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

Good afternoon, everyone. Before we get started, if you are a member of the press or media, please disconnect at this time. This is a restricted line. Any unauthorized party in this meeting or any unauthorized use of the information communicated in this meeting is subject to prosecution to the fullest extent of the law. Any unauthorized person, including the media that is on the line at this time, please disconnect. Please note, today's call is being recorded.

Good morning, everybody. I hope all is well with you. My name is Gary Mobley. I'm one of the semiconductor analysts at Wells Fargo Securities. Today, we have with us the management team from Cadence Design, actually one of my favorite management teams, and I'm not just saying that to all the management teams in TMT conference. We have with us Anirudh Devgan, the CEO of the company or soon to be CEO, I should say; and John Wall, the Chief Financial Officer. And we also have Alan Lindstrom, who heads Investor Relations for Cadence. And with that, I'm going to turn it over to Alan for just a few short remarks.

Thanks, Gary. So yes, before we begin, I need to 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 10-K, 10-Q and website for a discussion of risk factors and our use of non-GAAP financial measures. And back to you, Gary.

Yes. Thanks, Alan. I want to establish a context for a discussion. So I want to sort of go back in time, 13 years ago, in 2008 when Cadence was in a much different place. The EDA industry was in a much different place. The company didn't have the same ratable revenue recognition model it has today, and I think there was a failed attempt to buy Mentor Graphics, your balance sheet wasn't as good as it is today. Your profitability certainly wasn't as good as it was today. But bring us up to speed in terms of what's changed over those 13 years and then maybe more targeted, can you give us a sense of how [ Cadence's certain label ] market has changed as part of the Intelligent System Design strategy, start there with an overview if you can?

Well, Gary, that's a great question. 13 years, that's like eternity ago for us, I was not even in Cadence. I think John was in a much different role. I think 13 years ago, there were several issues. I mean one of them you mentioned there was -- the revenue recognition was more "upfront." It was not ratable. So that causes a lot of problems when drilling deals, and you know all the issues with that. So that's a big problem. Also, the -- we were not strong -- we were always strong in analog, but we were not strong in digital, okay? And that's a big issue because a lot of the big companies are -- have a lot of digital presence I think. And the third thing was we were not very strong with really the top companies. Because of analog, we were more like with the smaller companies. And I think the R&D teams were -- it was not a technology R&D-focused company, okay? There are some of these issues from 13 years ago. And I mean, a lot has changed. Almost all those things have kind of flipped around completely. We are fully ratable. We had a lot of flexibility in terms of booking timing. We are not worried about end of the quarter. I mean, we are, but not really in terms of bookings. The company is much more R&D focused, technology-driven. We have hired a lot of people to change the culture of the company. And we are in a very strong position in a lot of areas, including digital, analog, verification, okay. And I think this has been a journey over time. And then the question was, so it took like at least I joined 2012, I think Lip-Bu joined in 2008, and it took like a few years to fix the core business like digital, and we can talk more about that. But by 2016, 2017, we were headed in the right direction in core EDA, I believe. Now of course, it takes a few years for the results to show. But at least by 2017, we thought we were in a good position. And our destiny always, we believe, is to be the largest EDA company. And that's the same time that -- around the same time, John become CFO, started working closely with John. And so I think what we wanted to do after that is to see what is next for the company, the last 4 years. And that's where we focus on this what we call Intelligent System Design, which is -- EDA is always a great business. The question is, where else can it grow? And what else he can leverage from our strength in EDA. And one key area is expanding into adjacent areas, but based on our strength, okay? So because there's always competition wherever you want to go, and all companies want to grow. But a lot of times, they don't grow in the areas of strength. So it's very important to figure out, one, what's our area of strength. And two, where is the customer going and the intersection of that is the right direction, okay? So our area of strength is what we call computational software. So this is numerical, mathematical software. We have like 6,000 people in R&D, a lot of them with advanced degrees. And -- so that's what we are good at. We are best in the world for that kind of software. And where the industry is going, as you know, is more towards data and systems. So we look at the world as 3 circles, the outer most circle is data, then systems and then silicon, these concentric circles. And we have said that about 45% of our customers are already system companies. So when we expand beyond silicon or EDA, the natural adjacencies are system simulation because that's system software, but computational and then data analytics and AI from the third circle. So we are focused on being the best-in-class company in EDA and IP and then expand into system analysis, system design and analysis and AI and data analytics. So I'll stop there, Gary, to give -- it was just quick overview.

