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

Okay. We are finally live. Thanks for bearing with us. I have a pleasure today to introduce the next company at our CEO conference. And it is my great pleasure to introduce Cadence. From Cadence, we have Alan, who heads up their IR department, who's going to start with an opening statement. But also, I'd like to welcome Anirudh Devgan, who's the new Chief Executive Officer of Cadence; and also John Wall, who everyone knows well is the Chief Financial Officer of the company. Should be a great discussion. So Alan, the floor is yours.

Okay. Thanks, Gal. Before we begin, we'll mention the safe harbor statement. Today's discussion will contain forward-looking statements, and we'll 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. Now I'll turn it back to you, folks.

Well, listen, thank you. So to start, I'd like to kind of just focus on the exceptional strength of the market that you guys play in and how that has ultimately transposed into this double-digit growth that you consistently delivered over the last 3 years that kind of has not been the case in the past for Cadence or for the industry as a whole. So a few things to clarify first is when Cadence plays into the market of semiconductor design effectively, one of the big misconceptions that I come across still is the fact that your business is somewhat closely linked to the semi cycle itself. And Anirudh, maybe that's one of the things to start with is just kind of the thing. How much of that is true versus how much of that isn't true, but it's linked more to things like design starts, the research activity that people are doing. Where does your core business growth actually has come from? And why has it been kind of stronger over the last couple of years versus what we've seen in the past. What's changed?

Well, hi, Gal. Great to be here, and that's a great question. So we are primarily on the design side, as you know. I mean, we are linked to design our semiconductors and electronic system. So first thing I would like to say is that about 45% of our revenue is now coming from system company, okay? So -- and so it's not just -- of course, semi, we love all our semi customers, and it's a great business. But we are increasingly working with system companies, either with new products or they're doing more silicon themselves. So that's number one. And number two is that the amount of silicon being produced and the electronic systems being designed is, of course, going up. And in the past, maybe if you go back to the '90s or even early 2000, there were these like one main driver, whether it was PCs and there was mobile, whereas now we have talked about there, I mean, you already know, there are multiple drivers, 4 or 5 big drivers. And I think the need for electronics systems and semiconductors is going to go up. And there's a lot of design activity to make that happen. So therefore, there is some smoothening of that and -- versus maybe 20 years ago. And then the third thing is we have more -- we have TAM expansion. We have more products, more products that we sell at the system level. So -- and then IP is growing in software products. So that also adds to the revenue growth. So that's why we have now our 3-year CAGR is more than 11%, which in the old days, EDA used to be 6%, 7%, 8% business. And also, we have worked very hard in the last few years to improve our margin, operational efficiency. So like the latest guidance, we have our revenue growth of 11% and margin -- a non-GAAP operating margin of 37%. So we want to drive both those things going forward here.

Got you. So from what I'm hearing, the most important things that you mentioned are the fact that in the past, you've gone through cycles where there's been an important driver of growth. But usually, they were limited to 1 or 2 like smartphone was definitely a big cycle that helped you before it was a DC cycle. But now we're talking about the cloud computing. We're talking about the adoption of AI and machine learning in terms of the need to increase the power and performance of the chips, not just the area side of things. And then obviously, also when we think about the electrification of the things, right, Internet of Things, including smart cars and everything that brings with it, right, putting supercomputers on the wheels, as we may, right? So we're hearing that you believe that what we're seeing today is just the beginning of these trends that are all kind of happening at the same time, and it's the trend of digitization. So that to me sounds that it's more of a lasting trend because what lasting effect because of the fact that the trends are more diversified into other areas compared to what you've seen in the past. Would you agree with that?

