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

Great. Well, welcome, everyone, to the seventh edition of our software conference here in London. My name is Andrew DeGasperi, I'm a senior analyst here. And it's my pleasure to welcome Cadence here, John Wall, CFO; and Richard Gu, VP of Investor Relations. So I know you have a safe harbor you want to maybe talk about, Richard, so go ahead.

Sure. Yes. Thank you for having us. Let me mention the safe harbor statement real quick. 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. With that, Andrew, let me hand it back to you.

Thank you. So John, maybe I know many people might be familiar with you, but as given the audience maybe a good idea to start with an introduction of the company and your journey there.

Yes. Sure. Hey, thanks for having us again. We always enjoy coming to Berenberg. Yes, Cadence is an EDA company or basically a competitional software company that does a little bit of IP, a little bit of hardware. It was a company created by engineers, 4 engineers back in the '80s. And we largely -- I mean it's about -- it's largely 85% to 90% recurring revenue model. We sell to all of the big electronics companies around the world. I mean we're -- Cadence are in all the things that we all can't do without in terms of electronic gadgets. The company has grown revenue at kind of low teens the last 3 years, and margins have increased from like in 2020 -- sorry, in 2016, margins were around 26%, and we're guiding to 39.25% now. I mean, the business scales really, really well. We're executing against what we call our Intelligent System Design strategy. Cadence's kind of center of gravity is core EDA. It's what's engineers used to do chip design and chip verification. But we've been expanding this last few years into packaging and board and into system analysis. There's natural adjacencies into those markets and ultimately trying to create a single platform that covers everything from chip design right through to system analysis. But yes, it's just over 9,000 employees worldwide and just over $3.4 billion of revenue.

Great. So maybe just 2 topics that investors, I believe, really want to hear about in terms of your company is resiliency in the current environment and margins, right? So maybe let's touch on the first bit. How resilient is your company, given your industry has been seen as cyclical in the past, but clearly in favor, there's been supply chain issues. So that's probably helped you a bit on your end. Can you elaborate why you're not as exposed to the up and downs of the semis industry and maybe in what ways the business is insulated?

Yes. Sure. Great question. But yes, it is a remarkably resilient model. If you consider the whole semiconductor industry as like the gold rush, where the company is selling the picks and shovels to everybody searching for gold. I mean, I think some of the engineers talk to me but -- because I think it's amazing what they do and they tell me, no, what we do is not amazing, they say what our customers do is amazing that our customers are like today's Michelangelos and Picasos and they say we just provide the paint brush. But essentially, all those tools that they use that's because of the ratable revenue model, I think it's -- we have a very resilient structure. The companies use those tools to do all of the things that they do, all the design they do. We're part of the design cycle rather than the production cycle. But -- so we're somewhat insulated to all those supply chain issues. And then I guess, the visibility into our revenue is really, really helpful. That's -- it helps us plan. I mean we have a 10-year plan. But most of our customers sign up to 3-year baseline contracts. And then over the course of the 3 years, they will do maybe 4 to 7 add-ons that will co-terminate with the baseline contract. And then at the end of the baseline contract, we'll kind of roll all those into one new baseline contract and continue on again. So you're -- we're more farmers than hunters that we build the annuity over time. The 55% to 60% of the revenue comes from the top 40 names. There's considerable coverage in the top 100, but we have closer to 10,000 customers. We have 3 channels that we typically sell through. Most of the revenue comes through the direct channel with those bigger customers. Those are the customers that kind of get the concierge treatment, the white glove service from -- VIP treatment from Cadence. But we also have a channel partner program to handle a broader audience of smaller customers. And then we launched OnCloud this week, which is our e-commerce channel, and we already have 4,000 direct customers through the e-commerce channel that's -- and that's more helpful as we move into the more -- into the systems analysis area because systems analysis traditionally has a lot more customers. And what we're trying to do is set up a structure that allows us to service those customers without changing the experience for the VIPs. And then I think just the visibility in the model. When we start each year, we typically know 75% of the revenue. But -- so basically, because we consistently do 3-year contracts, and those add-ons, we generally average about 2.5 years duration in our backlog, like in our bookings every year. So at the end of any year, there's about 50%, 55% of the backlog that's going to burn off in revenue in the next year. That generally is about 75% of the next 12 months' revenue. And then the other 25% comes from a combination of upfront revenue from sales of our hardware and IP, which probably makes up about half of that 25%, that's about 12.5%. And then the other 12.5% comes from new recurring business that signed in the year. Like if you renew a recurring contract in January, you're getting another 11 months' revenue -- of recurring revenue in this year. A lot of the bookings in December don't really generate any revenue in the year. It's really for the following year.

