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

And we're live. Great. Thank you, everyone, for joining us for the next fireside. We are -- this is the next fireside part of our design software conference as part of Berenberg Thematic Software days. We have a great pleasure to introduce the team from Cadence. We have John Wall, who's Chief Financial Officer. And we have Nimish Modi who is SVP of the Marketing and Business Development for Cadence. And I'm sure everyone is very familiar with both of you, but we're very, very glad to have you today to host you again and get a little update on Cadence.

It's great to be here. Great to be here, Gal.

Glad to be here.

Great. To go right into it, maybe one thing that has shifted significantly, not just our rating, but the fact that what followed was, I guess, as we were thinking about Cadence through the years, and I remember when I first initiated back in 2018, even at that stage, Cadence was considered as a solid mid-to-high single-digit grower. Had a very, very decent exposure to a leadership kind of in a [ chart ] leadership in the EDA space. But since then, the story has really developed, and I was looking at my estimates back from there, for these years and how grown that was. And for once, we were wrong on the -- as analysts, we were wrong on the better side where, we didn't expect the growth to pick up. We didn't expect the margins to inflect. We didn't expect all these good things. So to me, especially when we talk about the growth rate of the business and what's changed in this 3 to 4 years, recently that makes Cadence becoming from this high single-digit to low double-digit grower. Maybe John or Nimish, which are the secular trends that are reshaping that are enabling you to do that and to really drive that demand to now be sustainably considered as double-digit grower as you get it together.

Maybe I can start, and then Nimish can add in. But before I begin, I should mention the safe harbor statements. Today's discussion will contain forward-looking statements and will make use of certain non-GAAP financial measures. So 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. Yes, just in terms of what you've highlighted and what's changed, I think the most significant change was we changed our strategy back around 2016, 2017 time frame, we're executing now to our Intelligent System Design strategy that has our EDA and IP portfolio as the foundation and leverages our computational software expertise to triple our TAM by going into newer markets. I think that's probably been the most significant change. In 2016, we kind of revisited our prior strategy and how to look at where we come from. And I think we had a clearer review of our identity or clear view of our identity at that time. At that time, we spent the previous kind of 4, 5 years building out an IP business. I thought we've done an exceptional job taking it from 0 to 10% of revenue over a short period of time. But what we've learned in that experience was that, that we signed up some business that with hindsight, we probably wouldn't have done if we had that time over again. And we realize that some IP business is great, some of it, not so great, that we love the royalty business in Tensilica, we love the high-margin, low-customized IP business. And certainly, you could see that you get that from STAR IP or what we call STAR IP, so it's like differentiated IP. There's less pricing power in commoditized IP. And it has an impact on profitability. So we'd rather focus on areas of higher profitability and sustainable profitability in the IP area. So what we decided back in 2016 was a lot of the investment dollars were going to the IP business because that was the fastest grower at the time. And we looked at it and we thought we'll apply the Rule of 40 to make sure that investment dollars were going to the locations that we're providing us the best return on for want of a better word, it's like take home, we were focused more on -- we thought we'd focus more on take-home pay than gross pay. And the revenue growth for the sake of revenue growth wasn't that important that what we wanted was to make sure that we were maximizing earnings for our shareholders. When we looked on a Rule of 40 basis over the prior years, what we discovered that while Cadence was priding itself on improving operating margins, they were doing so at the expense of reducing revenue growth and Rule of 40 metric kind of went sideways for 5 or 6 years. So in 2017, we adopted a different approach to the annual operating plan. And we directed the investment dollars towards the areas where we thought at the highest growth and the highest opportunity for sustainable and profitable revenue growth. And that served us pretty well. I think if you look at our CFO commentary from the most recent earnings call, you'll see since -- we always use 2016 as our starting point because that was the year before we adopt this new strategy and the new approach. But you'll see on the -- like I'd like to measure the 3-year CAGR, 3-year revenue growth CAGR because most of our customers are on 3-year baseline contracts. But you'll see the change in the revenue growth profile when we move to that model. And I think it's most of this down to the strategy to be fair. And I think Nimish, do you want to talk to any of the Intelligent System Design strategy because it's really executing its strategy that's driving those results.

