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

Okay. Great. All right. So -- and we'll certainly take questions from the audience as well as we go through, but I'm going to kind of run here for a while.

So first off, Anirudh, you've got a great technical background. You've been at the company for quite some time, but recently become CEO, you and Lip-Bu have helped to just turn this company into something that's pretty amazing today. But basically now as a CEO, as you look out and try to strategize on where this company is headed in the next 5, 10 years, let's see -- say, what are the biggest opportunities that you really see here for the company?

Yes. Mark, thanks for that question, and thanks for joining us today. I mean, we are in a good position. And as you know, the semi and electronic systems are in like a golden era, right? So for us, going forward, we primarily provide software and then some hardware and IP for the design of semiconductor and electronic systems. So the key 3 opportunities for us going forward, Mark, is, one is the semi business is doing well. So how well we can provide our software for semiconductor design. And that is a good tailwind. The second area is there are more system companies doing silicon, and that's a big trend that has happened in the last few years. And I expect that to continue in almost all verticals. So we have to make sure that we are critical to the semi companies that are doing more and more silicon now. And I think the third key area of opportunity is that our TAM expansion, we're adding more products for system design and analysis and that also naturally plays to these system companies. So these 3 vectors, I think, can continue for a while for the next 5, 10 years.

Right. You like to -- as we've had many conversations over the years, you'd like to frame this as kind of computational software. And -- so I think a lot of people think about Cadence as this EDA player that's really tied to semiconductors. And obviously, the industry ride -- $0.5 trillion semiconductors industry rides on your shoulders in terms of the ability to basically design devices and bring those to market, but you do way, way more than that. So maybe talk about just the bigger vision that you have for what Cadence taking that -- let me talk about core competencies and that computational software framework, what that can do to really take this company to the next level?

Absolutely. Good growth strategy is always based on what you're good at. So it's very important to know. I mean that's true for individuals, that's true for companies and true for organization. So it's a combination of what we are really best in the world at, plus what, of course, is happening in the market, right? So what we are really best in the world at is what we call computational software. So this is like numerical mathematical software. It's like CS plus math and traditionally has been applied to semiconductor design with EDA, but the algorithms are pretty general purpose. So if you look at all these whether it's matrix, algebra or numerical analysis, optimization, bullion analysis. So we probably have the largest scale of R&D for this kind of computational software. And the reason the Moore's law and the semiconductor industry has given the financial kind of backing for us to have this expertise. So as we -- so that's our core strength in computation software. And if you look at the silicon and the way we look at the word is that in these 3 concentric circles, so the outer circle is the data circle, then the inner circle is the system circle, and then the innermost circle is the silicon circle, okay? And this is happening in multiple industries. A good example is like an electric car, right? So you have all the data for navigation and then you actually have the physical car, which is the mechanical physical car with the mechanics and the software. And then the silicon that drives it. And then if you overlay on top of that, our strength in computational software. So in the design silicon circle, that's EDNIP, but in the system circle that is naturally system simulation and there's a lot of opportunity to do system simulation beyond semiconductors, things like thermal analysis and computational fluid dynamics and electromagnetics. And then if you go to the third circle, which is the data circle, AI is naturally computation software. So there's a lot of opportunities in AI and data analytics. So the way I look at Cadence strength is computational software, applied to silicon, which is EDA and IP and then expand into system analysis and expand into AI and data analytics.

Right. So how do you -- it seems like there's some breakaway opportunities here for you guys going forward. Just given the core competencies and kind of where the world is headed and where you can really be impactful around driving some of these broader applications and technologies. So how do you think about -- and this might be good for you too, John, how do you guys just think about resourcing the opportunities that you have out there? Because obviously, there's kind of unlimited things you could do you've got a broad set of capabilities, but you need to overlay those 2 against one another to make sure you get financial returns for investors. So how do you think about allocating resources against these opportunities?

Yes. That's a great point. Let me start, and I would like John to comment, of course, is, I think the -- first of all, we believe that small groups of people can achieve amazing -- all new products is always -- if you have the right talent, you can achieve amazing things, right? And then -- so it's very important to have the right talent, and we can talk about talent, and what we have done over the years and that's critical going forward. But at the same time, the allocation of capital is critical and allocation of capital with both revenue growth and profitability in mind and having enough investment pools that we can allocate to that. Maybe, John, you can go...

