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

Okay. I think we can get started. So very excited to have Cadence Design on the fireside chat. And Cadence, as many of you know, have been long-term participants in the NASDAQ conference here in London. So thank you for the great support. And so today, we have John Wall, who is Senior Vice President and CFO of Cadence since 2017. He is responsible for the company's overall finance, tax and accounting functions as well as operations, enterprise data and analytics, Investor Relations and Global Strategic Sourcing. So I'm wondering what you don't do at the company after reading all of those responsibilities. He also leads the real estate and workplace organization and manages 50 Cadence sites worldwide. He previously served as the controller and in other positions for more than 20 years, starting in Dublin, Ireland before moving to San Jose in 2015. So we also have Richard Gu, who is the Vice President of Investor Relations. So welcome, gentlemen. So I believe we're going to read a safe harbor. Gu, if you're going to do that, and then we'll get started because we want to make sure we're covered up here.

Sounds good. Thank you, Mark. Thank you for having us. So before we begin, I wanted to make sure you're aware of our safe harbor statement. The discussion today 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 our website for a discussion of the risk factors and the use of non-GAAP financial measures. With that, Mark, back to you.

Great. Great. Thanks. And so John, maybe if you just start off and level set our audience, a lot of them have -- know you, but I think sometimes there's new faces, so maybe you kind of level set everybody.

Yes, sure. Great. Thanks, Mark, and thanks for having us here. It's really a privilege for us to be here every June and December. Cadence is a little known company, but often the best kept secret in semis that we're kind of pulled into semis, but we're not really a semi company. It's a -- the company is a computational software company. 85% of our revenue is recurring in nature. The 15% is upfront comes from a piece of hardware and IP business that we do. But generally, it's a computational software company. And when you apply computational software to silicon, that's EDA. When you apply computational software to designing a phone, a plane or a car that's system design and analysis. And when you apply computational software to data, that's AI. So Cadence is kind of embedded in all the things that you can't do without. It was a company created 35 years ago, by engineers, for engineers. Basically, engineers realized if they were going to chase Moore's Law, they needed other engineers to dedicate their careers to building the tools that they would need as they advance down through the process nodes. So the company's culture is steeped in kind of a servant leadership culture, I guess. We spend our lives trying to help everybody do their jobs a little bit better. So you'll find us ingrained into all of those semi companies that you follow, they're probably all using our tools. And we partner with all of them, and we're very, very appreciative of their support and their business. We feel like that's -- we're going through an age where many of our customers, I think time will write them well in history as being like the Renaissance artists of our time. And Cadence kind of provides the paint brushes to these.

Great. And can you talk about how do your revenues typically track versus the industry, should we think about you guys as growing with the industry or better? I believe you've been growing faster. So maybe if you could just talk about how that dynamic works?

Yes. Generally, we grow faster than semi R&D spending that -- our revenue comes from -- I mean so systems companies provide about 45% of our revenue, but 55% is coming from semis. And if you look at semis companies, that some people ask us that, oh, if semi revenue gets cut in half, what happens to your revenue, we often tell them that you're looking at the wrong line in their P&L that our revenue will track semi R&D spending really. But we find that it grows slightly faster than the semi R&D spending. And that's -- I think generally, the growth in human population can't keep pace with the growth in complexity of our technology. So our customers need to leverage tools more and more to achieve what they need to achieve. And that change of spend, I think a few years ago, there's probably 92% of R&D budgets in semis were spent on people and 8% on tools. It is probably 90%, 10% now. And like you say, the human population isn't growing fast enough, even if more of them became engineers to keep up with the pace of growth and complexity. I think as AI tools become more useful for our customers in their design effort, there will be a natural shift more towards -- because more of that spend will be spent on tools versus people. So we'll tend to outgrow the industry as a whole.

Got you. And your top line growth has been accelerating over the several -- past several years despite the pandemics and export restrictions and slowdowns in the industry. And like how do you guys think about your operating model from a growth standpoint from here? Maybe if you could just talk about that.