Right. So the strategy, as John mentioned, we remand the strategy and got a very robust strategy, which we are very excited about. If you take a step back and look at the environmental kind of drivers behind it, we have talked about generational drivers, industry trends in 5G, hyperscale computing, autonomous. I mean all of these are accelerating the digital transformation of multiple end markets. And what that means is that -- or what that entails is driving the need for more high performance, low-power compute, high-density storage, high-bandwidth communications. And then we also talked about system companies increasingly investing in developing their own custom silicon. There's domain-specific computing, a whole bunch of new start-ups coming up. So all of this is, as we said, is fueling the golden era of semiconductors and electronic systems and leading to a kind of design activity. And so this has been providing a consistent tailwind to our business. And then for our strategy and our innovative solutions, we're just continuing to proliferate with what we call the market-shaping customers. So when you look at the -- you hear about Moore's Law slowing down, which is true, it's going down. But it still continues to march relentlessly forward, right? Customers, especially when you talk about mobile, hyperscale, AI, they are aggressively looking to get the maximum PPA, the power performance area gains, pushing the limits of transistor scaling. And at the same time, we are seeing an acceleration of the more-than-more front, right? A move from monolithic to modular design where it makes sense. You see a SoC disaggregation happening, the need for more sophisticated packaging. And so when you put this all together, right, our broad portfolio strengths in analog, custom, digital packaging, we're really positioned uniquely to benefit from both the more and more and then the more-than-more front. And you've talked about systems companies contributing to more than about 45% of our revenue coming from them. And so when we look at this, this has been a change, if you will, in the environmental conditions, which is providing a consistent tailwind, our innovative solutions and leadership solutions in the core EDA and IP front. And then we are expanding our portfolio beyond EDA as part of our strategy that John referenced into system analysis, for example. So all this is contributing to the growth of our business.

Really helpful. If I take a step back and I think about these drivers, right, yes, that's accelerated. So I asked you what's changed in the last 3, 4 years, and it's part of the strategy and execution, I guess. It's also helped the fact that your end markets are being obviously extremely relevant and everything that we're doing, and those don't seem to be cyclical drivers, it seems more like a structural driver, right? The proliferation of autonomy of electrification of smart-connected devices, machine learning, things like that. When I think about these drivers I still feel they were in early innings of it. Would you agree with that? Would you say that the things that are driving the -- from both hyperscalers, just being able to process more data and intelligently process data to having electric vehicles, having autonomous vehicles in the future, all these things are just about to come through? And for that, there's going to be a lot of developments going to have to happen both at a chip level, but also as like you said, packaging and the systems as a whole? So this isn't like a 2-year trend that you're seeing as an acceleration. This is a decade-long trend, I guess.

Absolutely, Gal. I mean these are multiyear kind of generation. The reason we use the word generational trends is exactly that. This is not for the next 3 or 2 years. I mean these are going to play out over several years. And each one of these trends is at a different kind of stage of maturity, if you will, when you look at them of their own life cycle. But collectively, they're still at the very early stage of that, right? And so we fully expect these to continue playing out and driving more and more design activity for all the various kind of factors that we talked about. And the innovation has to happen at the holistic system level, right? I mean the chip is obviously the underlying component and the engine, a very key part of that. But then you look at the -- our strategy, and it's very purposeful in the way that we have thought about this because we are addressing the customers' challenges at the holistic level, prudential in IP, next layer up as the chip, PSoC, which is where our [indiscernible] flows are, next level up is packaging, then PCB and now we're moving up into system analysis. So we're staircasing our way up the system stack, if you will, and taking up more and more of the customers' challenges and addressing them in that manner. The other thing I would say is while these are all horizontal in their own regard. But when you're trying to solve something as we -- as an example, on advanced packaging, for instance, sure, you need very sophisticated advanced packaging technology, but it's not just packaging by itself. It's the holistic view of -- you got to look at the implementation, the packaging, the analysis aspects, the thermal aspects of that which are all very, very complex in their own regard. You put it all together, and this is why you come up with a platform level view of integration across these different domains, which need to work holistically together. And that's where the challenge is and that's where the opportunities are and that's what our strategy is geared to crossing.

