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

Yes, I think so. I mean I think we have achieved -- so I mean, like this year guidance is about 17 -- slightly less than 17% revenue growth and slightly less than 40% operating margin. And then what we have done in the last 5 years is our incremental margin is 50% or better. So that means if we make $100 million more next year, the $50 million is profit, okay. And this is why we have invested in the new areas. And I don't see any reason that we can't continue that.

So you've been building out your go-to-market in simulation...

All in that environment. And so in terms of go-to-market, there's multiple parts to it, okay? So there is -- one is the direct sales force and which -- EDA or chip industry is mostly direct, okay? And then I think as you go to the system space, we are now investing in indirect and cloud, right? But one thing to remember is even for this -- so in the systems space, like -- some of the companies can have 80,000 customers, okay. That's the other exciting thing about the system space. There are a lot more customers. In the silicon space, there are probably a few thousand customers. But if you look at the top markets or the top customers for the system space, they are the same household names in the silicon space. And our strategy always is win with the winners, right? So we always go to the high end first. So there is a lot of commonality of customers at the high end or the household names are the ones doing all the system. And then we have to scale it to the 80,000 customers. So the direct sales force can help with the high end and then the indirect and cloud can help with the scaling.

Okay. And I am going to ask about chiplets in a little bit. We'll see if the audience has any questions. But you would -- so if EDA is the most difficult math equation and then simulation is also a different math equation. Where does Biosciences play in here? Because you just announced the OpenEye acquisition. I know it hasn't closed yet, but where does that fit in the road map?

Yes, that's a great question. And I'm very excited about that expansion. So as we go to the -- so of course, the chip space is critical, and I think it's going to do well for years to come. And then as we go to system space, the most profitable part of system space is simulation. And there are a lot of reasons for that. It has the highest growth, highest margin. And if you look at simulation at the system level, historically, there are 2 big areas is, finite element, which is simulating solids like cars or electromagnetics; and then CFD, which is simulating fluids, talked about airflow and thermal. So when we go into -- and these are established markets, that's about $10 billion -- $8 billion to $10 billion market and growing. But when we go to a new market, we just don't want to do what is the current areas. Of course, we can disrupt them with more like parallelization and more CS algorithms. But we also want to do emerging areas in simulation. So what is the biggest emerging area in simulation in my mind is molecular simulation, biosimulation. And this, of course, is going to be important in the next -- for the next 10, 20 years. And so we had a great -- we've been watching that space for a couple of years, and we had a good company like OpenEye. So we -- so I think the whole goal is to do simulation but also do it not just in the current market, but an emerging market in a very important vertical, okay. And the other acquisition we made, of course, is Future Facilities, which is going after the data center market and digital twins for data centers. So to me, data centers and biosimulations are very emerging areas that we should enter.

Okay. Any questions from the audience? Perfect. Switching to chiplets a little bit. There was a Wall Street Journal article mentioned a week or 2 ago, one of your competitors was quoted saying that they've seen a twentyfold increase in chiplet design activity since 2019. Why are so many semiconductor companies trying to pursue this chiplet technology?