That's great. So the arms dealer essentially. I know it's a bad word. But like either way, it's great when you have the other side of shovel. Just on a question in terms of what was the average tenure of your biggest customer? Like how long have they been customers of Cadence?

Many of them have been around for the entire duration of the company. But I mean, you'd be hard-pressed to find any customer that doesn't use -- or any electronics -- anybody creating electronic systems that's not using Cadence software. Yes, I mean, there's another company Synopsys you're probably familiar with that -- at the lower process nodes is really ourselves and Synopsys, but in many accounts, most of our customers will use both.

Interesting. Okay. And I'll touch on Synopsys in a little bit, but maybe on the second topic of margins. Your margin profile has been strong for a long time, I think. And what is the long-term potential for the business?

Yes. So on the margin side, it's a great question again. The -- back in 2016, I use 2016 as a baseline because I became CFO in 2017. But back in 2016, what I realized was that, I mean, of course, all of the revenue that we had in 2016, every dollar of revenue was costing us $0.74. But the EDA was always a great business, but it scales really, really well. The challenge was we didn't have anywhere to grow. And when we adopted the Intelligent System Design strategy, it gave us a path for growth into, like I say, the -- we had a packaging and board business for over a decade, but it allowed us to expand into systems. And systems is heavily simulation-based and a lot of our EDA -- core EDA revenue is simulation-based. So we thought we had the right to win in simulation because of the expertise we had internally in the company. And it is super, super scalable because like to enter the systems analysis space, all we had to do was hire some engineers with domain expertise in systems analysis simulation, add them to our existing simulation team and then we opened up a whole new market for ourselves. And so from 2016 forward, every dollar of revenue growth has cost us about $0.46, $0.47, but -- so as a result, we talk about incremental margins being over 50%. So for 5 years running, we've achieved over 50%. It's about 50 -- I think 53% on average for the last 5 years, and we're on track to do that again this year. But -- so when we talk to the teams and the R&D teams about their budgets and everything, when you've achieved 53% incremental margins for the last 5 years, and you're on track to do it for a sixth year, during like 2 years of a pandemic, there's a war that's broken out. I mean there's -- we had customers added to Entity List in China, but it's very, very resilient. So I don't use any near-term ceiling on operating leverage because as long as you can continue to achieve 50% incremental margins, you'll naturally drag the overall operating margin higher. I know -- we've seen that already this year. I mean we're up to 39.25% as the midpoint of our guidance for this year. And that's up from 26% like in 2016.

Interesting. And how do you leverage the R&D spend? I know you have a partnership with Dassault. Does that help on this line item or...

Yes. Well, partnerships always help that -- but in terms of leveraging the R&D spend, a lot of it is -- we measure everything. We have all kinds of metrics to measure everything. And on the R&D side, back in 2016, one of our issues was that revenue growth was quite slow. We were growing -- I typically look at the 3-year revenue CAGR because most of our customers are on 3-year baseline contracts. And if you looked at the 3-year revenue CAGR back then, it was like 7%, 8% in that kind of range. And I thought we just weren't -- I mean, for the spend that we had in R&D, we weren't releasing enough products, enough new products. But -- so we measured things for product quality. And what we found in product quality is that we had a tendency to release products too early, and when you release products too early, you kind of have some offset customers because there's too many bugs in the product. And then engineering resources get soaked into providing a lot of maintenance to update and fix those bugs. And that time spent on maintenance is soaking away time that they could be spending on developing new products. So what we did was that we put product quality metrics into people's bonus plans on the R&D side. We had deal quality metrics on the sales side, and over the course of -- like back then, we pretty simply measured R&D time and said, just bifurcate your time between how much time are you spending on products that are generating revenue in this year? And how much time are you spending on products that don't generate any revenue in the year? And about 2/3 of R&D time was being spent on products that generate revenue in the year, which, to me, meant that 2/3 of your time you're doing maintenance and you're not doing real product, new product innovation. It's less than half the time now on maintenance. It's more than half of our R&D spend, and a significant R&D spend is spent on developing new products. And you've seen an increase, if you forgive the pun and the Cadence of new product releases over the last few years. And in conjunction with that, I mean that's -- again I think you look at that over a 3-year CAGR basis. But with the -- or a 3-year basis, but if you look at the number of new products that have been released over time, that's contributing to accelerating revenue growth as well. That's also helping margin, of course.

