Synopsys, Inc. (SNPS) Earnings Call Transcript & Summary

Okay. We're going to kick off the next session. Rich Valera, I cover EDA and Technical software for Needham, and very pleased to have as our next participant the founder and now co-CEO of Synopsys Aart de Geus. And we're going to kick off with Aart sort of giving his State of the Union as it were for Synopsys in the industry, and then we're going to go into some sort of interactive Q&A. So with that, Aart, thank you. First, thank you for coming and look forward to hearing your thoughts here. Well,

Thanks for having me, and it's a shortened State of the Union bottom line is, I think, Synopsys is in really good shape. We had an excellent last year. And hopefully you saw that we gave strong guidance for 2020 with an objective to drive EPS return to mid-teens and $3.6 billion to $3.65 billion in revenue and ops margin, another 2 points up to 27%. And so I'm quite enthusiastic about the year. Now one warning, please don't read my body language today. I arrived this morning in coach, mid-seat, red eye with crying baby. And so as I look at the carpet, this looks like a bed. I will try to keep the energy up, but there are some physical challenge here. Be it as it may, we are also surrounded, I think, in a quite confident market, all in all, as you all have heard, hopefully, tomorrow, we'll see some resolution Phase I for China, but the overall market remains strong, but more importantly, the semiconductor market has a fairly high degree of confidence. You can see that last year, it had some big challenges right now, the numbers look up and the expectations look up. And most importantly, I think it fulfills what we have been talking about now for a number of years which is that every vertical is clearly embracing massive data and machine learning, and with that comes an enormous push for more compute, more storage, more connectivity, more sensing. And with that, the ongoing push for more of Moore's Law. You would say, Moore's Law, isn't that slowing down? Yes, maybe economically, it is. At the same time, I think there will be plenty of money to push it forward. And as a matter of fact, we are engaged right now very actively with 5-, 4-, 3-, and 2-nanometer technologies already, which is astounding in many ways. But maybe more interesting is the fact that if you wanted to have really great progress on machine learning, there's going to be a lot of invention in new architectures. And that is precisely what we're seeing. A lot of companies inventing and new opportunities there. If I look at our position in EDA, I think we're doing particularly well at this point in time. You may remember that over the last 5 years, we have invested in this notion of the fusion, meaning taking many individual products and smooshing them together at the algorithmic level and data representation level. And we introduced the first big product, pretty much exactly a year ago, Fusion compiler, and now a year later, as they go through the being benchmarked, being tried -- being tried on a piece of a chip, being on large chips and plan of record, it is moving forward fantastically well. I think we have won every benchmark of comparison, and we have already a number of large customers that have moved it into plan of record, which means they will continue to design with it for many years. Strong results also in verification. And this is an area where we had bet on the fact that the intersection between hardware and software is going to become more and more important. And indeed, that is the case. So the ability to verify the software in the context of hardware that is prototype, therefore, doesn't exist yet, is very valuable. We are a leader in that area. Doing well in deep technology also, the ability to model transistors, literally at the atomic level is turning out to be quite valuable. And so this gives us both insight and a push into the next generations as the number of advanced nodes continues to proliferate. And lastly, excellent progress also in custom design. We've always been strong in analog mixed signal verification. We are now also growing in the design side. This is especially true in the advanced nodes because we decided to not compete with an incumbent that is very strong in older nodes. For the new nodes, it's an open game and new capabilities are needed. Looking at the IP business, which we communicated has now passed the 20% mark. So we have substantial business there as well. This is scaling very well and clearly heading towards the $1 billion mark in the future and it's a difficult business. Difficult because these are extremely complex blocks and the word difficult is actually great for us because Synopsys has learned how to not only do very advanced protocols, but also move those to the most advanced silicon technology nodes and most of our customers no longer can do this themselves. As a matter of fact, many of the new customers, such as IP companies or hyperscalers, buy a lot of IP because they right away start with the building blocks that are available, not wasting time on them and using their precious resources and creating new architectures and looking at how they add value. So this will continue to be a push for the company and the areas where we see most growth is AI, cloud, automotive and 5G. Lastly, our business in software, Software Integrity. It had a milestone. 5 years ago, we started from 0 with a number of acquisitions and R&D investments. And this past year, we crossed Phase I, finished Phase 1, which is 10% of Synopsys is Software Integrity. And moreover, or somewhat surprisingly, we managed to get it to 10% ops margin. And so we had pushed very hard. In addition, towards the middle or end of the year, we introduced a platform that takes multiple products and brings them together. And so clearly, we're now entering Phase 2, which is looking at how to structure for growth, how to cross the $0.5 billion mark and ultimately aim also in the billion direction. Make sure that we've delivered the entire platform with more and more of the products integrated well. And that we are continue to scale the structure of the organization for the long-term growth. So in summary, we have a clear objective for this year to continue to optimize both the ops margin and the growth rate. And to that effect, we have adopted quite significantly throughout the company. The Rule of 40, which is sort of an easy mnemonic device to say, you add up the growth rate of revenue and the ops margin. And if you're at 40 or heading towards 40, you're heading in the right direction. Try to apply that everywhere in the company, and I dare to say that the vast majority of all Synopsys employees understand exactly that moniker and what to do. And so we're heading in a good direction. So I think good shape, a lot of hard work ahead of us, but we're well on our way into 2020.