Yes, that's a great question. And I mean, we work with -- first of all, when I was an IBM in '90s, we used to do some chiplets interpose. So chiplet basically is -- typically, right now, if you buy a chip like if you see the chip, there is only one -- there's a package and there is one chip inside the package. That's typical. But chiplet means that when you see the package inside, there are multiple chips on a package. So instead of like system on a chip, it's system on a package, right? And this has been talked about for a long time. Even in the '90s, there were discussions. And even with our big TSMC foundry partner, we are very close to all these big foundries, especially TSMC. So we did a whole chiplet solution like 5 years ago, 5, 6 years ago, but it didn't take off at that time, okay? But it's really taking off now. And I think this will be a big trend in the next 10 years. And over the last 1 year, 2 years, has really taken off. And what I want to say is that there are a lot of parts to it, but Cadence is in the best position because what it means is we have to do a system on a package. So we have all the chip tools, but we have also Allegro, which is a franchise tool on PCB and packaging, and we have the highest usage of advanced packaging in Allegro. And then we have all these analysis tools like thermal and electromagnetics. So Cadence is the only company that can put together a chip level, the package level and the simulation level in 1 platform. And the reason that you say only why is chiplets doing well, I mean, there are a lot of reasons to it. Okay. One reason is -- and you can see that in a lot of semi-companies. So let's say, you go from a 1 chip in a package to 6 chips in a package. Now some of them are big, some of them are small. So first benefit is that you can merge different kind of technologies. They don't all need to be 5-nanometer. Some are 5, some are 3, some are 28. And you can also -- one can also do more domain-specific, this heterogeneous integration. The other big benefit is that if there are 6 chiplets or 5 or 6 chiplets, when one goes to the next generation of that same product, you don't have to redesign all 6 of them. You just redesign 2 of them and you keep the other 4 safe. So there's a lot of financial and technical reasons where chiplets is going to be big. And I think Cadence is in the best position to capitalize on that. And we have been investing on this actually -- when we first started investing on the systems space, this chiplet was not a big thing, but now everybody sees the integration of chips and packaging.

So if you were able to put numbers around it, what was the increase in intensity or complexity with chiplets versus other types of advanced nodes?

Oh, like a -- I mean, I think to me it's difficult to give numbers because -- but to me, it's orthogonal. So the way I look at it is that there is -- so if you're following like Moore's Law or advanced nodes, so we are -- most of our -- most of the customers are manufacturing at 5-nanometer. A lot of our design is at 3-nanometer. A lot of our R&D is at 2-nanometer. So if you look at that thing and then there is one point x and one. So I see at least 3, 4 nodes on the plus -- at least 3, 4 nodes on the classical Moore's Law. So that's at least -- so each node is 2, 3 years. So that's at least 10 years, okay. And then when I look at 3D-IC is an orthogonal way to do scaling because instead of making more things on a chip, you can do multiple chips and you can stack them. So I talked to, for example, Stanford right next to us, and they predict that, oh, there's some papers predicting that, that can extend Moore's Law by 7 generations. Okay. But even if it's not 7, okay, let's say, it's 3 or 4. So that's another 10 years. So to me, they are like orthogonal axis, but I think it will happen in all verticals. But right now, it started more with high-performance computing.

So I mean that's a good segue to my next question. But chiplets, they're incredibly more power, not power, but the bigger compute, right? But that means that if they're larger in size. They produce substantially more heat. They consume substantially more energy. So when you think about priority in reducing those byproducts, the heat and the consumption, what does that mean for your portfolio? And how does that kind of interplay with some of the products you have?

Yes. So actually, we talked to one of the big foundries and they're investing heavily in 3D-IC and chiplets. And they told me that -- and this few years ago, that the biggest issue they had or had was thermal because you have a lot more chips and then you stack them. So thermal is a big thing. So if you look at it, we have all this investment in system simulation. Actually, we started it for the 3 reasons I talked to you about that it was synergistic for the customer, synergistic in R&D and higher margin, but the chiplet kind of brings it together. Chiplet is the integration of system and chip together. And so we could -- we did -- we have a very good thermal solution called Celsius, which we developed. And there's a lot of technical reasons why it is better because it's a combination of finite element and CFD. So this having this whole stack solution. So with Celsius on the simulation side, Allegro on the packaging side and then Innovus on the chip side. So we are in a unique position to do that. So this whole chiplet move also validated our whole move into the system design and analysis space. And we can apply a lot of those technologies back to the chiplet.

And I've got one more question, and then maybe we'll pay in the audience for questions. Any -- yes, sure. We have a microphone. One second.

So just 2 quick questions. First one is on your move into CFD. So ANSYS has been with -- they've done FEA and CFD for a very long time. And one of the things, I guess, is engineers build a lot of their models with ANSYS in mind, they've customized them for years. So do you have any kind of portability tools? Or just how do you convince the engineers to switch from ANSYS to, if you guys move into that -- if the functionality is the same, I doubt the engineers will make that move because it's going to be a fairly painful task. So just wondering how you approach that market? And then second, just on CapEx intensity, on the production side, especially for the fabs, that's been steadily rising over the last few years. You've done a pretty remarkable job on the design side. So at some point, do you think the strides you've made on design, they start lowering CapEx intensity? Or are those 2 just totally separate problems?