Yes, sure. We're farmers, not hunters. I mean, we partner with our customers. The -- and essentially, we always take a long-term view. Most of our customers are on 3-year contracts. We spent 35% of our revenue on R&D. So we're constantly introducing new products into the market. Our customers depend on those though and depend on that innovation. We found over the years, actually, there was a period of time from 2012 to 2016, I think where our revenue growth was decelerating. And I remember dealing -- talking with our CEO at the time, we were trying to solve that problem. And counterintuitively, the solutions to the problem was to release new products slower. But because you would think releasing new products at a faster pace is better, which is ultimately what we wanted to do. But that time, we were releasing products, they were too buggy, but we were releasing them slightly early, and about 2/3 of our R&D time was being spent supporting existing products because we launched the product, there will be too many bugs in it, and we'd have -- our engineering talent was spent maintaining existing products and correcting them. And by slowing down -- what we did was we identified some product quality metrics, put it into people's bonus structures in the contracts so that it encourages people to only release products when they were truly ready for a broad market. And by doing that these days in contrast to, say, 2016, it slips, there's probably 2/3 of that 35% that we're spending on R&D today. None of that is contributing to revenue today. It's all on new products introduction, only about 1/3 of that is supporting existing products. That's -- and I think that's helping our growth rates. The EDA has always been a fantastic business, that's -- like you said, we're like the oil in the system. But we didn't really -- it was a great business, but it wasn't clear where we would grow. I think in the early 2000s, all the fabs at the time, all the tailwinds were like social media companies and things like that, which doesn't help us at all. But when you have phones becoming smartphones, cars becoming smart cars, buildings becoming smart buildings and so on, those trends inevitably are more meaningful for us. And we always felt we were destined to be the largest company in core EDA. And that as a result, that's the center of gravity in our strategy. So you have core EDA expanding them into packaging board and then expanding into system analysis, system design and analysis. And ultimately, what we're trying to do with our customers is provide a unified platform that covers view from chip design right through to system analysis. And we're growing with their success, I think.

Great. And then can you -- everybody wants to -- the biggest question I get is, can you help us understand the undiscovered AI plays? So is Cadence and AI play?

Yes, it's -- so we've been doing AI for years, but Richard, thankfully convinced by [indiscernible] we just started calling as generative AI because people understood that. Like we launched Cerebrus as a tool in our digital portfolio as an AI tool, we always talked about that as being reinforcement learning. But essentially, like you said, what we do what we do well is computational software and computational software applied to data is AI. So I think we're very much an AI company. And it's from EDA to AI, if you like, is where Cadence sits. I think you've got a great example. Gu, you want to talk about the horizontal and vertical...

Sure, yes. So thank you. So Cadence has been a long-term partner with big AI platform, software companies like, hardware software companies like NVIDIA, right? You guys probably watch the Computex -- Janssen's Computex keynote, where he talked about the NVIDIA uses pretty much all of Cadence tools and we are the foundation for them. And it's a very mutual kind of relationship, right? We also -- in our most recent earnings call, we talked about how our high-fidelity CFD product can be a great example of like the GPU acceleration. We talked about the 9x acceleration in terms of the speed up and 9x kind of in terms of the cost savings and 17x in terms of the power savings, power efficiency. So -- but I think we think about the AI play from our lens, horizontally and vertically. But horizontally, we are definitely there, supporting and empowering all our big partners and customers in the AI journey, right? So it doesn't just include NVIDIA. It's all the other AI companies such as Marvell and Broadcom, AMD or the other companies. The beauty about our business is we're sitting nicely in the ecosystem. We manage and support all of them, right, all of our customers in terms of supplying them with our software and hardware platform. And very clearly, there are many, many good places because ultimately, when you look at AI, it's going to be a verticalized play. And HPC hyperscale is a great example of that, right? As you look at the past couple of years, companies like AWS, they've had resounding success in terms of carving out that custom design chip and journey. And this is just a starting point. So AI will be a great next chapter and other companies will follow through because they own the application and software, can platform themselves. And it also could blend into and support other verticals, key verticals like digital biology, right? I think in that aspect, what Janssen thinks about and what [indiscernible] vision is, it's super aligned. So very excited about the AI journey. We believe we are in the -- right in the center of kind of -- exciting kind of journey for our customers and ourselves.