Yes, Gal, I think that's a good way to put it. Even earlier, I think we were somewhat shield -- I mean, of course, EDA in the past 20 years ago, we'd go up and down because of the semi. But we are always somewhat shielded because we are on the design side, but still, it will do that. And also, the industry was not very mature. There were a lot of players. Pricing discipline was not there. Now our market has also consolidated, and we are the most broad supplier at this point. So that's the other thing. We have analog. We have digital. We have verification. We have system analysis. We have packaging and 3D-IC. So that's one -- so because as people do more electronics and digitization and systems, it's not just the advanced node that's critical, which it is like cloud and high-end smartphones. And we do really well at the advanced node. It's also all the sensors and the mainstream nodes and then all the memory. 1/4 of silicon is DRAM and flash and all. And so Cadence has a unique portfolio across all these things. And then on top of that, like you mentioned, the market of semiconductors and electronic systems will grow. But underlying this, I think there are some Cadence-specific things and industry-specific things, including consolidation and then Cadence being -- having the best product and a broad portfolio. So that's also helping us.

You mentioned a really important point that I don't think we touched on enough when we speak in this forum. And it's the fact that consolidation of the industry and also focused not just on the leading edge customers, but also on the customers that are effectively systems companies that can be huge, right? But tend to be -- it doesn't -- it's not their main priority to develop a chip, I guess, to develop a system and sell that product. And then as a result of that, you mentioned something that's really come across is better pricing environment. Would you agree then with the statement that 5 years ago, it was the pricing discipline that was kind of a different -- there was almost a detriment to your growth and margins, whereas that now has changed because the growth is a lot more balanced? So you feel, looking at John then, feel better about double-digit growth prospects going forward, but also then being able to generate sustainable high margins incrementally.

Yes. I can chip in and take that, if you like, Anirudh.

Sure. Go ahead.

Just on that, I definitely think the pricing environment has become much more rational over time. But when you look at the overall landscape, the thing that's been key to Cadence over the last 5 years has been at the roots intelligent system design strategy, opening up our TAM. EDA has always been a tremendous business. The challenge in the past was where do you grow? It's a sticky business, quite diversified, maybe a bit too exposed to semiconductor companies. And that -- I think what we looked at was how do you expand? How do you expand your TAM? Where do we have the right to win? And when we looked at the stages that people go through when they're trying to develop an electronic system and they're doing chip design, chip verification, packaging and board, system analysis was a natural adjacency and an area where we already had tremendous expertise in terms of simulation expertise. And we were focused. Anirudh and I sat down, what it like back 2016, 2017, and we talked about using the rule of 40 to allocate budgets between the different groups because we wanted to make sure that investments -- our investments were going to the areas that would provide the best return for our investors. And at the time, I thought we were probably more heavily investing in IP than we should be, that we were sacrificing profitability for revenue growth. And we wanted to optimize for both, and that's why we adopted the rule of 40 approach. And in doing that, what we identified was that you have tremendous businesses within Cadence on the analog side, on Custom IC, where you have virtually also a franchise where if you identify an analog design house that has 100 engineers, chances are they're using 100 licenses. They're buying 100 licenses virtually also from us. But we won't sell them 110 until they hire another 10 engineers. So your growth in that area of the business is going to be a combination of engineer -- the growth in the number of engineers using your software and pricing. And that's where pricing is important, Gal. Those types of tools are interactive in nature. Every license needs a driver. But on the simulation side, what we identified was this. There's potential for revenue going on linear because one engineer can run 20 simulation licenses at the same time. And...

I mean [indiscernible] a lot more than it is.