Thanks, Aart. And as you noted, you put up some strong numbers last year, and it certainly wasn't with sort of green lights across the board from a macro perspective, and I wanted to drill into two specific headwinds. One was, as you noted, the semi industry revenue was down round about, I think, double digits is where it looks like it's going to come out. And understanding, historically, your revenue is much more tied to design activity than actual semi revenue, do you think that at the margin created some kind of headwind. And conversely, does this year if, in fact, the semi market is up the way folks are projecting right now, does that create a little bit of a tailwind. And then similarly on the China side, well documented that several of your customers and one in particular, was put on the U.S. entity list. You had to stop shipping to them. Why it creates some tough compares. How do you think of China now going forward with maybe some sort of detente or just in general as an opportunity going forward. So I just wanted you to address those two?

It feels like a 17-part question.

Sorry about that.

Actually, an excellent point, because I always like to graph these highs and lows, and it is quite surprising how far down the semiconductor industry went last year. But in all fairness, a big piece of that is due to the pricing of memory. And so that doesn't touch to the supply chain all that much. But having said that, yes, you always want your customers to feel well and do well. And actually, they do feel well because I just yesterday saw the confidence index that GSA puts out. And it's well above the average and moving up rapidly. And in general, the expectation for this year are more positive on the semiconductor side. Regarding what that does for us. In general, during these down cycles, Synopsys does particularly well for two reasons: a, we cater to mostly the R&D part of companies. And so even in downturns, no semiconductor company will lay off R&D guys to then hire them back in 12 or 18 or 24 months later, makes no sense. You continue to invest in the future. Secondly, our revenue recognition and is quite ratable, so that hides some of the up and down. But more importantly, we have 3-year agreements. And so we navigate through these. And even going back to the '08 '09, that looks actually similar, but it was deeper and bigger. Synopsys managed to go through this on a yearly basis with actually no decline. And so I think we have a very high degree of stability. This is important because that allows us to continue to invest, which we've done now for many years in the new technologies and ultimately, our field never gets easier. And again, I say that with the glint in my eye because this is precisely how we differentiate ourselves. And after 33 years or so, I think we can honestly say that for 80% of our products, we have always been the state of the art, state-of-the-art tomorrow is hard again. But silicon technology continues, 3D ICs are coming on the horizon, new architectures, all of that will keep moving. Now you open up another aspect of the market, which is China. China, in general, has continued to do very well for us. We are all aware that last May, the entity list picked a number of companies to whom high-tech who no longer ship that included Huawei and a few others. That included us. And just to be clear on the record, our guidance for 2020 is based entirely on the assumption that that stays. The status quo doesn't change from an entity less point of view. Maybe with the agreement tomorrow, there's at least a little bit of cloud removal, and therefore, more discussion, but it's a complex matter, and so we're not counting on it.

Got it. I wanted to follow-up on another topic you brought up in your prepared remarks, which is Moore's Law. And sort of -- it's been the death of it has been predicted for quite a while. And as you noted, there's still a lot of nodes to go, and it sounds like maybe there's also sort of a pivot going on away from just smaller, faster and more towards architecture. If you can just maybe elaborate on how that still creates lots of opportunity for design essentially.