Yes.

So -- but that's kind of how we think of it on the R&D side, but the team is -- the teams that we have are fantastic. We measure all the talent and top talent. We haven't suffered much attrition in our top talent in the actual fact that we find the top talent is reaching out to us from other companies that we have had some attrition in under 30s.

Everyone is. Yes.

But -- over the year or 2, but in top talent, we've pretty much locked those guys down.

You have to tell me one day how you do that, but maybe just shifting to secular trends, right, in the market. You mentioned -- I think you alluded to this earlier, the growth trends are shifting between leading edge and the system design and analysis companies, which probably will provide more growth, I guess, going forward? And what do you expect in the next 2 to 3 years to look like?

Great question again. In terms of system analysis, that's a big area for growth for us. I mean it's off of the lowest numbers in our -- in all of the businesses that roll up under Cadence. That's probably the area for the biggest growth for us because we're still building out a multiphysics product portfolio for that business. So there's a lot of runway there. At the moment, I think we have -- like we want to go back again comparing to 2016. In 2016, we had probably 60% of our revenue coming from traditional semiconductor companies, 40% from systems. But at the moment, I think it's more a [ 55-45 ] from systems. I think it's only a matter of time before that gets to 50-50 because we're accumulating more and more systems customers. And particularly, I mean, if you look at a company like ANSYS, I think ANSYS talks about they have 80,000 logos that they sell to. We're probably touching on 10,000 now, but there's still a lot of runway there.

How much runway are we looking at in terms of logos generally?

I would like to say, I mean, if they're servicing 80,000, we should be able to get up to that kind of number over time. But now there will be smaller value transactions that typically will service those through channel partner programs and through our e-commerce channel, but that should be a source of growth on the systems side, which I think will skew that metric more towards 50-50 over time. And what would you think? Would you agree?

Yes, completely agree. I think just to add quickly, right, our strategy of system -- of Intelligent System Design and analysis, it's the like 3 concentric cycles -- circles, right? We are leveraging our core EDA strength, which is our roots. And then we are expanding that into the system design analysis space. Because as the world converges, you'll see more like multiphysics platform electromechanic meeting the mechanical side of the world. And Andrew, to your point, that's where we're maintaining the partnership with Dassault, right, in the PCB space because it's a very natural conversion point. And on top of that, we are infusing the AI/ML into our portfolio and Cerebrus is a fantastic product. We're very excited about it. We're now also launching a product called Optimality on the simulation side, basically combining simulation with the optimization. So with those 3 core tenets, that's how we are driving our vision and strategy forward. So that's how we foresee our business will look like in the next 3, 4, 5 years.

Interesting. And then in terms of the core business, the one you're known for, like, how is that -- is it still growing according to historical trends. This is all, in other words, the systems side and these other growth opportunities are on top of that?