Exactly, exactly. And that's a key point as well. But when you talk about how we've been really -- there's -- the nature of our growth seems more secular these days. And I think it's -- the number of electronic gadgets is only going to increase over time. And we're in the design cycle. But when you look at -- on the simulation side, we looked at what's preventing an engineer like that's running 20 simulations at once from running 100 at once. And we thought, well, it's access to compute capacity, and Anirudh thought it was really important for us to lead the way into cloud and make sure that we were bridging that access to compute capacity for our users. And I mean that will take time. Users get more comfortable with that over time, but we wanted to be positioned well for that. So I think in terms of looking at our growth, I mean, I'm very pleased that you'll see in our CFO commentary that you can pick up from our Investor Relations website, we publish every quarter the 3-year CAGRs. That the 3-year revenue growth CAGRs. And we're up to 11.5% now. I mean, I'm pleased that we're rounding up to 11% growth for this year. But the 3-year CAGR has increased again, and I think, for 5 years running now. And we've delivered incremental margins of 50% for 5 years running. But -- and I think that's because simulation is inherently more profitable than the average business at Cadence. I remember when we picked up coverage back in '18, the expectations were that this is a 7% -- 7%, 8% growing industry. And since then, you basically like you said, 50% faster growth than what the market expected. So it's kind of understandable. How does that growth from the past and what Anirudh and I were talking about in terms of the fundamental drivers that are much more secular, I guess, than they ever were before. How does that position you when you think about planning for the next 2, 3 years and the sustainability of that, John? Yes, sure. Great question. I guess, I mean, when I look at this year, demand has been strong across all of our businesses. I mean, core EDA software, hardware and IP and system design and analysis even though we're lapping some tough comps there as well. I mean, we typically look at the 3-year CAGRs because it's tough year-to-year looking -- or even quarter-to-quarter looking at some of the growth rates. But it becomes more clear over 3 years because typically, most of our customers on a 3-year baseline contract. And then throughout the 3 years, they purchase add-ons. So a lot of their budget cycle for what they spend with us is reassessed more deeply every 3 years, and that's why I think the 3-year CAGR is a good way to look at how we're doing with our customers. Because the area where we're growing into with Intelligent System Design with our strategy is not just generating R&D synergies, but they're sales synergies as well. But -- because there are [ cloaks ] adjacencies you're selling. In a lot of cases, you're selling to the same people. You just have a broader portfolio and a bigger multi-physics platform to sell. But just to highlight a few product areas, I think for Q3, digital and signoff grew 18%. System design analysis was up 17%. Functional verification, I think it was 13%. And like you say, our updated guidance for '21 is now rounding up to 11%, even though that's lapping a 53-week year as well in 2020. Let me see. So yes, our goal generally each year is to try to drive double-digit revenue books and then to try and have $0.50 of every dollar of that revenue growth drop through to operating income. At the product level, I think we would generally expect IP -- we typically plan for IP to just achieve -- the most important thing for us is to achieve sustainable and profitable revenue growth. And the sweet spot there is generally landed about low-teen revenue growth for us. Now I think there's some lumpiness in terms of like we had a very strong finish to last year. I think we finished last year at close to 25% revenue growth for our IP business, which is extraordinary for us because we're not even aiming for that, but we're focused on optimizing for rule of 40 there. But I think over the 2 years from '19 to '21, we're probably going to be at the low teen level of growth. And that's kind of generally where we expect to land with our investment in IP and because we're picking our spots there more carefully. So we're not worried about trying to grow with the overall market there. We're most worried about making sure that we continue to improve our rule of 40 in that business year-over-year. And then in system design and analysis, that's a really strong grower for us. I mean, we're entering into that market. So it's an opportunity for growth. And then the majority of the business that we do there, we've set up our contracts in a way that we recognize revenue ratably or a recurring nature over time. So there's subscription revenue there. So that will build over time, but it's a newer area for us. So we expect the growth there to be stronger than in other areas. On the hardware side, that's a bit more difficult to predict. And generally, it's just difficult to predict. I mean, there's -- because there's some lumpiness in terms of the timing of hardware deliveries and things like that. But secular trends in our own market-leading products there have probably driven strong demand for our hardware this past couple of years. But your kind of visibility into the next 6 months is good, but your visibility beyond 6 months is more difficult because that's more a pipeline-type business as opposed to like 85% to 90% of our revenue is recurring in nature. And you've got great visibility into when renewals come around. So I think when you look at all of the businesses, they're all slightly different. But we pride ourselves in the diversification it gives us and our coverage across what our customers are really trying to do from chip design through chip verification through packaging board and system analysis. We want one multi-physics platform to address all of those steps for our customers.