Sure. I think the death of Moore's Law has been predicted literally, I was in grad school, and I saw it happening. And so I think the lesson out of that is never underestimate: a, the power of invention, but also never underestimate that something that works really well, has an enormous economic push. And if there's one thing to me, that is really worth highlighting is how much after maybe 55 years or so of Moore's Law, which is a technology push, there is now an economic pull for it. Meaning the benefits of bringing machine learning, big data utilization to all the verticals are economically so high that the hunger will be unsatiable. Therefore, there will be continued push on Moore's Law. Please do something with that, but that's not sufficient. So the other something is, what about if we could put multiple chips in some package format that is efficient, close enough, high bandwidth, and most importantly, can manage power, refer to a 3D IC. I think that's going to continue to grow. But the most important one is the one you mentioned. If all of these techniques can get you, let's say, for argument's sake, a 10x, what machine learning wants is 1,000x. If we just had 1,000x. Unbelievable stuff would be possible. Therefore, it will be tried. And it's been remarkable to see and we, Synopsys, benefited from this. How many IP companies are designing a chip. They describe all the chip in the same way. It is the most amazing ever great architecture, fantastic chip that we're designing. Every single one is like that. And of course, they're going to compete like crazy, and the race will be on. Some will succeed, some will fail. There will be a recombination. But the reason this will go on for quite a while is that while we have seen some fabulous successes with AI. We have also seen that the human brain is orders of magnitude further along especially in its ability to compute versus the power utilization. Therefore, here's the goal, right? We sort of have these things in-orbit now. Here's the moon between sputnik and getting to the moon is a road, but it is very clear where the moon is and that's the human brain. And so that will continue for a long period of time, and it will be utterly supported by the financial benefits in so many verticals, be it from health to agriculture, to you name it.

Got it. So we're sort of talking about taking the conventional semiconductor technology and continuing to sort of push forward with that. There's a sort of branch developing on the side, quantum computing, which can -- promises subatomic or at least getting down beyond the -- where you can go with conventional semiconductors. What's your thoughts on where quantum computing can go and how significant it can be?

Well, I wouldn't quite see it as a branch on the side. If we said, okay, we're going to introduce some radically different materials in semiconductor. That will be branches. And the reason you would call them branches because it benefits from the entire experience base of the past. Whereas quantum computing, while having just a fabulous and outlook potentially in the future, that entire base of experience needs to be built. And so I do think that there will be continued investment because if you can still dramatically increase computation, including the fact that you do many computation at the same time. And so it essentially attacks the very notion of 1s and 0s. So the 1s and 0s and everything in between, right? Crazy notion but very powerful if you can make it work. I think there will be continued investments in that. It will take a while before that has impact on the products that we're going to see.

Synopsys has had some exposure to that. What you currently have today in terms of any kinds of tools or services that can facilitate that?

Well, we are right now designing actually almost finished a design system for Josephson junction. So those are our forms of semiconductors that work at very, very, very low temperatures. Not as low as where quantum computing is, but these are all steps in a direction that ultimately are puzzle pieces that will be assembled for these new types of designs. Again, these are -- these would be called research projects, but with a strong engineering sense of the company of do stuff that if you can make it work, we can make it practical. And I think we're making good progress on that.

That's great. I'm going to pivot now to a completely different business, your Software Integrity business. It was another big year for Software Integrity last year. You delivered your first platform. You had significant profitability. I think double-digit ops margin for several quarters in a row. So kind of give us the state-of-the art of Software Integrity and the opportunity you see for that business. And by the way, maybe you can talk about the acquisition you just made in the dynamic sort of verification space?

