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

Delighted to have the team from Cadence join us, John Wall, the Chief Financial Officer; and Richard Gu from the Investor Relations team, Richard said, that he will make an exciting announcement before we start, and then we will get on to the Q&A part of the session. So welcome, John. Welcome, Richard.

Thank you, Vivek. Today's discussion will contain forward-looking statements, and we'll make use of non-GAAP financial measures. Please see Cadence's most recent 10-K, 10-Q and our website for a discussion of factors and the use of non-GAAP financial measures. With that, back to you.

Very excited. Thank you. I appreciate it. So maybe would be good. I think to just to get on the same page, right? You reported very strong results. Recently, there were some questions about strength and so forth. So maybe we can start with just kind of the near term and then get on to, I think, the more important kind of longer-term structural advantages of the company.

Yes, sure. Great. So Cadence, I mean, it's a computational software company for those of you that don't know us, it's fundamental technology that you need to design any kind of microchips or electronic systems. About 85% of our revenue is recurring in nature. The other 15% comes from a small amount of hardware and IP that we sell, like I said, mainly a computational software company that has significant cash flows as we -- revenue just -- the guidance right now is $4 billion and $50 million for this year for revenue, 41.5% operating margins, and we've generated 50% plus incremental margins over the last 6 years consistently. But every dollar of revenue growth that Cadence tends to flow through in the kind of mid-50s of a $0.55 of every dollar revenue growth drops to the bottom line for us on a non-GAAP basis. And we spent about $0.08 of those dollars on share-based comp. So I mean it's a great business. our customers are on 3-year baseline contracts, and then they'll purchase add-ons throughout the throughout the 3-year duration. And when they purchase add-ons, the add-ons co-terminate with the underlying baseline contract. So typically, a customer will come back between 4 and 7x and purchase additional technology over the 3-year contract. As a result, the average duration of our backlog of bookings in any 1 year, typically ranges 2.4 to 2.6 years, averages about 2.5. And that's because you have the 3-year baseline and every add-on is less than 3 years because they terminate. And then we see the -- in terms of booking timing, because of the nature of our model, the timing of bookings doesn't matter a great deal. It's the annual value of those bookings because we're always trying to grow the annuity, the annual value that we're extracting from each customer. Our top 40 customers at Cadence, we generate 55% to 60% of our revenue from just those 40 names. And the top 40 is a bit like the Dow. It changes all the time. So the top 40 for this year will be very different. I mean there's maybe 50% of those names were in the top 40, 10 years ago, but it was still generating 55% to 60% of back then. But -- so we tend to win with the winners. We have strategies to grow our top 500 accounts. So we have a very broad coverage in terms of having individual strategies to grow the accounts. And our focus is mainly on profitability more so than revenue growth. But generally, we aim for double-digit revenue growth, which with improving margins tends to turn into kind of low to mid-teen operating income growth and then we use half of our cash to buy back shares, which results in EPS in the high teens low 20s every year with strong cash flow.

Very nice. So one thing I like, John, is in your CFO commentary, you gave that kind of 3-year view of right of the trend, which is, I think, the right way to look at the business also as you just alluded to. So when I look at that trajectory, Cadence used to grow at 7%, 8% CAGR, then it stepped up to 10%, 11%, and now it's kind of in the mid-teens, so walk us through what has led to that acceleration of growth? Because when I look at the broader semiconductor industry, that has not accelerated in the same way. So what has driven acceleration in your business?