We mentioned multi-physics and simulation in the Systems Design and Analysis many, many, many times now. It's a market that you've entered very consciously. And it's -- the way you described it was it's a logical continuation to going from a chip design to a system design. And that then includes other types of physics, not just electromagnetic. It also includes structures. It includes the -- obviously, the cooling, so computation fluid dynamics and things like that. In order to do that, we've basically gone from developing things internally, also adding very targeted acquisitions like NUMECA this year is a great example, Pointwise. Anirudh, when you look at the markets and the opportunity out there, you said it is a material market to go after today. But the way you see it is, the way you can define it for systems engineering, it can become much bigger in terms of the TAM as well. Are you in a position today also to have that your portfolio is complete in that way? Or do you believe organic and inorganically, we're going to continue to invest in this because it is a really fast-growing part of the business, and it's something that is an adjacent TAM to what we can do?

Yes, Gal, thank you for that question. So I mean, first of all, you always want to grow an area that you are strong in. So we are strong in computational software, which is this kind of mathematical numerical software. And EDA has done this for like 30, 40 years. We believe for this kind of software, we are best in the world, okay? We say that. So -- and because of all the history of EDA and chasing Moore's Law. And that gives like a search light because you go in the dark, you have a search light, gives you the direction. And I think that's what -- because EDA is a great business. We love EDA. And our strategy is core plus. So you always have to be the leading in your core. So -- and same thing we emphasize internally. Yes, we want to go into System Design and Analysis, but we want to maintain our leadership in EDA. But this computational software gives the search light, okay, and then into System Design and Analysis, into data analytics and AI, these are the 2 new areas that we are going into. And we are open to both organic and inorganic. I mean, I think organic is better because it's more efficient, and we are engineers. We think we can build stuff. But inorganic makes sense, too. And it has to be evaluated in the context of the strategy, in context of what price we can get, can we accelerate our go-to-market. So therefore, we bought 2 companies in CFD, like you mentioned, NUMECA and Pointwise. Now the addressable market there has increased significantly. Electromagnetics and it was 800-plus thermal. Now we are at $2.5 billion. Now some parts are still missing like structural. But I think with -- I feel with finite element and CFD, we are covering a pretty big market. That should provide momentum to grow anyway with the current set of products. So if you are making 11% of revenue, you can calculate maybe more than $300 million this year. And we are playing in a 2 -- $2.5 billion growing market. What we tell our teams is, "Hey, you have lots to grow." At the same time, we will look from time to time how to expand into other areas, but I feel that we have a pretty big addressable market. And then the core EDA market is growing and then the AIML market. So -- but the key thing is we want to build this for the long term. So you have more verticals you can go into and more horizontal products we can sell into those verticals.

Got you. That's really helpful. So that opens potential new customers as well down the line, right, in areas that you haven't before. And I wanted to touch on the go-to-market because it's something that we've mentioned a few times now. We also talked about it when you when you released your Q3 results. So it's definitely an important topic. I think that's kind of at the front of the mind. But before we do that, another topic that's been extremely important from both product perspective and also demand is a few new drivers that we talk about, especially not just AI as a customer, but AI as a means to solve problems in EDA. So you guys are talking about AI design chips through Cerebrus that you released. And I want to make sure that investors that are tuning in that maybe have not had a chance or opportunity to really understand what we mean by that, that we try to kind of go through that example to say, what is an AI design chip process means versus the traditional process? And what does that mean for you or your customers for the size of the teams from the way they go and you as an opportunity to go in maybe, obviously, incrementally get a bigger share of the wallet because you can get more software in that. Is that -- should we try to do that?