That's really helpful. And as a result of that, like when you look at market-shaping customers that you've had, where they continue to push the more slow to kind of its limits in a way, but then at the same time, changing the packaging in order to kind of get productivity gains from that side. On the other side, you said systems companies have outgrown, we're -- I think the percentage has moved over the last few years from 40-ish to 45 like Anirudh said recently. I don't know what's kind of -- when you look forward, what's the balance between your traditional kind of EDA customers in the core versus the systems companies in the future? What is 45% kind of what -- where you guys think this is going to normalize with because the growth rate is going to converge? Or do you think systems companies because the problems are so much more complex and we're putting more compute power to solve all this, it actually continues to outgrow the core EDA side that and could start being 50-50 over the next few years?

Yes. Let me start, and then John can add on. So Gal, if you take up -- as you mentioned, if you take a step back a few years ago, yes, we were -- systems company's contribution is probably in the high 30s or so and then it's a little step over time to 45%. Now the grant may not have been as steep, but that's a good problem in a sense because the tide has lifted both boats, right? So it's a relative percentage. And we have done very, very well with the semiconductor companies and also with the system companies and now it's 45% -- about 45% of our revenue. We would obviously expect to do -- continue doing well because we have on both sides, because there are 3 different things which are happening over here. Our -- we talked about our ISD strategy and the portfolio, the end-to-end EDA flow and being -- how that's transformed towards the course of the last few years, and it's now a leadership. And so we are continuing to make up market share -- segment share gains over there on that front. A lot more activity happening on the semiconductor design front as well. So that's one vector where we are continuing to grow. Then we talk about systems companies doing -- building their own custom silicon and the like, and that's hyperscalers, for example. That opens up a whole new vista of opportunities with system companies, not just at the chip level, but also as we move up the system stack with all the portfolio components that I referenced earlier, so that's another vector. And the third thing that we're looking at is that beyond EDA, as we are expanding into these new areas of system analysis, computational flow dynamic, there the mix of customers, obviously, is largely it's system companies, right? They're new customers. We have talked about going from having 1,000, 2,000 customers, right, and going into tens of thousands over time. And that mix is largely going to be on the system side. So we do expect that system slope, if you will, to kind of steepen and continue outpacing semi, if you will. But in the big picture, it's -- again, it's relative and it's all good news, it's all goodness. And semiconductors, the definition of the word, so what's the semiconductor company and the system companies, I mean, has also been morphing over time, right? Semiconductor companies are moving up the stack more to a systems -- system companies down -- coming down towards more semiconductor. One commonality, which we really, really kind of are looking for is more design activity, and that's happening. And our goal is to provide as much of that -- of the needs of our customers like full featuring, if you will, our portfolio.

John, go ahead.

It feels like the -- well, I would say that it feels like the customer base continues to expand. And as it does, I think there's a natural kind of tailwind on the systems number. So I think your assumption that at some point, you hit 50-50 is probably true in terms of the trend. It's just -- as Nimish says, both semis and systems, we've seen very, very healthy growth on both sides. So it's almost a race between the two.

Yes, better resolve right where it needs to [indiscernible]

Yes.

Exactly.

You know what's interesting to me, as you expand the number of customers, right, and like you said, Nimish, you used to have 1,000 customers, but even out of that, probably it will be -- if you count 300, 10 years ago, probably account for a vast majority of revenue, right? And now you're kind of expanding that. You've probably had to develop go-to-market strategies, kind of the different messaging, all that stuff, which sometimes involves hiring an additional people. You've been growing significantly the number of people at Cadence as well. But then that all sounds terrible from a margin perspective. And then I look at you, John, and like what, you've been able to kind of bring that pretty much 50% incremental margin over the last, say, 5 years now, 4, 5 years?

Right.

How were you able to do that? How are we able to expand? Is not the answer better pricing overall because the systems companies also come in at a very decent price? Or what else have you done in order to be able to kind of balance the two?