That's a great example of where we partnered very closely with our customers. But I mean, we use GPUs in our Open Eye business for where we're applying computer science and math to chemistry and biology. And then, of course, those GPUs are designed using our tools. So there is that symbiotic relationship, I guess, to a certain extent. And then as Richard said, there's the kind of horizontal and vertical view of things. but also with us spending 35% of our revenue on R&D, there's an internal use of AI as well as external. So externally, we're selling those to customers. But internally, we're using our own AI tools. We eat our own cooking. We use our own AI tools, in the development of our tools in our business. A good example would be, say, our Palladium hardware system that we use for emulation. That's a great example of Cadence on Cadence. So Cadence tools are used to design a custom silicon chip to use in a hardware system that allows you to emulate a chip. And it's very, very high. I mean that builds a wide motor on the business because it's hard to compete when you've made that chip internally using your own tools. So the next version of that when that comes out, that will, of course, use our AI tools as well.

Got you. So you are using your software on GPUs to design GPUs. It feels like we're approaching the singularity, right? So -- okay. And so like -- so you're using AI, right, and at the risk of making this a superficial question, but I will make it anyway. Like have you quoted or quantified X percent of our revenues are AI. And so therefore, we are an AI company. I appreciate that you have all your tools are embracing the technology for a while.

We haven't done that yet.

Okay.

So we're still in kind of a price discovery mode in how to price for those tools. We think it will take 1 or 2 contract cycles. We are contemplating packaging them in a way that is easier to quantify the AI tools from everything else. But we're tightly into woven into the entire design process.

Got you. Okay. And so -- I'm sorry.

We have quantified it with some customers, though, is if you take, say, a customer that's spending $10 million today on R&D in semis, that $9 million, that's probably on people and $1 million on tools. And if they just go naturally to -- by the time they doubled to $20 million, it will be $18 million on people, $2 million on tools. But we feel that if they adopt AI tools, that could go -- you probably get the same efficiency for 20 -- you'll achieve the same for $16 million with AI tools, then you can achieve for $20 million without this. And -- but we think at $16 million, you probably should be spending $4 million on tools and $12 million on people. So we think that over time, it will go from $90 million, $10 million to something like $80 million, $20 million or $75 million, $25 million. So from that perspective, directionally, that's the way we're thinking we have a pricing board that prices the tools because we sell licenses, individual licenses, and we're trying to track to sharing half of the value with our value creation with our customers. But it takes typically a couple of contract cycles for us to get that right.

Got you. Okay. And so you -- and you talked about this in your introduction, you're not just helping semiconductor companies, you're helping system companies also. And can you talk about that evolution and how that has impacted your model?

Yes, sure. I mean the -- again, it's applications of -- the application of computational software to different things. Like I said, EDA was always a tremendous business. The challenge was where do you go? And I think Anirudh's a visionary. I mean, he laid out a great strategy for us so that we could apply computational software to, like say, to the design of complete systems like cars or phones and it's the same. It's the same techniques. It's the same computational software. It's just applied to the entire system design and analysis. So what we find our customers use our tools for chip design, verification, packaging board and then the whole system analysis. And for years, we've built out a packaging business, and that's the natural bridge between EDA and system design and analysis because you see more and more companies using chiplet design now, they're stacking more chips on top of each other. And just the nature of doing that causes more heat and thermal issues, but -- which you resolve with system analysis tools. So it's a natural kind of progression for us. And again, like I said, computational software applied to silicon is EDA, when it's a play to products like planes or cars or buildings, its system design and analysis and when it's applied to data, it's AI.

And can you talk about like what percentage of your revenues are what people would understand is the classic chip design versus the system design and analysis? And how those factors are growing?

Yes. So system design and analysis now.

Yes. So system design and analysis right now. This year, we're expecting to have about the size of the business about $500 million, and it's growing at about 27% for the last quarter. And so it's about like 12% of our overall kind of revenue base. Majority is still EDA, but system design and analysis definitely is a faster-growing part of the business with higher profitability and margin opportunities as those continue to scale.

That's to do with the area for us. So it has been the fastest core for a number of years. And there's no let up on that in the short term.

Got you. And can you talk a little bit about the hardware part of your portfolio? Who uses it? Why has that been so strong?