Absolutely. I think the one thing I want to say before I talk about AI application is that AI is also computational software. The basics of AI, all this inference and training is all mapped in a matrix operations. So we have 6,000 people in R&D, right? We have more than 2,000 people in support engineers. So a lot of these people can do -- just like they can do system analysis, they can also do AI, okay? Of course, we hire people all the time, but we have natural strength that we bring to that market. Now what is exciting about that is -- and you just take a car analogy, for example, right? So we want to make the best car, okay, whatever your favorite car is. There are so many of them. You can pick whether you like Ferrari or Mercedes or Tesla. But -- and then that's like Innovus, right? We have a great tool or a great car that does thing, but how you drive the car also has a big influence. So you give the best tool like Innovus, Cerebrus but the skill of the driver or the engineer that's using it could be different in different companies. It could be different in different locations within the company. It could be different in different geographies. And to give you an example, like Renesas is a great example. We worked with them with Cadence Cerebrus. They were able to use Cadence Cerebrus to improve power by 10%, okay? That's huge, okay? That's like you have to go from one node to the next node. You have to spend like billions of dollars in CapEx to get that kind of improvement, but you're getting it from software algorithms, these AI-driven software algorithms. And the reason is that like if you run our Innovus tool, for example, digital implementation tool, let's say, it runs in a day, but the chip is not done in a day, right? The chip is done in a year or 9 months. So what's happening is the user is running on time, then they see what to [indiscernible], then they run another time, then they run another time, and that is iterated process. And that process historically has been manual, okay? So the driving of the car or the driving of our digital flow has been manual. And with this new tool like Cadence Cerebrus, we can do ADAS or driver assistance in that. So we can mathematically search how to change the parameter from one run to the next run. Even for the same car, even for the same tool, how you run it has a big effect, okay? So -- and we have examples from Samsung, for example, we coded in which the productivity improved by 10x, the timing improved by 25%. So it's like how well you are able to drive these tools and explore all the options. So in case of Renesas, for example, if they did it manually, you would have to run 4 million iterations, which is infeasible, right? Whereas we did it mathematically, we chose the right 200 iterations. We ran it on 10 machines and reduced the power by 10%. So this can be like an add-on on top of the digital flow. And in the end, I think a few years from now, I don't think people will use it by themselves. They will always use it through Cerebrus because it will just mathematically optimize and run these things much better than a human could do it. And what that will mean is that for us, of course, we can sell the whole solution in a much more efficient way. And for the customer, they can do more with the same number of engineers, and they can do much better results. So we are very excited. Actually, we have a lot of engagements on Cerebrus because this is a value that's easy to understand. And we are able to demonstrate it at a lot of customers.

Right. So I'm hearing that you're not replacing the design teams at your customers. If anything, you're empowering them in order to do more and to consider more options and ultimately get a better result, which right away increases the R&D efficiency is all about that, right?

I believe that AIML, of course, there's a lot of talk about AI, and we can talk about it forever. I think the rare value is in optimization and assistance. I think replacement is -- we are far away, and it may not even happen. That's why even -- I think the driver assistance or our advance and advance assistance is where it can provide value, get rid of mundane task rather than complete replacement. I think that's -- I don't think it's even mathematical possible in some of these complicated optimization base. But yes, helping them do more chips because the complexity of the system is going up exponentially, whether it's 3D-IC or silicon or -- and the design tool productivity also has to go up. So there is still a gap. So even if we improve our efficiency a lot, it can be absorbed. So I think what will happen is we will do more and more design rather than sometimes people were -- a good example was people would worry, oh, if you make your simulator faster, would they use less licenses? But that's not what happens. When you make something...

They do more work.

They do more work. They do optimize more. They search more design space because of the productivity gap. So we can improve productivity or speed of simulation or with AI this efficiency of the design. They do more work because the value of these things is so high if your power is reduced by 10%, why wouldn't you do it from a customer standpoint? So I think we still have a lot of silicon and electronic systems can be enabled with better products, better products from Cadence here.

John, how do you ensure that you take a little bit of those billions of dollars of improvements that you're effectively saving your customers overnight and apply it to the contracts that you said of 3 year in terms usually? How do you ensure that you get an upside on that and really monetize you bringing back and contribute to your growth as well?