I think that's growing stronger. I mean there's -- yes, I mean, back in -- like I say, going back to 2016, again, as a kind of a baseline year that back then, there were more EDA companies and more -- it was more competitive at higher process nodes. I think as we move down the process nodes that it's really become a 2-player game with ourselves and Synopsys. And I think the pricing environment has become much more rational. But also both companies have found their own identities. I mean, we have clear identities now than we had in the past. Like back in the '80s and '90s, Cadence and Synopsys were like the identical twins that everybody knew at school that wore the same clothes. That -- and it's kind of cute when you're a kid, right? That -- but we talked about it like we're in our 30s now. It's kind of creepy if you are wearing the same clothes as your identical twin. You have to form your own identity. But -- and with us that's -- Cadence was born back in the '80s as an analog EDA company. And Synopsys to be fair to them, they were like a digital version of Cadence. They are our roots. Clearly, over time, to become a core EDA company to service our customers, most of the design is mixed-signal design, you have to do both analog and digital. So we were going to encroach in each other's space. And we always felt that we would retain our strength in analog over time, and it would only be a matter of time before we got to 50-50 on digital. And the logic behind that is that like our engineers would tell you that analog to digital is a bit like orchestra to rock band. It's easier for an orchestra player to play in a rock band than it is for a rock band player to come the other way. So we felt that our destiny was always to be the largest core EDA company, and that would then be the center of gravity for our strategy. But -- and I think that's the way it's played out because we're probably 80%, 85% of the analog market. We're almost at 50-50 now on the digital side. We just signed a significant record new contract in Q1 with a marquee U.S. semiconductor company that I think will make significant inroads over the next couple of years to get us close to that 50-50 on the digital side. And I think we're -- like I say, we're still probably 80%, 85% of the analog market. So when your center of gravity is that you're destined to be the largest core EDA company, our customers, that the challenges that they have when they're doing chip design and chip verification, is like when they're doing packaging a board, if someone changes something on the package, and it alters the electromagnetic dynamics internally, it can alter the power performance in area that you get from chip design. And they want to be able to see these and try to do it concurrently on 1 multiphysics platform. So the natural thing for us to do is try to build that platform. I think we're the best position company to do that. By contrast, if you look at what Synopsys has done, I think they've done an excellent job. They've built out an IP business, and they've built out a software integrity business, which makes perfect sense for them because of where they stand in core EDA. But -- so I think what you find now is that 2 identical twins have now kind of formed their own identity and are kind of managing their own way in the world that we've never been more different in my opinion.

Right. Maybe just following up on that. You mentioned Synopsys went into IP more strongly. Why wasn't that attractive to you?

I think it's the right strategy for them. The -- yes, people say that it is not attractive. We do think the IP...

For you. Maybe for Synopsys it still is, but...

We do think it's attractive.

Yes.

It probably suits them better, but I mean they started a lot earlier than us. But for us, it's always been an investment choice, the opportunity cost of investing in simulation and building out our system analysis product portfolio. It's our highest margin business. Like if I look at -- if I break down the businesses that fall under the Cadence umbrella, IP is probably our lowest-margin business. It's significantly better than it was 5 years ago, but it's still our lowest-margin business. But in simulation, simulation is our -- I mean, obviously, software is going to be your high-margin business. And software at Cadence you could bifurcate into 2 types of software sale. There's interactive tools where every license needs a driver. Like so we have a Virtuoso franchise in the analog side. If there's a 100-person analog design house, they're probably buying 100 licenses at Virtuoso. We won't sell 110 until they hire 10 more engineers. So your growth there in Virtuoso is going to come from a combination of engineering growth or the number of -- growth in the number of engineers and pricing improvements because we've -- we're so big in the market there that we have significant pricing power with a product like that. But on the simulation side, if you have tools in simulation where one engineer can use 10 licenses or 20 licenses. And that is the most profitable part of software. So you could -- like the interactive tools are less profitable on software than the tools that can be batch processed by an engineer. So if one engineer can use 20 licenses, they could use 30 licenses or 40 licenses if they're given access to the right amount of compute power. And because of that, you can kind of have revenue go nonlinear in comparison to your expense base, so it tends to be the most profitable area for us. So when we have investment dollars and you're faced with the opportunity will I invest in IP and build out an IP portfolio -- a bigger IP portfolio, or should we invest to build out this multiphysics platform? I think this strategy is leading us down towards that. I mean, we're still investing heavily in IP but in select areas. We like to focus on differentiated IP. We don't want a broad portfolio of IP. A lot of -- I mean, there's a lot of players in IP and a lot of it can be commoditized. And there's no real pricing power in that, that you want to have differentiated IP. So you have a seat at the table for any IP discussions. But then after that, really our focus is on that section from chip design through chip verification, packaging and board right through to system analysis. We're trying to cover all of that for our customers because that's what -- that's the pain point for them.