Yes. Typically, Mark, I mean, we make sure that we have the investment dollars. We attract the right engineers because they want to work on new products for the future. We don't have engineers spending a lot of time maintaining existing products because the existing products are so good out of the gate that they spend a lot of their time. The majority of our engineers' time is spent on products that are not generating any revenue today. So that attracts talent in the first place. And then Anirudh gives them great direction and we identified the pockets. We use the rule of 40, we tend to apply investment dollars or give investment dollars to the areas that provide the greatest opportunities for growth for us over the long term. And we measure that consistently. And if we're going to fail somewhere, we'd rather fail quickly and redirect that investment somewhere else. And I think that's worked well for us over the last 5 years. Last 5 years, we've managed to achieve over 50% incremental margins each year at average 55% with this model. And I think this year, I think it's the largest incremental margin we've had in our initial guide.

Right. And would you think that this -- is this sort of where the ceiling would be from an incremental margin perspective? I mean you guys are going to continue, I think, capture more value in what you deliver. But obviously, there's some theoretical limit. Where do you think that is today on an incremental basis?

Yes. So on an incremental basis, Mark. I mean, over the last 5 years, like I said, we've achieved over 50%. It's averaged 55%. So that means $0.55 of every additional dollar of revenue growth is dropping through to non-GAAP operating income. And if you look at the midpoint of guide for 2022 and compare that to where we were in 2016, which was the year before Anirudh, and I first did the annual operating plan together. We're on track to grow revenue by just over $1.5 billion. Over $800 million of that will drop through to non-GAAP operating income. And over $750 million of that is dropping through to cash. So it's good quality income. And like you say, if we continue to drop over $0.50 of every revenue -- incremental revenue dollars through to the bottom line, you're going to drag up the overall average. So the operating leverage we should continue to expand. There's no near-term ceiling on operating leverage growth.

Right. That's what I would think. So I think last fiscal year, you guys did 37% non-GAAP operating margins. You guided to 40%, I think at the high end this next quarter. And so it would seem to make sense that if you can keep doing 50% or better incremental that we'd monotonically increase to that over the long run.

And I'd remind you that we're achieving this while we're investing heavily to build out a multiphysics platform in a new area for us and system design analysis. So maybe 55% incremental margin over the last 5 years is conservative. But we aim for achieving over 50% in each year, but we're very, very pleased with the model.

Right, right...

So Mark, one thing to say, sometimes people ask us, why don't you give a long-term target. And the way we look at it is it should be like a Super Bowl number; it should go up every year. So the right target is to have an incremental number. And then the actual margin, which is like 38% right now can move up over time.

Right, which makes perfect sense. So not only will you guys, I think, increasingly have higher margins, but it seems to me that the growth could accelerate as well over time. If you go back 20 years ago, probably people thought about your growth rate as totally tied to R&D of the semiconductor industry, add all up and say it's growing at X and that's what we should be growing at. 10 years ago, I think people probably ask you a lot of questions of, given all the consolidation that's happening in semis isn't that a headwind for you. But it seems to me, as you apply this computational software opportunity from a bigger perspective of other verticals outside of semiconductors that your growth could theoretically accelerate. So maybe talk a little bit about that, some of the things you're doing in simulation, no longer being bounded by 1 license per 1 engineer. Talk about some of those trends.