No, fair point, Gal, and good observation that -- I mean, take -- knitting that into your earlier question about we started as a kind of a mid-to-high single-digit grower. The EDA was always a great business. It was just where could it grow and where could you find growth. I think with our new strategy in the Intelligent System Design strategy, that we found that our center of gravity was core EDA, and we feel it's our destiny to be the largest core EDA company because of our strength in analog, and we figure it's only a matter of time before we're 50-50 on digital. And that was a great base to grow from. So we're naturally kind of moving into an adjacency then of packaging and board and then into system analysis. And one thing that's been very interesting as we've grown is on the system analysis side, like if you look at the different levels of profitability of the businesses at Cadence, the least profitable business is probably IP. IP often has a customization to it and still requires more manpower and some of it take on less services side. I mean some of it's great in terms of royalty revenue is 100% margin. But a lot of IP requires customization and there's a services component to it. So it can carry a lot of cost with it. The most profitable business by far, at Cadence, is simulation software. When we sell simulation software, the reason it's the most profitable is the revenue. And revenue is not tied to expenses in the way it would be for, say, an interactive tool. Like we have a Virtuoso franchise on the analog side and custom, which is great. But if you're selling to like a 100-person analog design house, chances are, they're buying 100 licenses of Virtuoso, they won't buy 110 until they hire 10 more engineers. So your growth there is always going to be a combination of volume and pricing. And probably a little bit more on the pricing side because the number of engineers isn't growing dramatically. On the simulation side, that relationship is different in that one engineer can use 10 or 20 licenses. And because you're seeing our growth in our Intelligent System Design strategy, taking us more into stimulation style tools, that you get that not only a relationship where revenue can grow faster than your expense base, and that's what you've seen with we've been driving incremental margins, as you pointed out, 50% for the last 5 years, it's averaged 55%. 5 years ago, we were like kind of high single-digits grower on the revenue side, with mid-20s operating margin. We recently reported strong financial results, and it's -- the '22 guide has a midpoint of 12% revenue growth and just over 38% non-GAAP op margin. A lot of that is coming from this, the profile of the business that we're doing. It's much more profitable business. The beautiful thing that I've seen in recent times of Cadence is, I thought that this is fantastic while we're growing into the system analysis side that there's -- simulation is so profitable for us, is that some of the interactive tools are now taking on the profile of those simulation tools with the advent and the new introduction of Cerebrus. Cerebrus allows 1 engineer to use multiple licenses of Innovus. So Innovus would also be an interactive tool where 1 engineer would use 1 license of Innovus. But with Cerebrus, 1 engineer using Cerebrus can use multiple licenses of Innovus and -- which allows that also to take the profile of a higher revenue grower for us. And I think that just leads us naturally down a path for 5 years running now with 50% incremental margin. There's clearly no near-term ceiling on operating leverage at the company.

That's helpful. And I want to piggyback on something you said at the end, which is really interesting because it's -- you introduced Cerebrus, if I'm right, about 9 months ago or something like that. I mean you kind of became commercial. And you've had a really strong traction. And outside of last quarter's results, it kind of came out that these things are starting to contribute to revenue growth on their own. They're expensive products, they're high value-add products. They use elements of machine learning and to optimizing the design, which basically means that customers can do more with the same number of engineers. So it's the first time in EDA, I guess, where we're really starting to see that notion of decoupling the license to the engineer correlation that you've always had. And I guess, does that mean that customers are happy to spend a lot more of the tools if they know they can develop better tools, more -- consider more designs and stuff? And does that ultimately potentially also increase demand for the core EDA tools that they buy because they don't tie them to an engineer anymore?

Yes. So Gal, I mean, Cadence Cerebrus, I mean, I think we talked about this in the context of AI-driven kind of chip design. It's absolutely gaining traction. We're really excited about Cadence Cerebrus. And the results are our customers are very pleased with them. It's really, as you said, it's incorporating sophisticated machine learning technologies, exploring the entire design space, something which maybe be on the scope of individual engineer to kind of think of all these different potential kind of combinations to try out. And then Cadence Cerebrus intelligently optimizes. So it's not just about productivity and getting the results faster. It's coming -- optimizing it to come up with a better answer faster and do so in a completely automated manner. So...

It's a different indication at the onset but it's like developing new things that they wouldn't have thought of because it's impossible.