Well, let's take a step back. So why did we go into that in the first place. And there were really two or three reasons. The first reason was already about 10 years or more ago, I'd observed that the semiconductor guys, had more software engineers than hardware engineers. And when you ask them, so how many? They say, "Oh, 75%". And they would immediately say, and our customers don't want to pay for the software. And my conclusion out of that was, hey, they don't know it yet, but they are selling functionality, the wrapper happens to be called silicon and a very sophisticated wrapper, but it's the whole thing that is actually the deliverable, which as a corollary says, well, what if the hardware is really good, but the software is sort of a little flaky, the quality of that can ruin the product. And so at that time, we started to put a lot of emphasis on quality of software ourselves because we were growing complexity. And we decided, "hey, what if we go in that direction because we're knowledgeable about software. And it's fundamentally the same verification techniques that we are knowledgeable about." At the very moment where we decided to go in that direction, another factor came in. Security, meaning a much bigger danger even than bugs came up, which is proactive bugs, being hacked in some form or another. And it became very quickly clear that the -- between hardware and software. Actually software was the weaker link, and actually, that's where most of the hacks are not that hardware hacking is not possible and isn't done. But ultimately, it could jeopardize the combination of the two. And so from there, it became an opportunity from business lines to say, can we create a third leg to Synopsys EDA and IP and so on. So a certain diversity. You have to keep it coherent. It's in the same problem space. It's the same competence space. It's the same difficulty level. We can maybe make a contribution, acquired Coverity initially and then a couple of other acquisitions and put a lot of R&D in it. And so that's how we arrived there. Now as you observed, platform is essentially the saying, can you do something better with a set of products together? And the better can be both technical, it can be an efficiency, but it can also be in terms of management of software engineering. And many IT managers that have been confronted with this wave of hacking and security issues and buy this product and try that one and this one and that one. For large companies, this is unviable. A, you cannot rely on companies where you don't know if they're going to be around for a while. Independently of how good they are. And b, you cannot manage all these slices. And so we saw the opportunity of the Polaris platform as something that we can make available at the IT management level that can be used both on the web or in the cloud as well as domestically inside of companies and can be looked at bottom up as really easy to use and top-down as now I can monitor the progress on security in a more aggregated fashion. Now having said that, now we're adding inside of Polaris, the different products we have. And the small company we just bought is a dynamic assessment company. So it's one more of those little slices that we intend to plug into that solution, gradually making it better around it and actually being directly responsive to what some of our customers were asking for.

I think perhaps related to the platform now that we've got a broader solution, and I think tying back to the services related acquisition you made a bit ago. It sounds like you've been able to raise your average sort of sale and make it maybe a bit bigger, more strategic sale than historically was the case for that business?

Absolutely. And actually, that's one of the key things to track. And in many ways, that is also what we learned doing EDA for many, many years, which also started as point tools and also started as small transactions and then gradually grew with the big customers. This, however, is a little bit different than the EDA market because there are so, so many more customers, many of which we've never even talked to or they themselves know yet that they should talk to us because they have not had the security issues or quality issues or not visibly so. And so this is one of the reasons why, as we now look at the next phase of this part of the company. We need to think through how do we balance well, the moving the customers to gradually build commitment, so the higher dollar transactions. While at the same time, have enough seeding effort with all the companies that just need to get to know us. And where very small transactions start the interaction.

Got it. Continuing on the theme that your biggest customers, even ones that are nominally semiconductor houses are doing lots of software development, sometimes have more software engineers than others than hardware engineers. You have a business, hardware-based verification which has, I think, gotten significant traction within that sort of software side of your customer base. Can you kind of talk about that business? It's obviously been a strong one for you in the past few years. How it's evolved? What investments you made in it that made it successful? And what your outlook is for that business?

Yes. So maybe let me use an example to sort of describe what it is. So specifically, let's take the automotive industry because it's an interesting one because 7, 8 years ago, they had no interest in super advanced technology. And now suddenly, they are supposed to be autonomously driving. And that's been a shocker to the OEMs, the OEMs are the automotive companies, they typically work through the Tier 1s that provide subsystems, who in turn work with the semiconductor guys that provide chips. And of course, we knew the semiconductor guys well. We worked somewhat with the Tier 1s and so at the right time, we're able to engage with some of the key driving automotive companies with the following proposition, which is you now have to develop much more sophisticated electronic systems, on that runs software and you want this all to go-to-market at the same time because if you wait for the software, that car is going to take another 5 years to get there. And so the value proposition was, what if we could, on a computer, maybe using hardware acceleration or not, create a model of the hardware before you have the chips, before you have the system and already run the software on it. And moreover, you could send that model to your suppliers, develop the software, check it out for functional safety for other things and then bring it all together, just-in-time to then be able to be assured that it will work together without having surprises at the last minute. And that progressed really, really well. Now incidentally, that very same notion applies to many other products, too. Maybe not for the same time to market reasons that are very long in the automotive industry, but if you do a cellular phone, and you would like to know if certain applications are going to tap the battery like crazy. There's no reason you cannot have a model. You run the application, the software on that model, and we have now a capability to analyze on the basis of estimates. How much power will it use? Well, this is all essentially part of what we call a shift left, meaning know things earlier and find out bad stuff earlier. So when it's easier to fix or optimize. And that's why I think that intersection between the hardware and the software is so interesting. And by the way, there's no reason that one could not ultimately also trial security things on this. So what traditionally in our field for chips was quality of results equals primarily performance and power. Now you will have to add functional safety for cars. You may have to add reliability over time. You may have to add various forms of security. So the number of demand is growing. In other words, we've moved from Moore's Law scale complexity, more transistors to now systemic complexity. This angle, that angle, it all matters together. And that is a fabulous evolution for us as a supplier because it's difficult stuff. And you have to know about different domains, and that is sort of what we're aiming at as a company.