Yes, absolutely. So first of all, if you just look at the numbers, like 4, 5 years ago, our revenue growth was about 7%, 8% and the margin was 26%. Now the margin is 38% like we talked about with 50% incremental margin. But the revenue growth, the 3-year CAGR is 12%. So it's already improved, and it is improving over the last few years. And the reason for that is, one is, first of all, semi itself is doing well. So it's always good if the core business does well. But also, we are more diversified with this competition software. And also when we go into a system design and analysis, not only it provides revenue growth, it also has higher margin profile than EDA, right? So that's our additional benefit. But the reason that it has higher margin profile and also has higher revenue growth. So one is that in simulation, system simulation, there are a lot of applications. The semi is one application, but there's applications in automotive, there is application in aerospace, application all the way in medical, right? A lot of the blood flow and all the CFD is used in that. So the number of customers is much bigger. So in the semi space, maybe we have 1,000 customers and maybe the top 100 drives most of the revenue. But in the systems space, there are probably 70,000, 80,000 customers. So the number of customers increases significantly. Now the top customers are still the same because they are the same household names that are doing semis and systems. But the number of customers is more. And then also simulation is more batch-oriented, so 1 engineer can do a lot -- can run -- instead of running 1 job, can run 10 jobs in parallel and especially with access to cloud and all this parallel compute. So that's another exciting area that drives revenue growth and margin. And the other thing is AI. So we have a new tool even for our core business, which is Cerebrus, which uses AI to automate what used to be done manually. So typically, if you have an interactive tool, and you have 100 engineers, you buy like roughly some ratio of like 100 licenses, but if you have more of this AI driving the tools, then 1 engineer can drive 10 of these in parallel because it's no longer a manual process, it's more of a mathematical way of doing things. So what I expect in that is, over time, as a percentage of our customers' R&D budget, they're spending some of it on our software and some of it on hiring engineers which is getting harder and harder. So I expect that to shift. We have an opportunity to get a bigger share of wallet on automation rather than on headcount, okay? So that's another exciting thing with this kind of AI-based automation of the design flow.

Right. And maybe we can talk about that a little bit more because it just seems to me that given the complexities of system design and whatnot that AI and machine learning ought to be really impactful from what you can deliver to the marketplace. So the problems you can solve with that. Cerebrus is a breakthrough product. Where are we just thinking in the arc of capabilities of the tools that you're delivering with an AI influence? Where do you think we are in that sort of progression over time?

I think it's still very early. I mean, it's still very early. And now we are very happy with the results of the two. So what Cerebrus does is instead of the -- typically, just to give you an example. If you're a designer, you will run in [ OS ], which is our digital flow. And then you run 1 or 2 times, then you adjust some parameters, then you run. So it may take like 6 months. Whereas Cerebrus will do it mathematically and may do it in 1 or 2 weeks with much better results. So the results are very promising. It's like 10% to 20% better timing or power, which, just to give you an example, that's the kind of improvement you get when you move from 1 node to the other node. So the industry is spending billions and billions of dollars to get that 1 benefit which we can get with better algorithms, okay? So I think the response is phenomenal. A lot of customers want to engage with it. I think what we are early on is like what I'm curious to see is what percentage of designers switched to Cerebrus as the only way of running stuff. So is it like 5%? Is it 10%? I mean I think it should be 80%, 90%, 100%. So we are still early in the penetration of how many people use that as the way to do things. Just to give you example, you were driving a manual car and now it's automatic. So how many people switch over to automatic? In that case, 95% switched over, right? So if the switchover like that to Cerebrus is a lot like we expect, then it's going to be a meaningful effect. But we are still early because we just launched the tool and I mean, it's a few years old. I mean we launched it last year. We've been working on it for the last 2, 3 years. So we're just curious to see the uptake on it. John, do you want to add anything?

Yes, I'd just say like a macro level consideration would be that, I mean, the level of design complexity is outpacing the growth in the number of engineers. So I think naturally, a greater share of wallet is going to have to go to the tool bag that the engineer carries and it's going to have more ML and AI and -- so -- and I think the beautiful thing about it is that it's naturally deflationary for customers because customers will get so much efficiency from us that we can increase prices and increase value to the customers and take a bigger share of that R&D spend. I think if you look at semi-R&D spend and compare it to EDA revenue, you're probably at like low double digits or low teen percentages that -- but over time, if you don't have the growth in the number of engineers, and you need growth in the complexity of the tools or the capability of the tools, that percentage will have to increase over time. And like I say, I think customers will get a great deal out of this and it will be good for us.

Yes. I mean I'm convinced this is the wave of where things are going to go. And I think it's just a massive expansion of TAM for you guys. It obviously takes a lot of capabilities to go do this. So who do you think from a competitive framework perspective, who else can go after that similar revenue opportunity you think?