So we've been very data-driven. I remember -- so I've been with the company a long time, 26 years with Cadence. I took the job is like a 1-year, 2-year gig, thinking I'll go find something else. Nobody knew who Cadence was then, they still don't know who they are. But -- so 26 years there, but I moved to California in 2015. And when I arrived that we were at the sales kickoff and my CEO at the time was Lip-Bu Tan, Lip-Bu was on stage handing a blank check to the head of our IP business. And blank checks scare me anyway. But I was a bit freaked out by the blank check and I went to talk with them afterwards and I just asked what's the problem? And most of my conversations a Cadence start with, what's the problem you're trying to solve and the problem he was trying to solve was that revenue growth had been decelerating because revenue growth, if you go back from like the period 2012 through 2016, Cadence revenue growth, I think, was going like in 12%, 11%, 10%, 9%. And at the time, he was a bit worried because I think our forecast was 6.6% revenue growth. And he was like, we got to turn this around, and he was, I think, looking at what Synopsys was doing and thought, well, their revenue growth was mainly coming from IP, and we thought that was a good opportunity. If we want revenue growth, we'll do that. So that was the problem we were trying to solve, revenue growth. So I just dived into the data, pulled a bunch of data with my team. And what we looked at was -- and this is the most unintuitive thing about Cadence. Although we're spending -- back then, we were probably spending close to 40%. But generally, we spend 35%, 38% of our revenue on R&D. You won't find many companies spending that amount on R&D, but it's a very innovative space. But -- and what I found at the time was just a rough straw poll of the leading engineers in the company. I found 2/3 of their time was being spent on products that we're generating revenue today and 1/3 was being spent on products that didn't offer any revenue in the current year. So for me, I wanted to bifurcate between how much of your time is maintenance, how much of your time is new product innovation. And then when I looked at why they were spending so much time, the reason that we're doing that was our customers are the greatest people in the planet. I mean, they're artists. They take our technology and they design all of this wonderful technology that we can't live without. The -- and when we have a solution to something many of our customers, they don't care if it's -- if it breaks down every 30 seconds. If you have something close to a solution, they'll take it the rest of the way themselves. And our engineers trying to keep customers happy, we're too quick to want to release something. But when you release it too quick, you've a lot of bugs. Customers are -- not all customers might have the same talent level that customers are upset. Our engineers spend a lot of time doing bug fixes. So what we did was we started capturing things like customer call center volume, customer change requests to try and identify which are those products that we release too fast. And when we identified those, the crazy answer I went back to Lip-Bu with was, I said, Lip-Bu, I think I have the solution to your problem on how to turn around revenue growth slow down product introduction. . And we did it like he trusted the data that we slowed down product introduction so that the engineers weren't spending time after the product release, maintaining those products or dealing with the bug fixes and everything, and they could spend more of their time doing new products, innovation. And so back then, if it was 2/3 maintenance, 1/3 new product innovation, it's flipped the other way now. It's probably 2/3 new product now. So despite the fact that we have 41.5% margin guidance for this year, with probably 35% of our spend is in R&D. 2/3 of that is not contributing a dime to that revenue this year. But it's all generating revenue in the future. So once we turn that around, and we started increasing the pace of new product introduction, then we saw revenue growth come with that. And then you've seen that, we track a 3-year revenue CAGR because we're looking at accounts. Sometimes you'll have accounts where it's hard to get the customers see the value initially. So we'll try to get the -- proliferate some of the technology into the account, maybe for 1 or 2 projects and then leave room for customers to come back and purchase add-ons later. So we did that, and you tend to get that growth that follows through. The other thing we did was we established these deal quality metrics. So -- and again, very simply, we just did a calculation to figure out what's the average selling price for every product. And then we compared what a salesperson was getting on each account. So whatever your configuration was, we knew what the average selling price was for the annual value of those products. And then we could figure out a deal quality metric. So if the annual value that everyone is paying us, the average price was $10 million a year, and you are buying from us from $8 million, I would look at the sales guy, they're selling to you and say, okay, you've got a deal quality metric of 80% here. It freed the sales guys like tell and teachers, we're judging you based on the grades of your students. They would say that's not fair. But what is fair is how does the grade improve while you have responsibility for the account. So once we establish that, okay, the sales person, you have an 80% DQM. The beauty of this was you could look at one account, you could look at one product, you could look at a portfolio of accounts. You could look at regional VPs that have thousands accounts. And as long as you were measuring the accounts they had in their control over a defined time period, you could tell which ones -- whether they were improving disimproving. And then when we did our individual strategies for each of the accounts, we looked at -- we started with the top 100, bear in mind, top 40 is 55% to 60% revenue huge coverage. And we did what we called the 4Ps. But again, trying to keep it simple for everybody. What we did was we looked at the people that were on the accounts, the products that were in the account and the profitability in the account. And then what I asked them for was the plan to double profitability on the account in 4 years. In some cases, they came back and looking at the deal quality metrics, customer was already paying more than average. And we said, well, we need to play defense here. In other cases, customers are paying too little here, you need to be more aggressive and play offense. And just following the data and knowing whether you're playing offense or defense, I know it sounds silly, but that's the type of stuff we were into 6, 7 years ago, but we've adopted those -- that strategy and there's a lot more discipline in pricing. There's a lot more data driven approach to everything now. And what we've seen is that if you follow the 3-year CAGR, like you say, we're -- we've gone from 7%, 8% up to mid-teens. And all the while, we've done that improving operating margin and improving operating cash.