Interesting. And then on simulation, I mean assuming it's your most profitable product, right? Just based on ANSYS and Altair, actually, you can see how they have evolved. Just on that topic and maybe benching ANSYS, you mentioned them twice, I believe. I mean, do you see them as a competitor on simulation at this point? And are there any other emergents?

We admire them.

Okay.

Yes. And Jim -- you got Jim here, Jim Scapa from Altair. I love Jim. But we admire those companies, I think they're really, really good at what they do. And I think we can all peacefully coexist because we're all trying to solve the same problems for our customers. But -- and there's no reason why you can't put all of us into your design flow. But -- but yes, we would very much admire what they're trying to achieve and the problems we're trying to solve for customers. And we know them well. The -- yes, the profitability profile of that business is very attractive to us. I mean we made a conscious choice when we entered that area that, hey, let's not be so disruptive that we disrupt the market because it's beautiful the way it is and the high margins that they have, and we wanted to keep that. So we've deliberately structured our contracts as well on the systems analysis space as recurring revenue contracts. So the vast majority of our systems analysis revenue is structured so that it's like our EDA software contracts where you take revenue over time. That's why we have such a high recurring revenue percentage. And the nature of our relationships with our customers is that we're like a strategic partner for them as opposed to just a supplier. Many of the C-suite will talk to us about what they're trying to do before maybe the rest of the C-suite even knows what they're considering. But -- and the -- like I say, when you're doing chip design and chip verification, then next steps, naturally, are the packaging and board and system analysis. So it's natural to carry on the conversation to cover all of those aspects of the design process.

Just a follow-up on the fact that simulation generally tends to be viewed as more resilient, it kind of fits your model, right? Do you provide any services like high-performance computing to your customers, like maybe they have some models they wanted like they really need a lot of processing power. Do you provide that kind of service? Altair does that a bit, so -- yes.

Anirudh was -- like a few years ago when we talked about launching, so Anirudh created the -- our CEO, Anirudh Devgan, he created the -- our Intelligent System Design strategy, and he actually did his thesis in Carnegie Mellon on system analysis. But he loves the area, he is very, very knowledgeable about the area. But what -- when he looked at taking Cadence down this path, he wanted basically that -- he thought it was a natural thing for Cadence to do expand into system analysis and solve the problems that those customers would have, like when you follow through that whole process. But super attractive for us [indiscernible] the question.

No, no. I was just curious in terms of the high-performance computing side.

Yes, yes. So when we launched OnCloud, Anirudh wanted to make sure that we were first at the cloud. We talked to him and said, "Hey, our customers aren't ready to design in the cloud. It's like it's really -- they want to protect their IP." And he said, "Well, they'll get there in time. We felt that it's not a case of if, it's when. And he said that, but if we get into the cloud because what's stopping an engineer using 20 licenses versus 5 licenses is probably compute power. So he said, let's make it as easy as possible for them to get access to that compute power. So when we launched our new e-commerce channel OnCloud, that's all on the cloud and it's easy for you to get access to whatever compute capacity you need. And we've got arrangements with all of the large hyperscalers and providers of -- cloud providers.

Great. Maybe I want to talk about your international opportunities. I mean, in terms of where do you see the greatest upside? Is it APAC? Is EMEA? Just maybe touch on that a bit. And then are there any challenges? I know you work in China, for example. Are there any challenges they face there? Are there any potentially emerging competitors that you might see there on EDA?