Well, I think the good thing about -- that I feel that the Cadence position is that we have the broadest portfolio. So we are always strong in analogue. And then over the last few years, we've become very strong in digital design. And then if you look at going forward, there's a lot of activity in 3D-IC and chiplet. And we are the leading provider of packaging solutions with Allegro. And then there's all this thermal and system simulation that we have entered with computational software. So I truly believe that Cadence is in a unique position to really address -- there's no other company that can address the whole entire flow. And I think EDA also becoming more like enterprise software, like in the old days people who put their own flows together, and they'll buy 1 tool from 1 company, 1 tool from other company. But at these lower nodes and these complicated design, it's very difficult to do that. So they expect the provider like us to do that. So we are, I think, in a unique position. So I think it's our opportunities to take for this combination of the entire kind of system full flow, right?

Great. Let me -- I'll ask one more question but let me see if there's also -- okay, great. So as the mic is coming over to you, I'll just ask you another one here. If we can bring the mic up to this person here would be great. So you've talked a lot about just how the expansion is going to happen here with customers and orientation. And maybe just talk about where we are in the evolution of percentage of your business from traditional semiconductor companies and then on the system side. And just talk about how that might develop. We've obviously seen a lot of consolidation in our sector. There's going to be more of it. I think you'll see more and more companies that aren't traditional merchant semi guys build capabilities around semiconductors. We see it at Amazon, Microsoft, Cisco, you name the company. So what's that mix like today? And how does that ultimately sort of change the business fundamentals?

Right. And I think the other thing is like -- I mean there are semi companies and there are system companies, but some of the boundaries are getting fuzzy, right? So like what used to be semi company becomes system companies, look at NVIDIA is truly -- or like Qualcomm, there's 5G and all that. And of course, some of the system companies like we talked about, like Tesla is doing chips or you know about the cell phone companies. So there is -- the definition is a little bit fuzzy as you can imagine. But if you just use the same definition and we look at it over the last few years, right now, about 45% of our revenue is coming from system companies. And maybe 3, 4 years ago, that was about 40% of our revenue was coming from system company. So that is increasing because system guys are doing more diligent, and we also have more system product. So what could it go to? I mean it's difficult to predict because the semi guys are doing well, too, right? So -- but I think overall, it should edge up, but it shouldn't -- I don't think it will edge up that dramatically because there is a lot of strength in semi too, but we'll watch that.

Yes. Great. Questions from the audience? No one got you a mic?

I think you just answered it, but what portion of revenue is not semiconductor related?

I think that the way we break it down is the system -- the revenue we are getting from system companies is 45%. Now some of that is "semiconductor related" because they're doing silicon design within the system companies, but this is not a merchant semiconductor kind of revenue. And then some of it is just selling tools which are system design tools. But 45% roughly is from semi -- from systems companies, sorry. Yes?

I had 2 questions. The first is, as you get into systems design, are you going up against some of the simulation guys like Ansys? Is that who you're going head-to-head with or not?

Yes, that's one of the companies, definitely, yes.

Okay. So they have a pretty good legacy in that space. What do you bring to the table, I guess, if you're competing against them?

Yes. So I think the way to look at that is, first of all, there is -- as you go, the chip level goes to the board level and goes to 3D-IC and systems, there is more interplay between electronics and system simulation. But what we really bring what I was saying in terms of computational software. So what you have to remember is when you -- even our EDA business, our core business, about 1/3 to 1/4 of it is simulation, okay? Because historically, we do simulation for chips, right? And the complexity of that simulation is much higher than what the traditional simulation companies have done. So give you an example, if you're designing a 7-nanometer chip, there are like 50 billion transistors on that chip. So we have algorithms that will simulate like 50 billion variables efficiently because of our background in semiconductor automation. If you go to like something like this, and you want to do a thermal simulation of phone, so it's like 30 million by 30 million. And that sometimes may be considered like a big problem. So the -- but we have used to doing 50 billion by 50 billion simulation. So the amount of algorithms and complexity that is there at the chip level can be applied, and that's our core competency at the system level. So when we launched our first tool clarity at the system level, it was like order of magnitude faster than all these other tools because of the parallelism and the algorithms we have in the 30 years of experience in EDA that we can apply to the system well.