Yes, you're exploring the design space, a mathematical exploration of the entire design space, right? And so with engineers, you come through with experienced heuristics and the like and you're trying out these different options here, which might take weeks to months or even year to get to that. And here, you're seeing results in a couple of weeks, in many cases, 10% better power or 15% more timing improvements. I mean these are significant kind of gains, if you will, in a very short amount of time. And basically, we've seen that in the history of EDS, things have moved up the stack, I mean, the synthesis. You had engineers kind of manually kind of laying out -- drawing out transistors and the like and then you have synthesis with that. But that doesn't reduce the amount of engineers, it just changes the task that the engineers are focused on and typically to move up the stack, think about bigger problems, bigger challenges, which in turn, fuels more innovation and the need for more help, if you will, from us. And so I think it's really a virtuous cycle in that regard. And then at this point, from a Cadence Cerebrus perspective, I mean, we're still in the very early stages. As you pointed out, it was just a few months ago that we introduced the tool. And Cadence Cerebrus is a new product, sold as a separate product. And at this time, the focus is on the digital flow, right? It basically looks at our digital full flow and leverages the underlying engine. So you need to have the best-in-class engines for synthesis for placing route and the like. And then get necessary [indiscernible] on top of them and deploys those in a parallel manner to do the design space exploration and then optimization. And so it's still early days, but the results have been really good. We're excited by the momentum and it's too early to quantify the overall monetization opportunity over there. But we'll have a better idea as more customers adopt it, and we can see the percentage of users who are adopting it and proliferating and the like. And so stay tuned. But so far, so good and customer feedback has been very strong, very positive.

I guess what you're saying is it's really hard to size the opportunity at this stage. But it's multiple times bigger than what it is today, and it's really starting to kind of contribute to your growth rate. And it actually might be multiple. It's interesting because when we first started reading about it during the year, even before when it was kind of in concept, the notion was that it's going to focus on there. We look at a power performance in the area of segments. The area was the first one that was kind of always optimized for and something that if we can be more mindful of how the layout is we can be more productive, I guess. We can also shrink the area unnecessary. But then you're saying now that you've been able to kind of patch the other parts of the consideration. And now all these 3 kind of play together as what is the ultimate solution that takes into account all the PP&A.

Yes. It depends upon the value metric of which the design or whether designer chooses to optimize into their optimization point on. And so again, it's -- the machine learning or algorithm is reinforcement learning. So it kind of gets reinforcement when it goes to a certain path, whether it's tracking to where you want the end goal to be or not. And so you can -- it depends on your optimization point and -- but it's about -- the thing is it's about quality of results and productivity, right? It's a combination of both. And both are equally important. Together, that's awesome.

When I look at Cerebrus and the concept of AI within the design and verification and the whole kind of flow, I understand it is like a platform. I like that notion when I look at that and interestingly, in mechanical design, we've been talking about generative design for years now, at least half the decade. And when I talk to some of the users that or potential adopters, there's been a lot of kind of fear of this thing might take my job away, these things might threaten me or if I give to my boss, they would have said, why didn't you do your job and like if a tool can do your job, what is your role? Have you had any pushback at all in terms of the adoption of the technology when you kind of go and introduce? I guess you alluded to the fact that engineers are kind of moving and their jobs are changing. But what has the adoption been in terms of just -- or feedback been when you kind of come in with a concept and have you had any pushback at all?

Yes. At this point, again, you think about getting an Cerebrus as an engineer-assist kind of tool, right? It's assisting the engineer to get to the answers, as I said, in a better manner, faster, more efficiently and more effectively. And it does not really require any fundamental kind of significant changes to the floor, right, because it's sitting -- you already got -- you're using the Genus, Innovus and the underlying engines, that's part of your existing point, getting Cerebrus is sitting on top of that and just kind of naturally blends in. So it's not disruptive in terms of them having to reorder their entire methodology and flows and the like. It's a natural kind of extension, if you will. So there hasn't been any -- that seamlessness of adoption, training has not been really that a big concern either. So it's just been -- the adoption has been very smooth in that regard. And then, of course, the results are going to speak for themselves. And so you got -- think about this as it's using that analogy, I think I have used that last week as well that think about this and say, you've got a really good car, but now this is like you're basically creating a better driver, right, as well. You're kind of helping them drive it in a much more efficient manner as well. And so no, there hasn't really been any pushback in that regard. And it just kind of opens up the vista of opportunities for that engineer to do much more. And it's just going to -- and the other aspect of this is it democratizes the ability to do these things, right? You're starting up with a new team somewhere, you don't need a very seasoned, sophisticated kind of engineer right from day 1 on this stuff. You can either -- it takes on more of the more automated because of the automation and the fact that it kind of is an assist kind of thing, it actually democratize that as well. So a lot of goodness is there. Having said all of this, again, it's early days, and we are in very close discussions with customers as well in terms of how can we even further optimize this to get better results, if you will. But yes, it's been pretty seamless.