Yes. So big question. The -- so our geographic mix of revenue is consumption-based. It's based on where the products are used. We probably have about 45% of our revenue in North -- coming from North American usage. It's probably 1/3 now at this stage coming from China and Asia Pacific. Maybe 14%, 15% of that is China. That is about 19% then is the rest of Asia, probably 17% in EMEA and then maybe 6% in Japan. The -- but it's very, very hard to predict the geographic mix because the engineering community moves around. And like in China, although it's maybe 14%, 15% of our revenue, maybe there's probably 12%, 13% of that is sold into China and there is a couple of percent of that, that's outside of China, to like, let's say, if we did a deal with somebody in the Valley, and they decided that, okay, so 10% of these licenses are going to get used by the team we have in China, then we will represent that 10% of that revenue as China revenue, even though all the sales activity and everything happened -- we're even providing the licenses to the team in the Valley, and then they're putting it on their servers in China for use out there. And we have this call home technology in our software to track where software is being used. The -- what we've seen in the past, I think the customer consolidation that we saw about 5 years ago was quite informative in terms of how quickly our customers can move the geographic mix of their revenue. But we had one customer in America that acquired another international customer. And they decided to eliminate their analog design team in Sweden, and we thought that, okay, there was 18 months left to run on that arrangement and we had 18 months of revenue to go. And we thought from our forecasting perspective, okay, we're not going to get that. They've just got rid of that team, so that's not going to get renewed. What we discovered then was that some other customer picked up that engineering team, and then they bought licenses from us for that engineering team in Sweden. So for -- in 18-month period, we almost doubled up on revenue because we're still getting revenue on the old contract, and we were getting revenue on this new contract from -- when those engineers got picked up by another customer. And then although we didn't expect the contract to be renewed the customer that it acquired that entity did renew when the 18 was up because they were -- they had moved the design activity that was being done in Sweden, and they just -- they were doing it in North America now. So we found that geographic mix shifted. We also have -- I mean you mentioned China. But I mean, there's been geopolitical tensions in China, and we had customers added to the Entity List and you can see how resilient the business is. I mean, when you step back and look at what's happened in the region, we're still growing at a healthy clip despite being prevented to sell to some large customers. But what tends to happen is if you can tell to those customers, well, then the engineers at those customers, well, what are they going to do if they don't have the tools. So they end up going working somewhere else. And when they go working somewhere else, they need to buy tools for those engineers. So it's like when you look at our customers' R&D spend, you could simplify -- again, bifurcate it into 2 and say that your R&D spend is so much on employees -- on the engineers in your R&D group and in so much on the tools in the tool bag that you give each engineer. Back a decade or so ago, it was probably 92% spent on people and 8% of the R&D budget was spent on tools. But now it's probably something like 88% and 12%. We think over time that, that's -- I mean it might take multiple decades. But we think over time, that should get to 50-50 over time. And the reason we think it gets to 50-50 is that the pace of innovation and the pace of growth and complexity of electronic products and what we're all demanding as consumers of those will far outpace population growth. So the engineers -- number of engineers can't keep up. So you're going to need to leverage the tool bag more, and that's why we've been developing things like Cerebrus as an AI product because if you can leverage AI and machine learning more, you're getting more value from the tool bag. And the -- and it's quite interesting because the -- I mean Cerebrus is just an incredible product because it's taking -- remember I said about the software. There was -- you could bifurcate it into interactive tools and tools that where one engineer can use multiple licenses. So Innovus is our digital -- is a digital -- very popular digital product for us. And that's an interactive tool. But Cerebrus allows you to use Cerebrus as a cockpit to control 10 licenses of Innovus. And it's like the self-driving that Anirudh tells me that -- like if you imagine a world where self-driving cars are possible, you said chances are there's not going to be many people in the cars that you will have the driver sitting in an office like this controlling -- they might be controlling 10 or 20 Amazon trucks or Tesco trucks to -- around the country and you might have some kind of operational team to go and help get one out of trouble if something gets stuck somewhere in the grid. The -- and Cerebrus is a bit like that, that it allows one engineer to operate at a cockpit. So when you see trends like globalization over the last few years and what everyone has done like they moved to lower -- like offshore costs in terms of their R&D spend. But then you end up with supply chain issues that's because your supply is so far away from you. If you see any kind of trend towards deglobalization that's really expensive. I think inflation is one thing. I mean engineers in America cost a lot more now than they did last year. But if you were looking to deglobalize and take some engineering work from offshore and bring it back onshore to the U.S., that's going to be a significant increase in cost in R&D, so you're going to be trying to do more with less engineers. And the way you do that is you spend more on the tool bag. But -- so I think we're pretty inflation resistant that we should at least maintain our 88-12 relationship in terms -- like if the cost of engineers goes up or employees goes up, the cost of the tool bag should naturally go up with this. But I do think that the percentage of R&D budgets that will be spent on tools because I think we're naturally deflationary to our customers' R&D budgets. And I don't think it's -- the impact is going to be on people. It's just there's not enough people coming into engineering to pick up the slack. So I think it's just a natural thing that happens over time.