Okay. I have a second question, which is on the core EDA side. As Moore's Law slows or stops, what does that mean for your business?

Well, I think, I mean, there are multiple aspects to that. So first of all, I mean the death of Moore's Law has been predicted for a while, even when I was in IBM like 20 years ago. And I think it just [ morphed ] in different ways. So I mean, we still have several process nodes to go -- and then on top of that, the 3D-IC happens. So the industry always finds a way to evolve, and the Moore's Law becomes more at the system level than at the chip level. You have more chips connected on a package and things like that. Now even if there is some slowing of the Moore's Law eventually, our business is pretty diversified. We have digital tools, but we have also analogue tools. We are the leader in analogue, which are more mainstream, 65, 180, also this advanced packaging. And it also gives time for like things like Cerebrus to be deployed because if you are not getting advantage from the process improvement, the advantage is going to come from better algorithms and design flows. So what is happening is as Moore's Law is slowing as we go from 7 to 5 to 3, the importance of EDA is increasing in the design flows. So even at 3-nanometer, we do a lot of special things for 3-nanometer to produce jointly with our foundry customers to give that advantage over 5 nanometer. So it's not just purely coming from process scaling. There's a lot of EDA algorithms that even at 3-nanometer and that's going to become more important at 2-nanometer and 89 strong or 49 strong. So I think there's still a long runway. And even if there is some slowing down, there is more time to run the tools and also more time to get benefit of our R&D investment. So I'm pretty optimistic this can go on for a while.

Allow me just to add to that, that a little understood thing, I think, is that as Moore's Law slows down, we take -- we get a longer time to harvest what we've created because our R&D, although we spend 35% to 40% of our revenue in any given year on R&D, more than half of that R&D expenditure is in relation to products that don't even exist today that are not generating revenue today. But -- so if Moore's Law slows down, you get longer to harvest what you created, and it just naturally will improve profitability.

Question right here.

How should we think about the cyclicality of your customers' businesses impacting your business?

Well, I think if I look -- if you go look back even in the like last time, there was some cyclicality like 10, 15 years ago. So one thing is the EDA business is always one level removed from it. So because we are tied to the R&D spending. And even if there is some correction usually the R&D is the last thing that's get affected, okay? So that's what we have seen in the past. Now going forward, first of all, we don't see any -- especially on the design activity side, we don't see any slowdown right now. And then we have diversified now to the system companies doing semi and then, of course, system design tools. So I think we are in a pretty good position, and we are, of course, well run financially, but we'll see what happens.

I have just 2 very quick last questions, one for you, John. Clearly, it seems like we've got an opportunity here for pretty strong growth and very good cash flows as margins expand over time. So how do you think about return of capital versus M&A or other things off the balance sheet?

Well, generally, Mark, that the first priority is investing in the business, investing in R&D and organic growth. Second priority would be tuck-in M&A, and that's where we can accelerate our road map typically, that's where we can attract top talent or top technology or like you say, accelerate the road map and importantly, at the right price for us. And then the third thing is returning cash to investors by way of share repurchases. Typically, we aim to repurchase at least 50%. We use 50% of our free cash flow to repurchase shares. It will be more than that this year, and we're repurchasing $250 million worth of shares in the first quarter of this year.

Right. And just last question for Anirudh. So you guys are an incredible company. You've done a lot of great things certainly; investors have a pretty good understanding of the company. But what's the one thing you think investors don't know or under appreciate about Cadence?

I think I would say that is not as well understood that Cadence is mission-critical to semiconductor and electronic systems. I mean these -- without this kind of software and hardware product, it's just not even possible to design any of this. So you go back to the dark ages if you don't have this kind. So I think that's something that is a great position to be in. And I think the other thing I would say is that we have this computational software expertise that lets us expand into other verticals. So the combination of how critical we are to the semiconductor and electronics ecosystem, and how much we think we can apply to other areas like simulation or AI. So that would be it.

Wonderful. Well, thank you guys for being here. Thank you, guys, for attending. And we'll go on to our next session.

Okay. Thank you.