That makes sense. And it sounds like because it utilizes the existing workflows that it also doesn't require as much training as up-front. So adoption has been pretty seamless when there is adoption. Okay.

And Gal, I probably should have included that in my explanation as why profitability is improving as well because our engineers are using things like Cerebrus internally as well, which means headcount growth doesn't have to grow as fast.

Application engineers, especially.

Yes.

That's very interesting. You guys have both done a lot of interesting M&A, and we're going to talk a little bit more on the whole CAE market. I want to focus on that and that strategy. But the other part of your very important kind of expansions have been partnerships. And it was interesting to see the tie-up you kind of announced last week around the results as well with Dassault. When you think about kind of -- especially around packaging and the PCB side, it's a market where we see obvious ECAD and mechanical CAD kind of connections, right, that are coming together. What drives you to kind of think about, oh, maybe we'll develop some things in-house, maybe we'll buy something versus we're going to partner just at a high level? And like specifically for Dassault, what does the partnership bring to you? Like what are some of the interesting motivations that kind of drove you to sign that agreement?

Sure, sure. So yes, and we announced this partnership last week, and we are really excited about -- it's a very strategic partnership and it brings together Cadence's Allegro platforms and Dassault's 3DEXPERIENCE platforms to provide an end-to-end electromechanical and product life cycle management solution. It's at the enterprise level at this point of time. And it's really -- we always look at these things from the outside in, meaning what are the customers end requirements, how can we help our customers better. And in this case, it helps our mutual customers accelerate their overall system development process. So Cadence, it brings 2 leaders together. Cadence is a leader in electronics design. Dassault is a leader in 3D mechanical CAD, PLM, data management. And this collaboration kind of brings together these 2 leaders to optimize the customers' kind of mechatronics, if you will, challenges. But when you're coming down from the systems -- or the first electronics aspect that you see is PCB, for example, or going up the system stack up from the chip, PCB is key intersection point. So effectively, this collaboration with the strengths that the 2 companies bring enables some end-to-end information flow, right? So when you think about this from concept to PCB design to manufacturing, you just get a seamless flow and you can do all of this, right, in a real-time collaborative environment. So I think this is very -- this is going to be very useful to our customers. We -- obviously, we talk to usual customers, have been working with them, and it is very exciting. It's very -- the beginning of this -- the beginning stages that were here, and we've already been engaged in like a multiyear collaboration and totality of the solution to get to this point. So we think that the entire system design analysis space, which includes PCB, is entering a period of accelerating growth. And then as we look at this in the totality of the holistic electrical, mechanical and also the data aspects of it, we think this partnership is going to yield some really good results for our customers.

It could also potentially help you go into these new areas and new customers that you don't have to directly employ your own sales force or find your own channels too because like just the solid works channel is probably the broadest channel in mechanical kind of discrete manufacturing side. So if we can start utilizing some of those, is that the idea as well?

Yes. So the focus right now is on the enterprise front. But yes, effectively, our sales teams are going to co-engage in these discussions with customers, right? And so Cadence's customers and PCB customers, okay, we're going to co-engage and get in the Dassault solution, introduce them to it and vice versa. So there's going to be a lot of synergistic co-selling and co-engagement opportunities for other mutual sales forces to go together with. And then stay tuned. We are exploring other collaboration opportunities but right now the focus is on enterprise, in those areas I mentioned.