So is this mid-teens growth rate sustainable, why or why not?

So again, I focus on take-on pay rather than gross pay that someone can tell you that your gross pay taking home less you won't care, right? But so there's a lot of growth out there, and we have these new AI tools that we're launching. I mean, there's huge potential for growth there. We think that will turn up over maybe 1 to 2 contract cycles though, because the important thing at the initial stage is going to be proliferating us. But when you look -- our model generally, when we talk to the Board, when Anirudh and I talked to the Board, what we committed to them the double digits. We want to...

Double digit could be like 10 or 99.

Well, it's -- I would rather double digit more profitable double-digit revenue growth than -- like there's a lot of business that's available. You could pretty much print any IP revenue you want at the moment. I mean there's companies willing to outsource their IPT. I have a team of engineers that cost me $10 million for my own in-house IP, you guys take it on up here $20 million now for the next 3 years, saves me $10 million. But you get $20 million revenue immediately pretty much because you've already delivered the IP, but you're signing up to expense a perpetual stream of expense. We don't do that well. So we steer clear of that type of business. But we're -- we always describe kings as being farmers, not hunters that we're better farmers than hunters. We basically plant now and harvest later, right? So we will invest in R&D despite the fact we're spending 35% of our revenue in R&D, 2/3 of that is not generating any right now, we're planting now for harvest later. But we want to do things now that allow us to harvest cash flow and revenue into the company later. With some IP business, it's an offer, you can do the opposite. You can take revenue now, sign up to the expense over a period of time. But that's a hunter mentality and it doesn't suit us. But -- so we tend to focus on more profitable activity. What I'd like to do is measure the operating income per engineer including the cost of stock because everyone -- a lot of our engineers I guess share-based comp. Now if I compare -- like, so if we go back to 2017, it was first year I became CFO. The -- our competitors Synopsys had like 40% more revenue than us and 33% more operating income if you include the cost of shares. The -- so when you look at it by engineer, I think we were generating 115% to 117% of the operating income per engineer that they were. And that was because we had dedicated our engineers to more -- to higher margin activity like software development as opposed to IP or services. That's 5 years on, if you compare '22 with 2017, the gap has widened on revenue. We were 40% behind. We're now 45% behind. I'm proud to be 45% behind. The -- on the operating income side, it went from 33% behind to probably 3% behind. And right now, we're generating probably 175% of the operating income per engineer than our competitors because we're very data drive because we're focusing our attention. We think the scarce resource is engineers, and we need to focus that attention on the highest profitable activity for benefit of our investors.

So when you look, John, at the new product pipeline, and your investments, then do you think this kind of mid-teens growth rate is sustainable?