That's a great overview, and I think fully encapsulated -- encapsulates your growth strategy. Not only is the EDA market accelerating and we can get into more detail there, but you're moving into these adjacencies. And related to that, move into adjacencies with maybe the 5 or so multiphysics simulation software tools that you have developed internally and inorganically. Maybe can you give us a sense of how that initiative is translated into revenue growth impact your backlog and then as well maybe the margin accretion dynamic to that?

Yes, absolutely. So I mean, the other good thing about system design and analysis that's where the customers are going, number one, that's where we are -- we have expertise. But the third thing, like you mentioned, it has a great financial profile. The market is growing and also it's a high-margin business. I mean EDA itself is higher margin, but system design and analysis is even higher margin than EDA, okay. Because simulation inherently is higher margin. So about 1/4 of our business even within EDA is simulation, and that's high margin because one engineer can run a lot of simulations. Whereas an interactive tool, one engineer can run one tool. But if it's a simulation tool, one engineer can run tens or hundreds of simulations to do a lot of experiments. And the R&D cost is fixed, but the revenue cost -- the revenue can be higher because of the number of seats, so the margin is better. And we are seeing that already -- even though we are investing a lot in the new areas, we are already well profitable in those areas. And I think we break out system design and analysis as a separate segment. It's about 11% to 12% of revenue, and I think John can give you more details on that. And it's growing pretty well last 2 years because the question was, okay, if you're going to do it organically, and you're entering in trench space, can you really be successful. And I think last 2 years, we have grown rapidly. And we have grown rapidly even with taking revenue ratable. So even in system design and analysis, we are taking revenue fully ratably compared to some other company and still able to grow pretty fast in revenue. So that means bookings are growing even faster than revenue.

John, do you want to add?

Yes. So Gary, I mean, it's a great point. In terms of system analysis, we do find that, that's the most profitable area or simulation generally is the most profitable area of business for Cadence. And if I may, I contrast it with IP. But I mean, there's parts of IP that are very profitable, particularly, I mean, you've got a royalty stream of revenue is coming through with pretty much at 100% margin. But if you're doing design IP work where there's a heavy customization and a lot of like labor-intensive customization, the margin profile on that could be low. But on the simulation side, that the reason it's typically got the highest profitability levels is the revenue can go nonlinear. I mean if you take say -- I mean we have a tremendous analog franchise in our custom IC business. I mean if you picked randomly a 100-person kind of analog design house, around 100-engineer analog design house, chances are they're buying 100 licenses a Virtuoso from us. But we won't sell them 110 until they hire 10 more engineers. So your growth in that area of the business will come from a combination of the growth in the number of engineers and pricing. But generally, because we're such a large part of the analog market that you're not going to get a lot from margin -- or from market share gains. It's really going to come from growth in the number of engineers using your licenses and pricing there. In digital, you probably get a little bit more from market share gains because we continue to advance there. On the -- but on simulation tools specifically, I mentioned Virtuoso because it's an interactive tool. You could probably bifurcate a lot of our software tools into those two groups, interactive tools and tools that can be batch processed. And on the simulation side, you can have one engineer kick off 10 or 20 simulations and use 10 or 20 licenses. And when Anirudh and I were looking through the profitability profile of that 5 years ago, we were saying, well, what's stopping them using 50 licenses. And we've thought, well, it's probably access to compute power, and that's kind of job Anirudh say, well, we needed to be first to the cloud then. But now the industry might not be ready to go there, but we want to be there and ready for when they are. But because it's clearly beneficial to us.

That's a great overview. I wish I had more time to talk about that, but I do want to get on to the topics of Moore's Law and the slowing of Moore's Law arguably the longer the time interval between the transition from one process node to the next smaller might slow the pace of innovation, chip design, which is the lifeblood of EDA licensing and IP licensing. But on the other hand, when you get more versus relying on Moore's Law, [ more than Moore ], so to speak. And that might include new packaging schemes that might include new way to lay out transistors, 3D transistors, for example. So what is the net impact to a company like Cadence from the slowing of Moore's Law? Is it accretive to your growth or dilutive to your growth?