Other area, as I mentioned earlier, there's been very interesting just in the whole space of industrial automation software has been linked to simulation. And the way we think about that, I guess, is mechanical verification, mechanical simulation is still at a stage where a lot of people play in the physical prototype centers, labs. We saw during COVID, a lot of these labs were shut down for 3, 6 months. And the only way to do work -- R&D work was through simulation, virtual simulation. And it's interesting because that doesn't happen in EDA, does it? People don't play around with potential concepts in physical world. So you've designed a full flood from design to verification to basically production virtually. Not something that's happened in mechanical, like I said. Is that the opportunity? Is that the point where -- John I heard you say it before, it's like the whole rock band versus the orchestra kind of analogy that you've used. Is that the point that EDA industry has shown the whole world of how you can virtually design and manufacture something before it actually gets into the production? And because there's a lot of R&D dollars being spent in automotive, aerospace and everywhere else, you believe that there is a natural drive to kind of increase that 6, whatever billion market it is today to $20 billion plus, as you said, because of the fact that you can bring it more into the virtual world. And hence why you started making acquisitions in things like Pointwise and NUMECA and potentially even in addition to that?

Yes. So I mean the system design complexity is just exploding. It's rapidly increasing. So you've got that thrust of system design complexity and then you got time to market pressures coupled with that. And then the cost of failure in terms of, hey, look, if you don't get it right, so it's increasingly important. You hear the shift left paradigm, and you're seeing it play out in front of our eyes accelerating, right? You try to flush out as many of the issues as you can at early stage through simulation. And it's not just that. It's also going to validate your design ideas, right? You may have different ideas on the design itself. Okay, let's validate this and see what's going to best -- which is the best option to help us meet our design goals, product goals, if you will. So all that's driving the need to do more simulation earlier, right? And that's basically the thrust of the acceleration of the system analysis or simulation market. Right now, roughly it's $7 billion, $8 billion TAM market, growing at low double digits. And we just think from our strategy perspective, John referenced that earlier on, it's -- we think our core competencies in computational software, we feel we can really bring a disruptive kind of view approach to this by leveraging the core competency and moving up the staff with massively parallel architecture, innovative algorithms and offer these products. And the vision is to build out a comprehensive multiphysics platform. But we go about doing and we are going about it in a very structured and methodical manner, right? We started off with electromagnetics with our Clarity product and electrothermal with Celsius. And then as you pointed out, Gal, last year, we expanded that into computational fluid dynamics with the addition of NUMECA and Pointwise. So collectively of that overall TAM, we're addressing about $2 billion, $2.5 billion TAM of the system analysis time, I mean. And the thinking over here is that we are providing an end-to-end simulation of flow, right? You go from reprocessing, meshing, solvers, optimization, post processing, and it's also enabled us to have -- adapt into getting new customers and several verticals which we have not participated to a degree before in turbomachinery and marine, for example, and obviously, not automotive and aerospace and defense. So I think we're going to continue to invest to expand our platform. The goal is to, again, land and expand, right, is to proliferate with these customers and also continue building out the portfolio. M&A, as far as M&A goes, we are very disciplined about M&A. We look at means as a means to add either scale, domain knowledge, customer reach, portfolio augmentation. But M&A, every one of these that we do has to fit and further our strategy and, of course, make financial sense. So again, it's a very exciting market. I think we have brought some really disruptive products out -- to the front out there. And then on CFD, as we mentioned in our prepared remarks as well, we added close to 100 customers, new customers last year. So still early stages but excited by that and stay tuned.

And in terms of the profile of the business, it kind of fits right in as well, right? Because...

Oh, absolutely.

It's slightly accretive to the growth, but also margins are -- at least potential margins are well in line with what you guys are doing in the core business as well.

Yes. So the 3 things which you think about strategically, right, is it a market worth getting into? And then you talked about that large TAM growing to low double digits. So that's #1 checkmark. Number two, what's the value that you bring to that by entering that? And this is basically leveraging our core competence in computational software. We've already proven that we can come up with a disruptive angle there. And the third thing on this is we are absolutely in a steady state. We expect this business to be equal, if not better than our EDA margins. Absolutely.

And everything is a make or buy decision, right? I mean we consider how long it would take us to make it ourselves. And I mean, often, what you see us doing a small talk in M&A because it accelerates our road map.