It's a trick. I mean, we're not guiding for next year. I think if you look over a longer time horizon, like if I look at the AI tools that we've launched, and we have a suite of AI tools now across all of our products. When you look at the benefit that, that provides that you're a typical semi company, if they're spending $10 million on R&D, they're spending $9 million on people and $1 million on tools. And what I would tell another CFO is that, okay, if you don't do -- if you don't change anything, at some point, that $10 million becomes $20 million, you're spending $18 on people and $2 million tools. But if you spend more with us, we can save you more. If you spend more with us by the time you get -- you can pay $16 million to get that $20 million of value if you spend $4 million with us. But -- so I do think the potential for someone that's spending $10 million in R&D now, $9 million on people, $1 million on tools, might only have to spend $16 million to get double the value but they'll have to spend more on tools because you'll leverage AI to get more efficiency. But it really is the pace at which that we can get that adopted. It will be certainly slower than NVIDIA, right? NVIDIA is I mean, tremendous been great partners. I don't know if you saw the...

The announcement, right?

Yes. So we've always been very close collaborators with NVIDIA. And we generally follow where they lead that like when we do -- our emulation systems are super popular. We would collaborate with NVIDIA or a good partner of ours to help us develop those systems. But when you have, say, NVIDIA succeeding with the GPU, what tends to happen is our customers will use the GPU, they'll have to broaden it, they run a bunch of projects. They'll see value in a particular niche market. And then when they see that value, they'll look at how do they optimize for that and they probably create a custom chip and that the solution to drill in further where there's opportunity for value is to create some custom silicon. And when you need custom silicon, you're going to need tools from a company like Cadence. But -- so generally, their success tends to lead more custom silicon. And the more democratization there is a custom silicon development in the more startups, design start-ups that you have that generally is a leading indicator for revenue growth for us. But like I say, I mean, some customers will adopt it quicker than others. I mean I have described you the sum is. I think our pricing power gets better over time. That -- the best analogy I can use is, if you go to a Warriors game, and you stand up for a minute, your view of the court is amazing that -- but then, of course, if everyone else stands up, obviously, as have used this with me as well. But if everyone else stands up, what your view is back to the same as it was, you just all a little bit more uncomfortable now. And obviously as will tell me this, so I'll pay for AI tools, but they never want to pay for AI tools, and I'll be no better off competitively, but you guys are getting all the AI dollars, right? That the thing is a couple of contract cycles in imagine being told to sit down. We don't want to sell you AI tools any more. That the pricing power you have with customers kind of builds over time because the more dependent are on tools and the technology and the more sticky those tools become. So I think it's -- we have like a partnership relationship with our customers. We certainly don't price out to anybody. But we do look for a fair return on the investment that we make. Our strategy normally, I mean, if someone doesn't want to pay like if someone is moaning or complaining about paying extra dollars upfront for the benefit, we believe that they'll derive. Typically, we will offer to invest alongside them. But if you don't want to invest, if we can afford to invest more than 80% on your R&D, maybe we'll invest another 20% with you, but we have to share the outcome. But if we invest alongside you, we share in the outcome in some kind of bonus element or royalty -- and once you take it down that path and they realize how much you want to get paid for investing alongside them. They bring the discussion back to okay, well, what if we pay you the price a bit more on price. That's -- so the pricing discussion is often take that path. But over the course of contract cycles, we just become stickier and stickier.

Is there a simple way, John, to quantify the AI benefit to Cadence? Have you already seen it? Because I imagine that all the new products, right, whether it's the TPUs or the H100 GPUs or what have you, they started to get designed a few years ago. So are we already seeing AI benefits kind of run rate in your business today? Or you think that benefit can drive further acceleration in your business over the next 3-year cycle?

There are certainly huge benefits that we've seen. I do think it's project by budget base that's, I think, like take the H100 as an example, in 9x faster speed up?