Yes, those are good questions. So on the -- so we actually break out our system design and analysis revenue, okay? So it is -- because the best way to answer the question is by results, right? So if you look at our results, it's about 12% of revenue is expected to be around -- our guidance for the year is about $3.5 billion, so 12% of that. And it is growing more than 20% for the last several years. So the customers are adopting our solutions. So even in Q2, the system design and analysis segment grew by 28%. And you talked about ANSYS and some of the other folks. So one thing I also want to emphasize is that we take all our revenue ratably, including in this space, whereas some other companies take it upfront. So we are growing 20% plus in a ratable model, okay? That means the real growth in bookings is even higher than that. And that is just based on -- this is what the customers are doing, right? So they are adopting our tools based on strength of our product. So what happens in simulation is -- so what you have to remember is in EDA, about 1/3 of the business is already simulation. So we have a lot of expertise in that. And also, our products will run like 5x, 10x faster than what's available. And when you go to the system space, I would say there are 2 kind of products. There is interactive products and then there's batch products. Interactive products are like Autodesk, right? Your architect is using some interactive tool to do your floor plan whereas simulation is a batch process. And we have similar tools on the chip side. Virtuoso, which are our franchise analog tools used by almost all analog designers, it's an interactive product. So when you go into the system space, I believe it's very difficult to replace interactive tools or like a best example of interactive tool is Microsoft Office. I was in IBM. IBM launched like Lotus 123. What happened to that, okay? So I think interactive tools are -- and your architect is unlikely to change there. So into -- what tools they're using to do their floor plan. Now simulation -- any tool is difficult to change, but simulation is the easiest thing to change, okay. So when you go to the system space, we naturally don't go after these interactive tools because that's not where the value is. The value is in the simulation tools like FEA and CFD and optimization. And I believe those can be changed. There's a history of that over the years even in EDA, right, simulation tools, people change. And also simulation tools, customers want more than one. See, you don't want 2 Microsoft Office tools or 2 AutoCAD tools. But simulation tools, even in EDA, they will have multiple tools. So I think in that industry, in FEA and CFD, they don't have that much choice, but the customers are demanding choice. So I think, first of all, it is easier to change than interactive tool, and you can see it in our results. And two, the market can sustain multiple players. So I think they can do well, too, and we will do well just -- so I'm not worried about that. As long as we have good products, and it's in the simulation space, I think there will be opportunities for us to do. Now on your second question on fab, we are mostly on the design side. And -- but I think the need for semiconductors is going to be pervasive. I just came back from DC, right? Yesterday, they had a big -- today is the signing, but yesterday, we had a big kind of CEO summit. And you can see like even U.S. government and EU and all these governments are putting a lot of money in it. So I think one thing I do want to emphasize that there may be some cyclicality in the -- and you see some reports from our semiconductor customers, maybe there is some cyclicality in their business. But in the long run, there are strong bull trends for semiconductors. And for us, okay, the reason we are powering through this, right, we expect to power through this is, first of all, we are on the -- we are mission-critical on the R&D side, okay? So there may be some -- the line you have to watch and our customers is R&D spend, not revenue. And then there is more and more R&D spend, and we can also get more and more part of their R&D spend with automation in AI, okay? So -- because we are essential to their product design, okay, that's the first part. And the second part is we are ratable, right? We are mostly ratable business. So we have a lot more visibility than some of the other companies. And third thing is we are highly diversified. So all geographies, all end markets, they all have to use our products. So for that reason, I believe we are much more resilient as we go through good times or bad times. And the fab, I think there is good to see all this investment, and we will benefit indirectly from it as they build more IP and fabs. But what I want to say is that we are highly diversified. So these things are in the end is good to have, and there is some benefit, but we are not dependent on it for the growth side.

Yes. Okay. Well, we're almost out of time, but I'll save my hardest question for last. What do you like so far about our Vail conference?

I love the location.

Great. Thanks, Anirudh.

Thank you.