So Gary, first of all, a couple -- you made very good points, but I just want to highlight 2 of them. So one is there is a lot of nodes still to come in terms of Moore's Law in pure silicon scaling. We are -- most of the volume manufacturing is starting at 7 and 5 right now. There is some 5, a lot of it is 7. So we had 3 to come, then 2 to come and then 1.4 or something between 1 and 2, okay? So that's at least 3 more nodes, okay? That's like 2, 3 years. So that's still like 5 to 7 years of pure scaling still to come at least. And then now -- you're right that when the scaling happens, the things are getting smaller, but they're not getting any -- they're not getting as fast as they used to in the old days of scaling, like 10 years, 20 years ago. So you get a lot of stuff on the chip, but it may not be running faster, okay? But I think we will get more integration happen on a chip for next 7 to 10 years or 7 to 8 years. Now on top of that, the industry is evolving to this 3D-IC, which is more chips on a package. And this has a lot of advantages. And the reason is that it naturally extend Moore's law by having more capacity. And for applications like AI or GPUs or even CPUs, if you remember, like 10 years ago, you would have one CPU, now you have 8 CPUs. So they may not get any faster, but there is more integration of things on that. So there are more of them that you can use. So the software has to make use of these more CPUs or GPUs. So that's why we are doing parallel processing for years now, actually, ever since I joined Cadence, we have a big focus on parallel processing because I saw this coming in like 2,000 in IBM. We were -- the classical scaling has been in problem for like 20 years, okay? So that having been said, it plays very well to Cadence strength. So one, we are already doing pretty well in advanced node and that will continue for digital. We are a leader in advanced packaging with Allegro. And so we are very well positioned with 3D-IC that we have analog digital packaging and then system analysis on top of it. We can talk more about that thermal and electromagnetics. So we don't see any issue with the profitability profile continue to improve, okay? And I think I would like John to comment on our focus on incremental margin. We like to have relative target, not absolute target. So John, maybe you can comment on that.

Yes. Actually one thing I wanted to mention, Gary, is that -- I mean, you know that we spent like 35% to 40% of our revenue on R&D. But most of that R&D effort is spent on products that don't generate any revenue today. So one impact of Moore's Law, if Moore's Law were to slow down and people spend longer on process nodes that the impact on us is we get longer to harvest what we've created that because the R&D costs associated with creating products at various process nodes is already sunk upfront. And then you're on time-based licenses for revenue. So if you get longer to harvest, what you've -- basically what you've planted in terms of all of the -- all the work that's gone into previous process nodes, then that's very, very good for profitability and continuing to improve margins. That's -- yes, and on incremental margin, Anirudh, that's -- yes, Gary, I mean, you know that we scale very well as a company, but we're trying to focus on dropping $0.50 or more than $0.50 of every dollar of revenue growth to the bottom line. And we're on track to achieve that for the fifth year running. So we're very, very pleased with it.

Good, good. Yes. So the semiconductor design community has been changing for quite some time. What we're seeing is a couple of different trends. We're seeing more system OEMs developing their own chips internally. And to put some numbers around that, one of the projects we're working on is trying to compile a list of all those system companies automotive OEMs, hyperscale is developing their own silicon. And it's about 3 dozen long right now that I can think of. And if you go back 10 years ago, that list was probably half the size. We hear Ford, GM announcing initiatives to take more control of their chip design and align with the right manufacturing partners. And so my hypothesis is that this shift from the merchant chip design community to captive chip design is accretive to the growth rate of the EDA and IP industry. And so with that in mind, for those -- that 45% of your revenue that's tied to system OEMs, are those average deal sizes typically larger than they are with your traditional merchant chip company? Is that the way it becomes accretive to your growth rate?

