comment on that.

All right. I guess the point I was trying to make is that over the next few years, a lot of leading capacity is going to open up to new potential customers for the foundries, and that should enable your customers to start doing more development at the bleeding edge, if you will, besides just the top over 5...

Absolutely right, Tom. Anything that promotes more design activity is very, very welcome to us.

And, Tom, just to add there. I mean you're right. If there is more foundries, then, of course, we have a very good position in the market, we believe. So all those foundries need to enable our software and our IP so their customers can use it. So it's a net positive, but we are still in the early innings. Of course, we are very close to TSMC over the years. This is well known. TSMC and Cadence are very close over the years. I think majority of TSMC's internal usage is Cadence based. We have become very close to Samsung and we talked about Samsung in Q2 earnings call. And then, of course, there is new entrants in the U.S. with Intel. So we are very optimistic to see that there's a lot more foundry activity.

Great. Okay. And then the last kind of new recent tidbit is on a call a couple of weeks ago, Synopsys talked about how their end markets and the strength there, enable them to grow double digits and they're projecting to continue to grow double digits for the next few years. When you look at your end markets, do you see a similar trend as far as ramping up of activity across all your served markets and to potentially get into that consistent double-digit growth rate?

Yes, Tom, we heard the commentary on the kind of long-term industry growth rate on that call. And I welcome as clear -- clearly, that's an indication that the industry is much more rational regarding pricing. But we don't have a stated long-term revenue growth target. And naturally, we look forward to talking to you about revenue growth for 2022 when we're done with '21. But when we talk about -- like you heard Anirudh talking earlier about our Intelligent System Design strategy. That opens additional growth opportunities for Cadence, and we're very pleased with the progress we've made over the last few years. We've generally seen accelerating revenue growth over the past 5 years. You'll see that if you look at the CFO commentary that we published on our Investor Relations website, we call out the 3-year CAGR deliberately because I think that's a good way to judge the growth in our customer base because the revenue growth in our customer rates because so much of our revenue is recurring in nature. And many of our customers tend to purchase on a 3-year cycle. Average duration is about 2.5 years because a customer will typically start with a baseline 3-year contract. And then over the course of those 12 quarters, they'll come back somewhere between 4x and 7x and purchase add-ons to that technology that they purchased in the baseline contract. And when we add it on to the baseline contract, typically, we'll co-terminate. So if you're halfway through, you probably have an 18-month contract that co-terminates with the underlying baseline contract. And then at the end of the of that baseline contract when you're renewing, you kind of bundle everything in together and you start the whole cycle again for another 3 years. So I like looking at the 3-year CAGR. But generally, we believe like EVA and technical software industries can grow faster than the overall economy. We're very happy with the pricing environment and the focus on profitability. Another thing I heard was a lot of focus on Rule of 40 metric. And as you know, we've been focused on improving our Rule of 40 metric for the past 5 years, too. And I think that helps maintain good discipline in pricing. So our goal is to generate strong profitable growth while flowing through over 50% of incremental revenue to non-GAAP operating margin. And because of that, that's why we don't really foresee any near-term setting on operating margins either.

That's great. Okay. And I know the bleeding edge, leading edge gets a lot of the attention, the 7-, 5- and 3-nanometer nodes. What are you seeing in the more mature nodes? And how has that been as an incremental driver for you? We've seen a lot of capacity being added at the trailing edge nodes. And it seems like over time, they would be more EDA intensive as well.

Yes, Tom. That's a great point. So I mean, apart from logic, let's call it, logic, advanced logic, like 7, 5. And a lot of our R&D right now is at 2-nanometer, okay? What is really interesting for Cadence is that, we, historically, is very, very strong in analog, and we're still the preferred solution for almost all analog mixed signal implementation. And the more mature nodes are inherently more analog and mixed signal. So whether it's 20-nanometer or 65-nanometer, 40. And then they also have a more -- like we have more specialty technologies integrated with them, like embedded flash or other RF and things like that. So traditionally, and we continue to be the leader in that space. That's why I'm not just with Virtuoso and Analog. But also RF needed acquisition of AWR from like National Instruments, integrate that with our Virtuoso platform and the Innovus platform. And these things are also more mixed signal. And there is a lot of activity with this automotive design. So we are in a very strong position in all the automotive players. I mean some of them are public. So inherently, automotive is mixed signal. So traditionally, they had done analog in digital, but now they do analog and digital, both in Cadence. Traditionally, they had done analog in Cadence. So things like TI or -- STMicro, all these companies are -- we have deep collaboration with all the major automotive companies and automotive chip supplier. And the other thing that's happening is -- and the more than more, let's say, call it. So the other thing that's happening is, of course, the integration of advanced packaging. Because when you have these disparate technology, sometimes they're on a single ship. Sometimes you have multichip solutions. And again, we are in a very good position with Allegro being the leading platform for advanced packaging. So that's the strength of the Cadence platform that it is not just digital, it's digital plus analog plus advanced packaging. And it's not just 3-nanometer, though we're working with all the top customers at 3-nanometer. It is also mainstream nodes and packaging. And then we want to extend that to analysis. That's even more number of customers with system analysis. So I really feel that the breadth of the portfolio, and we have worked very hard in the last few years to make it best-in-class. So to have best-in-class portfolio and have the breadth, we can really reach most of the people doing design, whether it is advanced node or mainstream nodes, specialty nodes. And including memory, let's remember, memory is a big part of the semiconductor ecosystem, DRAM, flash, A lot of them are using our analog mixed signal and digital platform, too. So it's logic, advanced node, mainstream node, memory, packaging, system analysis. I think all this has to come here.

I think that's a great place to leave it. A nice overview. Appreciate it. John and Anirudh. I appreciate your time today and look forward to hearing more about as the story evolves over time. Thank you.

Thank you.

Thank you, Tom. Thanks for having us.