Yes. So I think I think the way we think about the [indiscernible] from AI is both horizontal and vertical, right? Horizontally, we're in the midst of supporting all the AI companies, right, like infrastructure, platform companies like NVIDIA, NVIDIA and Broadcom and all these different -- all these different companies. And importantly, also on a vertical level, we have a products, right, all woven together ready on one common giant platform on an open source basis, right? So all these AI products, obviously, they command a premium for pricing. And also, over time, it's going to drive adoption curve and driving more quantities. So the way we think about AI, we feel like this is a great thing, great tailwind for our business for many years to come.

But are the benefits still on the come? Or do you think they are represented in your current growth rates?

I think it's just -- there just getting started on this. When you think about the H100, I think over time, you're going to see AI chips because it's not just compute, right? It's compute, networking, storage applications as those start to proliferate and disrupt these different industries. I think you're going to see more and more of activities in that front. I think we as a company sits beautifully in the midst of all these activities and driving and supporting our customers on their AI journey. I think it's a very stage for our growth.

Yes. Got it. One very kind of near-term question, John, which is how is the visibility around the second half of the year, right, when you're supposed to start building up, right? Your bookings and your backlog.

Yes. So the annual value of our bookings has been increasing steadily. The -- we did some record contracts with very, very large customers first half of last year. And of course, they don't come up for renewal for another few years. So you've got -- you have some kind of headwind on RPO as a result of burning revenue on that every quarter. But the first half of this year, we didn't have a lot of contracts expiring. So you didn't have a lot of renewals happen in the first half of the year. But from a profitability perspective, an annual value perspective, anything we did really renewed at a higher annual value than before. And that drives some growth. But look at -- I think in relation to the second half, typically, I hedged the second half for hardware, particularly. Our hardware emulation systems are super popular. The demand has been off the charts the last 6, 7 quarters. The -- we ramped up production capacity last year and increased prices by 20% and still ended up with a longer lead time in backlog at the end of the year than we had at the start. So we've pretty much doubled up capacity for production capacity at the start of this year. And Q1 was significantly high revenue record quarter for hardware. But a big part of that was because every system that came off production when we went to a customer that had purchased this. And when you deliver those, you trigger up -- we get immediate revenue on delivery of the hardware, the maintenance piece gets taken over time, but the hardware gets taken immediately. Now when we produce hardware we typically have 4 constituents that are waiting to receive it. The biggest one are customers that have purchased and waiting for delivery, and that triggers upfront revenue. We also have our own hardware cloud infrastructure. We put it in the cloud for smaller customers. So the price of this hardware is a bit like a big Ferrari or something. But -- and if you don't have a use case for Ferrari all year round that -- and you might only need it for a month, you'll rent it. And typically, people do that in the cloud. There's premium pricing for it, but that has been starved of hardware for a while because there's been such long lead times from the customers that are purchasing. Then we have hardware systems that we go out and seed future business from, like if a customer has the ability purchase one, but hasn't bought yet, that we will maybe put a system on site for them for 1 or 2 quarters and then convert that into a sale later. And then there's a -- there's our internal usage requirements as well. But now in the first quarter because all of the hardware systems that came off the one to kind of eat into that lead time up. We sold systems in Q4 in December, even though we told customers, you might have wait 6 or 7 months for the hardware to be delivered just that much of a backlog. And people were still buying them. We're still buying them and prepared to wait. And then most of the time people -- you'd only see the opportunities turn up in the pipeline about 3 to 6 months before a project starts. So before we tell them wait 6 or 7 months. We're pretty much asking them to delay a project to use our systems, they should go off and buy something else, but they were still waiting for our, still willing to wait for us, which is great. But that's not something that's sustainable. We have to ramp up production, but we want to eat through the lead times. We did that in Q1. In Q2, more of -- we'll produce the same number of systems in Q2. But some of those systems will go to set up second half activity. It will be demonstration machines. There will be some cloud infrastructure machines. We will -- we're trying to get down to like 13 week lead time for sales that happened at the end of Q2, beginning of Q3.