Yes, Gal, I think it's difficult when you look on a product-by-product basis in terms of the contribution to overall revenue. I'd like to think of it more like a Netflix style. The fact that we keep producing incredibly good content, it means like if you've got an hour to spare tonight that chances are you put on the TV, you may go to Netflix and look for something to do. We would like our -- any engineers when they're thinking of doing something to go check out Cadence first and see what we have and the fact that we're innovating at such a healthy clip. I mean, we've launched 13 significant new products this year. It's not just Cadence Cerebrus. But with 13 significant new products, the content that we keep adding will keep broadening our moat and have engineers come to Cadence first. But -- and what we've been trying to do in terms of expanding, I think there's great opportunity to expand from the volume of customers that we serve. But if you have a look at Cadence traditionally, we -- I think we've called it out regularly the top 40 customers. We generate maybe 55% to 60% of our revenue from just 40 names. But of course, there's the 40 global names that everybody would -- we all started making a list, we probably hit the 40. But -- and when you look at core EDA, a lot of our revenue comes from maybe like, let's say, 1,000 customers where we have a really, really good direct sales channel with really good relationships with those customers. So you get substantial coverage there. But when you're playing in the system analysis space, you've got to learn how to serve up to 80,000 different logos, 80,000 different customers, which requires us to build out different trading channels. So we have a tremendous direct channel. And we use a Walt Disney-type analogy in which -- Anirudh [indiscernible] Walt Disney. But -- so effectively, with the 1,000 customers that in our core group for core EDA, they're customers that are VIP customers for years that are used to white-glove treatment and concierge service and that VIP treatment. We want to not change that experience for those customers. But at the same time, we're building out our channel partner program to make that more robust. So it's kind of like on the Walt Disney -- in Walt Disney terms dealing with the package holiday companies, that will give us access to a lot more customers. I think that's an opportunity for revenue growth. And then also, there's the other kind of Disney+ over-the-top model, where we're building out our e-commerce channel as well. I don't know, Anirudh, anything to add to that?

John, that's a good summary. So Gal, I think the go-to-market innovation is as important as product innovation. And like John said, we have great relationship with the top companies, and we always start with win with the winners. So we -- any new product has to win in the top. Otherwise, we can [indiscernible] go home, right? So our focus always is to have the best product be successful at the best companies. That's a recipe of success. You can never go wrong. If the best companies use your best -- your products, [indiscernible]. But then in certain areas, the market is more diversified. So we need to go scale it to the rest of the market. And I think the indirect channel, so we recently hired new people to build our indirect channel like John mentioned, and then a SaaS or a cloud offering and also partnerships in those indirect and SaaS, I think, is critical going forward to reach the long tail that is there. Now we do have some experience with Allegro. So the semi guys, maybe 1,000 customers. Allegro already has 10,000 customers, but we need to scale that 10,000 to like 70,000, 80,000. And I think cloud is a great way to do it. I mean that's -- because the smaller the companies get, the benefit of the cloud is higher. If you go to some big company has like 1 million [ cores ], maybe they can do it better than the cloud providers. But if your company has 10 people, they're not going to see...

Yes. It's too high. Okay.

So the benefit of the cloud will be there more to the smaller companies. And so therefore, like SaaS and e-commerce makes sense, and then indirect makes sense because of partnerships and all. And then, of course, [indiscernible] we need to win with the winners to give this white-glove service that John [indiscernible].

Yes. And that's a key part of our identity. I mean, we've said many times before that Cadence is a company that was created by engineers for engineers. Our entire goal has always been to be an engineer's best friend. But we're always providing like new innovative products to try and make -- turn good engineers into great engineers. I think Cadence Cerebrus is a great example of that. But -- and I think the more we do that and the more consistently we do that and build out a broad multi-physics platform, broad portfolio to address everything that those engineers might need, we consolidate ourselves as being that engineer's best friend.

Sure. The other big trend that you've been benefiting from is the fact that you have a great exposure to not just the digital design, but also a delay out in the PCB and the fact that you're able to kind of bring the system together. So in terms of the 3D-IC market and really investment that are going into that, it sounds like again, it's something that's been in the works for a while. But it's still just kind of starting. People are starting to understand the problem, what it is, especially now in practice. Is that another one of those drivers that are kind of industry horizontal, and it's going to be helping you? And again, does it change the conversations you have now with companies that previously was focused more on Moore's Law itself now going beyond the Moore's Law?