I guess -- yes, that's great. One question I have is like how easy it is for an engineer to adopt Cerebrus and AI-driven chip bus?

So you might come in, Richard. I know that Anirudh tells me that -- we were ready to launch that 18 months before we launched it and Anirudh stopped it because he wanted it to operate from what he called a cold start, but no. He will explain on that. That's a cold start, Richard.

So how it works is, obviously, we're very excited about the tool, right, the technology itself. And it uses sort of like everyone talks about machine learning and artificial intelligence, but this is truly using the reinforcement learning mechanism and algorithms to manage and automate via the full flow, meaning like we -- when you think about the digital full flow, you start with RTL synthesis and then you get the implementation place on routing, which is quite manual in the past, and then you got to sign off, right? So there's a lineup of products we launched or the different phases of the digital full flow. And as John mentioned, what Cerebrus does is it's a cockpit software, where it helps automate the whole thing. So there are 2 great benefits you can get out of this. Number one, you dramatically increase the productivity for the engineers, right? Instead of like manually going through all these permutations and paths given the parameters that the machine has done it for you, and it's done a better job on that front because from a human perspective, we just cannot exhaust all these different options, number one. Number two is you get better results, okay? So as a result of that, you get dramatically better PPA. We have seen 5x productivity gains for engineers, and we're also seeing like 10% reduction in power, which is quite beneficial from a customer standpoint. So still early phase, but our goal is we wanted that tool to be pervasive and proliferate through 80%, 90% of the customer base and products. And another great thing, as John mentioned, is also it's a natural pool, right? Because you need to have a lineup for the products in order to get a benefit out of that. So there are multiple fronts we're working through right now.

Have you maybe -- it sounds like this is a key pillar for your growth going forward. So have you disclosed like how penetration rates of any kind or how big the market could be for this?

It's still quite early. I mean, we launched it last year in July. But I think it's -- Andrew, if you look at the -- again, I came back on the strategy of the 3 concentric cycles -- circles, right? It's 1 of the 3 prongs of the pervasive intelligence, and we've launched this a year ago, and now we're applying that from the silicon side to the system analysis side with Optimality. So a lot to come. But I think from a customer standpoint, to see great benefits. One of the major customer was talking to Anirudh was like, "Hey, I wanted that tool to be everywhere in my company." So I think it's just a starting point of a major proliferation of technology across the board.

That's great. Maybe just one on partnerships. You alluded earlier to Dassault. I think you have one with Schneider Electric you signed in 2021. Is that right? What -- just maybe why did those vendors choose Cadence, first? And then -- because they're big companies. And then second, why -- what was the rationale behind those partnerships from your perspective?

Do you want to take that?

Sure. I'll take a shot at this. I think when you look at the partnership with Dassault, right, both parties are contributing to the partnership. It's a very highly complementary kind of relationship. If you think about the sole strength, obviously, mechanical they're coming down from the physical world, right, and from a design automation standpoint. And our strength has been the core EDA from its silicon chip design standpoint working upstream. And the PCB becomes a very natural kind of touching point and convergent for these 2 pieces to come together. And that plays right into our strategy of one single platform of multiphysics platform, where you can manage -- using digital twins to manage the entire kind of design automation flow from a customer standpoint, both digital and the physical and electromagnetics on that front. So I think that's how we foresee the partnership working. And Dassault is bringing, obviously, the 3DEXPERIENCE platform. We're bringing Allegro. Those best-in-breed products will -- we are forming that relationship. Again, it comes back to our strategy and our investment philosophy of we're not just going about these markets all by our ourselves, right? We're thoughtfully leveraging the partnerships and kind of expand our footprint that way.

Yes. And you mentioned Schneider. The -- I mean, in fact, I mean, we've all seen the trends that are happening in the world. I mean phones have become smartphones, cars have become smart cars. But in Tesla, when they launched the Dojo chip, they used Cadence tools to create the Dojo chip. The -- and buildings becoming smart buildings, cities becoming smart cities. It's -- I mean, Schneider mainly are in buildings, but the -- but I think there's a natural convergence of what we do and what those other companies do.