What's the lead time on hardware right now?

So it was, like I say, it was 6 to 7 months in December trying to get that back to 3 months. It used to be -- the lowest it was before, it was like about 6 weeks. I don't see it ever going back there. That's -- I just don't think it's possible. It's -- it would be nice to get it into the 10 to 13 weeks, but right now, we've targeted 13 weeks.

So when you say you have a conservative outlook on hardware for the second half, does that mean like flattish half on half, does it down half on half, part of the outlook...

The thing that was hard to predict for the second half is that when you have long lead times like that I know how I reacted when there were supply chain challenges that if you wanted to or something and you found all of the year, quarter 4, right? But -- so we had huge demand a big backlog. I didn't know how much of that -- will that demand -- when we start meeting all of that demand, will that -- will we have taken some of second half demand forward or not, I don't know. But the other thing that I was cautious of was that with every system coming off the production line going to customers that had purchased it, we're selling a lot of hardware to customers that already have 10, 12, 15 systems that those new adopters that we typically put out a demonstration machine for a quarter before converting it into a sale for the last kind of 5, 6 quarters, there's not been a lot of opportunities put them on machines out there. We want to get to the point to have those demo machines out there so that trigger demands in the second half. So I mean, we've done all the things that we need to do to improve the second half.

But as part of the outlook that you have already given, do you think hardware has -- what is reflected in that outlook that's been already is hardware flattish half and half? Or is it...

Oh, no, no, no. First half is -- it's much higher than been second half in our outlook. And there is definitely room to take up the second half with more hardware demand. The tricky bit is how long will it take to get those -- the hardware installed. We really have to eat into the lead times. But if I can eat into the lead times, get some demo machines out there, there's is room to take up the second half.

Got it.

But -- I mean we'll tell you more about that in July.

Of course.

Yes.

Do you think industry has kind of addressed this China restriction risk adequately. Like when we look at the last set of restrictions that were put on right? I think they were fairly benign. But to the extent there were restrictions, were you informed 3 months before, 6 months before a year before, like how long of early on system do you have?

We were informed within a year, but we did have -- we had a subpoena on our China business because the government wants to understand what we were selling and to who. And thankfully, it was very straightforward because our sales into China are basically, you have a contract or a product at an invoice and then delivery, it's all very, very straightforward. There are no joint venture structures or anything like that. But -- so it was all straightforward that they came in, they look at that, they were trying to understand what our -- what our customers were using for. I think what we learned from the government was they were very focused on military end use capability. And I know you say the activity has been benign, probably benign to us, but I wouldn't say so is benign to say equipment manufactures. But they seem to be focused on equipment manufacturing mainly. They seem to be happy with design activity as long as the design activity results tape out outside of China that you take it to Samsung, take it a TSMC or Intel or whatever, Global Foundry [indiscernible], didn't want the activity happening in China. But -- so it seems that -- I mean our customer 1,500, 1,600 customers in China now, a lot of design activity going out there that most of this is getting taped out. It's probably TSMC.

I see.

[indiscernible].

I won't claim to understand the workings of the government what the motivation is behind this. I do know that like on the design side, what we've seen is some restrictions on being able to help lower process now designing and manufacturing in China. But then we saw that GAFA. There's a restriction on this GAFA technology really applies at probably like 2-nanometer design at TSMC. Now the good news for us, I mean, we've always been telling the government if you're imposing rules want us to stop selling certain technology, tell us plenty of time in advance. Don't tell us about something that we made 10 years that so with GAFA, we have new products that are -- it's kind of like diet coke and regular coke that you can have. There's a version of the product that has a GAFA in it. There's a version of the product that doesn't have GAFA in it. So that we're able to at least service both markets. I mean we we're free market people. We would like to -- we service all of our customers as much as we can within the boundaries of the law.

Terrific. I think with that, we have run out of our time. But thank you so much, John. Thank you, Richard for your participation. Thanks.