So Gary, let me start. I think, first of all, that's a fabulous trend that you mentioned, system company is doing more chip. And we can talk a lot about -- we can talk for a long time about that and the reasons for that. And this is not going to change anytime soon. And there is so much silicon that is going to be consumed. One question is, will it affect the semi companies, but I don't think so, I think it will be overall accretive and -- for multiple reasons because these are more specialized chips for particular applications. Now in terms of EDA, IP is fabulous, because we have more customers, and these are big companies. These are big system companies doing their own chips. And even in our Q3 earnings call, we talked about Cadence working with Tesla for a while, and that's a remarkable example of a system company or a new company doing chips. And the -- and your question on deal sizes, I mean the semi-companies have a whole range of customers, right? So some customers are big, some customers are small. So there's a whole range of deal sizes even in the semi companies. But the good thing with the system companies is because they are very big if you're doing your own silicon, I mean the small system companies are not doing silicons. So naturally, these are big efforts, and they are on the bigger side of the deal sizes that we have seen. But I still think that it's early days for them. So the other interesting trend is not only there doing chip design, when I look at several of the system companies, they plan to expand that chip design over the next few years and do more and more chips. And then the other thing is because we have the system design and analysis tools, we can sell them not just EDA tools, we can sell them the other tools on top of it that interact with the system and 3D-IC. So overall, this is a very positive trend for Cadence, I believe, and the industry.

Okay. Another example of where we're seeing more fragmentation in the chip design community is China, which today represents about, what, 14% of your revenue? And then if you exclude -- or if you focus specifically on customers who are domestically oriented in China or domiciled in China, it's probably, what, high single-digit percent of your revenue. But yet China consumes about 30% of global semiconductor output. My question to you is, do you ever see that China's chip design activity and as a representation of your overall revenue ever being close to what they consume from a chip industry perspective? Can they ever be a 30% mix of your revenue?

I think you make a very interesting point there about the level of semiconductor output. Not quite sure what the comment on that, but I mean there might be some imported components in that 30% number that you're quoting. Generally, I mean, I'd expect that our EDA revenue is more likely to be in line or above average compared with the design activity that actually takes place in China. And design activity has been quite strong there in recent years that -- and I know -- I mean, you made the point that our China revenue that's sold in the region is likely less than what we report because I think that's an important thing to understand about our geographical mix of revenue that is based on consumption rather than sales activity in the region. Our China revenue will benefit from allocations from sales to multinational customers based outside of China. So it's a -- we sold to a multinational customer here in the Bay Area and let's say they renew their baseline agreement with us. If they tell us the 10% of the licenses are going to be used in China, then we'd allocate 10% of the revenue to China because that's where those licenses are being consumed. Now if the previous baseline contract with that customer had no design activity in China -- performance in China, then the 10% allocation on the new contract is going to all show up as China revenue growth. But of course, if the previous contract had, say, 15%, and now it's declined to 10%, you might find that, that contributes to more muted growth in China. So all in all, it's kind of -- it's generally easier for us to predict how a contract is likely to grow. But it's a huge challenge to try to predict where our customers plan to use the technology that they purchase from us. So we don't even try to predict the geographical mix of revenue, but we do disclose it because we know that by the time -- we know the numbers by the time we're reporting our revenue. And -- but I think it's really important that people understand that when we report the geographical mix of revenue, it's so our investors can see where our customers are using our technology and it may not necessarily be indicative of purchasing in the region, but it is indicative of consumption. It's the amount of design activity that's happening there.

Okay. I realize I'm asking you all so very high-level questions. But more specific relating to China that process of dealing, engaging and servicing customers who are domestically based in China, has to go through a process, a licensing process where you have to essentially get permission from U.S. Commerce Department to service that customer and prove that they don't have the affiliation with the China military. Is that process becoming more or less onerous compared to when first started a year or 2 ago?

It's a bureaucratic process. There's no doubt about that. The -- I think one thing to understand about our business, I think you understand it very well, Gary, is that a 55% to 60% of our revenue is coming from the top 40 customers. So we do have a high concentration in the bigger names. So in terms of impact, the actual impact on the dollar numbers of our revenue in the region is not significant. It's just -- it takes longer to get a startup to know those customers, I guess, so that you can sell to those startups in the region. And then, of course, salespeople being salespeople, if you've got a quota for the Asia Pac region, and you've got a heavy bureaucratic requirement in one country compared to others, where you're going to spend your time and efforts that -- So that I do think that it's evolved over time. Initially, there wasn't a huge amount of guidance in terms of the amount of paperwork that you had to complete and everyone was feeling their way around it. And there's more and more guidance in terms of the type of kind of putting process around capturing the information. So it's probably got easier from that perspective, but it's not any less bureaucratic, should we say.