Yes, absolutely, Gal. So I think we want to align ourselves in areas of where the future is going like often used term of where the puck is going to be, but also be in a unique mission-critical position. So of course, 3D-IC is going to be huge, and there are a lot of reasons for it. We can -- and we know this for a while because of our exposure to packaging and analog and digital. But we are also uniquely positioned to benefit from that because we are the only company that can put all these things together. So we have digital flow. We have analog flow. We are leaders in advanced packaging, and now we have system analysis. So all these 4 things are critical to build effective 3D-IC. And this move of when we talked about semiconductors to system, the natural coupling point is now 3D-IC and advanced packaging. And so we are uniquely positioned. So not only the future is 3D-IC. Cadence is in the best position in that. I think that's the important part. So not only you want to be in the growth area, you want to have unique IP and mission critical in that growth area. And the reason is because we can combine all these things. Allegro is the most advanced packaging, most widely used packaging tool. And then we have Integrity now that's a single cockpit. And this thing is going to be huge 3D-IC because of several reasons. I mean, instead of, let's say, instead of having one big chip, you can have 4 chips which are still advanced node, right? And so your size of the chip just went up. The size of the system just went up. But also when you have to redesign the system, you don't have to redesign all 4. You just redesign one of them into a new process node, and that's the best form of IP reuse, okay? This is like IP reuse at the packaging level is much more powerful than historically IP reuse at the SoC level because you are only designing [ 1/4 ] of the system, only manufacturing [ 1/4 ] of the system in terms of silicon, right? And then the rest is at the package level. So this is going to happen in multiple markets, and you see that in hyperscalers and automotive. You name it, right? And we have -- I feel good where we're positioned because, for example, not only is the implementation part, we're talking to one of our leading partners. And we are working with all the main foundries who are also doing this now. And they say, well, one of the biggest issues is thermal simulation, and you pack so much power. And we have Celsius, which is leading thermal simulator. It's very accurate for these things. I can get into why it's very accurate because it combines finite element and CFD. And so you couple that with Clarity, couple that with Allegro, with Virtuoso, Innovus, you get a state-of-the-art solution.

Interesting. I'm going to squeeze one question in. I know we're running close to the time. But when you mentioned activity, it's not just activity at the latest at the 2-nanometer node. It's activity that happens at an older node that actually helps you. But looking at John then, those things tend to be very, very profitable for you. So when there is activity on all of [indiscernible] on the -- not just the latest, the leading edge. It actually helps because I assume there's more known about it, there's less handholding that the customer needs because they don't know what they're doing, right? So you mentioned thousands of people that you have committed to helping customers solve really complex problems. When you go to a problem that's better known, we're just kind of stacking it up differently. It's bound to be more productive as well. So in light of that, I imagine you still feel good about that 50% incremental margin going forward.

Yes. Gal, I think there's a few important things to understand about 50%. We've achieved it 5 years running. 2 -- in 2 of those years, we've gone through a pandemic. But I really can't see a huge amount changing in 2022 that we haven't seen in the last 5 years. But the point that we were trying to make in terms of Cadence was always a great business, and EDA was always a tremendous business. The challenge was always where do you grow? I think with Anirudh's Intelligence System Design strategy, we found a decade-long runway for growth here. But -- and it's a huge opportunity because we scale really, really well. And the thing I'd point out to you is that we're achieving this 50% incremental margin while we're building out -- investing heavily to build out a multi-physics platform. The other thing I'd point out is that we spend upwards of 35%, 40% of our revenue every year in R&D. I think more than half of that R&D spend is with engineers that do not touch any products that generate revenue today. But -- so that is through new product innovation for the future that we sink the R&D cost upfront. And the longer people stay at process knows, the longer we get to harvest what we've created. But it's an incredibly profitable company and a great cash generator that -- and I think is just a magnificent strategy designed by our new CEO, Anirudh.

Awesome. There's no better way to finish it. We're going to see how that continues to play out. But thank you so much, Anirudh. Thank you, John. Thank you, Alan, for joining us today. As always, appreciate your time. And I'm sure investors really did the same. We look forward to having you hosting one of the events soon again.

Excellent. Thanks for having us, Gal.

Thank you.

Thanks, everyone.

Bye.