Excellent. We have about 5 minutes left. I don't know if there's any questions from the audience. Yes, in the back, sorry.

[indiscernible] like-for-like pricing [indiscernible]

Yes, sure. Great. But again, if I go back to baseline year for 2016, 2016, when we started it, but mid-single digits was like the expectation for core EDA growth. And we thought, well, that's nowhere near enough for what we provide in terms of the value that we provide into that whole process. And we had to get more constructive in pricing and more disciplined in terms of how we price things that we formed all kinds of deal quality metrics to measure quality and to improve pricing. I think we are more disciplined. We managed to get up into the high single digits. And I think of late, the last few years, we're getting into double digits and a little bit higher. But -- and that's been mainly this leveraging the tool bag more that when we deal with a lot of our customers, we show them how using our tools can save them money and then we try to partner with them. And like if you're going to save $0.5 million a year, can we get $200,000, $250,000 of that. But -- and then we continue to partner with those customers and get more benefit from that. I do think the pricing environment has gotten much more rational over the last number of years as well because we compete on the technology. And as the companies have become more different, there are -- I mean I met Prashanth from Analog Devices, the CFO there last week we were having dinner and he was telling me that, oh, he says the 2 companies have never been so different. He said it's the pain at times when they're trying to buy tools from us because there is no natural competitor for some of the tools, which kind of benefits all companies in our space. The -- so I think that trend has been -- we were in high single digits, but we've been overachieving that the last few years as companies have realized just how much more value they can get from leveraging a higher proportion of their R&D spend on tools versus employees. I don't know how far that goes, though. Does that answer the question?

Sorry, in the back, all the way in the back.

[indiscernible] what was that?

I think that was -- if I go back to kind of early days maybe 20 years ago. Yes, yes. I mean it's been slow. But I think it was slow because there were a lot of EDA companies back that time and competing on pricing. And as it came down to a smaller number of players, I think pricing has become more rational. And I'm measuring that against like this traditional semi R&D spending compared to, say, EDA revenue for all the EDA companies that were around at that time. Well, back then, there probably wasn't a lot of systems in there. So...

[indiscernible]

Yes, actually, it would probably include systems. Yes. Because I mean just looking at our -- I'm taking our revenue in the industry and comparing it to, again, kind of percentage of semi R&D spend.

[indiscernible]

Yes. That's a good question. That's -- because we've been working through how to price that because we have a pricing committee and a marketing group that works with them to try and trying to establish what the right price should be. They're trying to establish the value that a customer can drive from it and then take a percentage of that. The -- but there's a lot of pull-through business that comes from it as well. Like when you -- when someone buys Cerebrus, they have to buy like 9 extra copies of Innovus. So we're seeing some pull-through benefit from that. But it's early days that right now, we're kind of in a proliferation stage and getting it out to customers. Anirudh shared it with one customer and asked them to try it out. And within 2 weeks, the CEO called him and said, "I want it everywhere." But -- so it seems to provide significant benefits to customers. The challenge, of course, is that I think when -- we had the team produce a whole bunch of metrics to try and come up with the pricing, but the sales team's feedback to us is that the most significant comment you can make when you're selling it is to highlight that your competitor is using this, but -- because if your competitor is using AI tools, you probably need to use them quickly. Yes.

Yes. Sorry, last question. I think you had one at the front.

The recent Harvard Business -- there was a recent Harvard Business Review article on labor costs [indiscernible] saying that on the chip itself the labor costs have enormously reduced. [indiscernible] not had any major efficiency [indiscernible] and now the industry is depressing that. That is a major tailwind to your business?

So the question was that there was a Harvard Business Review suggesting that there's not been a huge increase in so we -- can you repeat the question?

There's been as labor costs working on the silicon itself [indiscernible] there has been no decline in the labor cost ratio of services on the [indiscernible].

Right. Right. Yes. Yes.

Is that a major tailwind [indiscernible]

So I think any trends towards lower labor -- labor becoming a lower percentage of the spend is positive for us because we have most of our revenue coming from the providing tools and services. If it remains the same, then it's not really a tailwind for us, I would say.

Yes. [indiscernible]

I think we're out of time. I'm sorry, but I appreciate the questions, and thanks very much, John and Richard, for coming.

Thank you.