Well, it's quite interesting. I mean -- so there's like a secular trend in demand for verification that's -- nothing is getting less complex, should we say. And our verification team always tease our analog designers and our digital designers because they say that you only have to cope with maybe doubling the number of gates on a chip with the number of transistors on the chip. And on the verification side, they need to -- whatever the number of gate improvement is, they need to verify 2 to the power N, so to speak. But so they say their job is multiples [ harder ]. But generally, what we see is the time line for new product introduction has shortened many of our customers. Many of our customers want to have a product launch every year. And if you're doing a product launch every year, it really tightens the whole time line for everything for the design phase, for everybody involved in the process. So your chip designers need to have -- they need to have a virtual chip available for the software designers to develop software for before you have a physical version of the chip. So what they'll do is they'll -- so in some cases, they'll use a prototyping system to bring up the virtual version of the chip and the software people will design their apps and their software to run on that chip, while that chip is sent off to the foundry to be stamped up. The -- if you're using an emulation system, the chip's not hardened yet. So that's kind of distorted the natural flow of -- the natural order of things, where software people are often big champions of ours that they want the chip designers to use more of our emulation systems because they want to collaborate more closely with their chip designer colleagues. So when they're designing some certain software, they ask for a chip to be tweaked. And with the emulation system, you're able to do that. And then when you have your chip finalized, then you will transport it over to a prototyping system and free up the emulation system for more collaboration. So allowing our customers, engineers on the chip design and software design to collaborate more closely together, we provide that benefits them. So that's been huge for us. I mean the demand there for the last 6 or 7 quarters has been off the charts. We haven't been able to keep up. We keep building more and more systems. We ramped up production at the start of this year. And you see that that's in our guide for this year. But Q1 was a record hardware quarter for us on the back of that.

Yes. And it seems like more and more semiconductor companies are adding software capabilities to their own portfolio. So that makes a lot of sense. So profitability for the company, how should we think about long-term operating margins? I believe you've talked about 50% incremental as a bogey going forward. How should we think about that?

Yes. I mean again, Cadence has been a fantastic business, and EDA has been a great business, and what we discovered is the scales really, really well. Our general model is we -- by what we commit to the Board is we're aiming for double-digit revenue growth, to try and turn that into mid-teen operating income growth. And ultimately, with using 50% of our free cash flow to buy back shares, we would hope that, that then translates into high teens or low 20s EPS growth. And we published a CFO commentary each quarter when we produce our results. And if you look on an annual basis from -- certainly from 2016 forward, our 3-year revenue CAGR continues to expand. We were in the -- we started at the [ 8s ]. We're up into low teens now for revenue growth. From an operating margin perspective, on a 3-year CAGR basis as well, looking at the growth in operating income, that's been up in kind of low 20s, sorry, mid- to high teens and then EPS growth has been in the low 20s consistently on the back -- like we look at things over a 3-year period because most of our customers have 3-year baseline contracts with us and they purchase add-ons throughout the -- through that cycle, but add-ons typically coterminate with the baseline. So every 3 years, there's a complete refresh in a new 3-year baseline contract. So a lot of our people are measured on how they grow an account through a complete contract cycle. So we typically measure EPS growth, income growth and revenue growth on a 3-year CAGR basis. But the model has worked really, really well that we dropped over 50% of revenue growth into operating income every year. And we've done that consistently for, like say, last 6 or 7 years.

Great. And it looks like we just ran out of time and maybe, John, I'll give you a last word. Is there any final message you want to leave investors with?

I just think that we're not as well known as we should be. We're farmers, not hunters, but you won't see us -- I mean, if you're trying to pick the winners in AI, they'll all be customers of ours. But -- so generally -- generally, I mean it's a pretty safe bet in Cadence. You cover AI by investing in us. Now if you can pick the winner, you're better off picking the winner. But if you want to kind of put a broad swipe across the industry with all these technology companies, semi companies, systems companies and AI companies, you kind of have more covered with Cadence.

Great. Well, John, Richard, I'm sorry, we ran out of time, and I'm actually hosting the next fireside chat, so I would infringe on the next one, so we're going to cut it there. But I'm sure you can meet them in the call. John and Richard, thank you very much for supporting the conference and for the great presentation. Thank you very much.

Thank you.

Thank you, Mark.