Got you. Okay. Well ...

But the only thing to add there that China is still growing pretty well. I mean there were some hits because of some entity companies, you know about that. But overall, I think there's a lot of design activity. And we are in a good competitive position in China. So we'll continue to monitor it carefully. It depends on even bureaucratic work, there is a lot more design activity in China than a few years ago, okay?

Okay. Well, to bring all this home into some conclusion. I want to ask you about where we're at in terms of market growth rate for the EDA in the IP industry. And prior to 2019, we were sort of in this paradigm of mid-single-digit percent growth for the EDA industry. But more recently, it appears to be accelerating to something closer to 10%. And so would you now consider that new paradigm of growth to be the baseline for your business off which you can try to grow organically and inorganically?

So I can take that one. I'm very pleased with our progress across all of our lines of business that we were very, very happy with the level of growth. I like to look at the 3-year CAGR. That's my personal preference for tracking revenue growth. It's because most of our customers are on 3-year baseline contracts, and we published the 3-year revenue CAGR metric in our CFO commentary every quarter. I mean, typically, a customer will enter into a baseline 3-year contract and then maybe 4 to 7x through the 3 years, they'll come back and purchase add-ons and that add-ons typically contribute to growth and then any growth in the baseline contract when it gets renewed that contributes to growth. But we've seen the 3-year CAGR rise from kind of -- it was 7% 5 years ago, it's -- it got up to 11% last year. It's -- I think we'll end this year with more than 11.5% as a 3-year CAGR. And this year, we're on track -- like if you look at our guidance, we're on track to achieve 11% revenue growth for the year. And when you look at the mix of businesses, I'm delighted with the way IP is performing that seems to be a kind of a low teen grower for us based on the way we set ourselves out. I mean we deliberately pick and choose the business that we want to write there. And that's -- we're optimizing there for kind of our rule of 40 metric to -- it's not -- we're not chasing revenue growth for the sake of revenue growth, we're looking to optimize for profitability. But -- and that tends to result in low teen revenue growth. System analysis, as Anirudh said earlier, is doing really, really well that we have a ratable revenue model there, but it should contribute to significant growth for us. And then as we talked earlier about the analog, the custom IC business, that's probably your kind of high single-digit grower because it's a combination of price increases or price improvement and number of engineers, maybe a little bit higher, I'd expect from the digital side because I think we're still gaining some market share there. And then, of course, on the verification side, the hardware, we just launched our new hardware systems at the start of this year, and those have been tremendously popular. And I think their kind of demand is outstripping supply. And you saw, I think gross margins are up to 92%. I wouldn't expect them to kind of stay at that level. Typically, I model for like 90%, 91%. But the pricing environment is strong right now because there's so much demand there. So overall, I think market is very healthy. We're not guiding '22 yet, but we'll talk to you about that in February.

I tried. All right. On the thread of your hardware business and your hardware product cycle, to conclude, you having an easier time perhaps in securing the ASICs and the FPGAs underpinning or central to those Z2 and X2 products.

Yes. So we've -- we moved to a just-in-case, I think, is the way the guys put it, just-in-case approach to inventory management a few years ago rather than just-in-time to make sure that we had the components. And then, of course, we work with some suppliers in terms of building those systems. And it's -- we're building them as fast as we can, and we've ramped up the production. We've taken up as much time as we can with our suppliers to build all the systems that we need to meet the demand that's out there.

Okay. I wish we had more time. I can talk to you guys for hours on end, but we're out of time. And John, Alan and Anirudh, I appreciate and Peter for that matter, I see that he joined. I appreciate all of you joining us today for the fireside, and I wish a good day for the balance of the day. And so again, thanks everybody for joining us.

Thanks.

Take care.