Yes. Awesome. Maybe if we shift to focus a little bit towards what's been happening recently, the outlook for the year, which you've introduced. And John, when I look at the kind of results last week, what really kind of -- what really jumped out was the fact that you have a pretty good visibility, especially in the first half of the year. Based on that, you're dealing with kind of this record backlog. Part of it obviously being driven by hardware as well. But it seems to me that when you kind of look at the rest of the year, you didn't go out and just project or extrapolate this trend from Q1 potentially in the backlog you have. It's more kind of, I have a pretty good confidence in the year because I also believe that if the strength from the first half of the year repeats, especially on these things that are less predictable, more one-off in nature than this potential upside to it. Would you say that's the right assessment of the way you kind of guided for the year and why you kind of like to end up versus what you were able to introduce at the beginning?

Yes, that's reasonable, Gal. The -- I think what you highlighted immediate at the start of your question is exactly right. I mean with the -- we have tremendous confidence in our numbers because 75% of the year's number's already done. It's already in that backlog. So $2.5 billion of that $4.4 billion will revenue in 2022. And then it's really just the other 25% and where that comes from, it will come from a combination of new recurring revenue business and new up-front business. But in the backlog, there is kind of a slightly higher up-front revenue profile that's hitting Q1, specifically, that's why you have the spike in Q1. And I called out on the call that -- the last year, we had 88% of our revenue was recurring and 12% of the revenue was up-front. In the -- at the midpoint of guide, I'm assuming 87 and 13, so 13% up-front. I know 1% isn't that much, but it is, 1% of $3.35 billion is $33.5 million. And a large portion of that is hitting Q1. That's why you see the spike in Q1. And like I say, it's a combination of our up-front revenue typically comes from IP and hardware. On the IP side, we only grew 7% last year. I've assumed low teens in the guide. So it's slightly more IP revenue. But we finished 2021 really strong on the bookings side and booked a lot of hardware. We have a lot of inventory to make that hardware, and we're making the systems as fast as we can. The backlog that we have at the end of the year, I don't think it's possible for us to ship all of it in Q1. It's going to bleed over into Q2. There's that much demand and that much of a backlog, if you like, in terms of meeting customers' needs. So we had $116 million in inventory at the end of the year, which I think to put in context last year, all of our product and maintenance COGS, if you exclude the amortization of intangibles, was $175 million. So about 2/3 of the year's worth of COGS in inventory. So we have plenty of components, and we're getting more time with our manufacturers to make the system. So I think we'll be able to make more systems in Q1, but we're servicing our customers as quickly as possible. That's why you've got a spike there. But you had tremendous visibility into the business. I -- your question was in relation to -- have I projected that out because I did say on the call, I expect 50% of the revenue to fall in the first half, 50% in the second half with this current guide. And from that perspective, I derisk the second half from a downturn, let's say, in that -- I didn't want to extrapolate that strength in hardware that we've carried into the year, into the second half of the year because that's notoriously a pipeline business. It's -- you tend not to see the opportunities until they're within 6 months. And there's so much going on in the world right now. I thought it was safest to derisk that.

That makes sense. And you haven't seen any significant on the hardware side, supply chain issues that will stop you from being able to kind of provide the -- or effectively put out the orders that are in the backlog?

But like I say, our approach has always been that like a few years ago, we moved from just-in-time inventory management to just-in-case. We thought we'd just -- we'll continue to beef up our inventory just in case we need it. We did a full assessment of our supply chain to make sure that we had second source suppliers for -- and ideally domestic second source suppliers for the more significant components and the critical components. Where we didn't have that, we wanted to stock up on inventory. So that's protected us pretty well through all the supply chain challenges that we're seeing everybody else have. But of course, it's not a policy. I mean, like I say, about 2/3 of the year, we still have to source probably over 1/3 of the inventory throughout the year and -- but those orders has all been placed last year. Yes, so we feel good about it. But like I say, the longer that you have supply chain challenges, the more likely it is to hit everybody, but it will hit a whole bunch of companies before it comes to us.

That's great. We are over time, so I'm going to have to thank you for your time today. Thank you, everyone, for listening in as well. And if you have any follow-up questions, I'm sure both John and Nimish and Alan will be happy to follow up with. Thank you, John. Thank you, Nimish.

Right.

Thanks, Gal.

Thanks, Gal. Thanks for having us.