And if you go back a few, I'd say, several years, people thought of that business as fairly cyclical. You've seen several years of growth in that business. Do you feel this dynamic with using it in the sort of software bring up development realm has changed that dynamic?

I'm glad you bring this up because today, and just on the one-on-one, we must have had 3 or 4 questions where words cyclical came up. And I'm rebelling against this word because I feel that in our domain, this is long gone, long gone because what cyclical was, it was demand goes up, supply cannot hold up. And so something comes crashing. Prices go either up and then somebody layers in a lot of supply, at which point in time, demand is fully satisfied, prices come down. And so you get the cyclical behavior of the semiconductor industry, and was quite remarkable throughout the '90s. How every time that DRAM pricing went down. Samsung actually laid in more capacity, so that by the time it would move back up. They already satisfy. It was -- it became an art on how to play the cyclicality of those markets. Now we don't have that situation in the semiconductor or barely have it because there's substantial capacity. And so you don't get the ups and downs because of that. You may get ups and downs for other reasons, but not really technical cyclicality. Now you say, well, but is it not true that the nodes change? Yes, that is true. And they change at periodic places, but because they're so advanced, they don't get adopted by many people in immediately. There are one or two companies that really bet their future on the advanced nodes, and invariably, they're really good at it, and then everybody else watches. If they don't keel over, the next ones come in. And so on. And so it's much smoother. And then on top of that, all the things I mentioned earlier regarding multiyear deals and ratability. Cyclicality, I don't see it.

That's great. With that, we have a couple of minutes left. I was going to see if anyone in the room wanted to ask a question.

Just a follow-up on China. You mentioned that [indiscernible] was on the list. How many companies do you have on the list in total?

I don't know the answer. And that tells you that I think there were maybe two more that mattered somewhat. And then a few small ones.

Okay. [indiscernible] do you see impact on the business? How would you sell [indiscernible]?

Well, we don't characterize it. One of the things that we have said is that China is less than 10% of our business. And that we are doing -- continue to do very well in China. So there's a lot of other business.

[indiscernible] area is a set of different sub-verticals. But do you think the industry will ever consolidate them?

Oh, I think -- when you say this industry, are you talking about EDA or semiconductor?

EDA, simulation, [indiscernible], all of those things that are virtual representations of testing in the physical world, will that ever become more [indiscernible]?

Yes. That is de facto multiple questions. Because if you talk EDA, I would argue, a large amount of consolidation has happened. We were proactively acting on that. In EDA, we must have done 80-or-so acquisitions. And it was a direct reflection or preparation for what happened with customers that consolidated as well. You may remember in the late '70s, people were saying, hey, this is the end of EDA because when you have 3 customers left, what are you going to do? Well, it turns out, we're doing just fine. And -- but the consolidation is also a necessity because in any field that continues to move very fast in technology, but simultaneously relies on the experience base of every lesson learned, you need to have critical mass to be able to support that, which I think the EDA industry has today with relatively small number of players. Now by the time you say, the modeling of many physical things. Now we have opened up a space that is so large because it can go from chemical reactions to mathematical functions of how kids grow or what have you and all of that exists. Now there's certainly some companies that are particularly focused on 3D simulation, for example, in various forms. And this would come to mind in that context. And actually, we have a very good working relationship with ANSYS because there are some pieces that they have that are precious to us, and we can use inside of our code. So that's a good way of saying this, that's not a consolidation, but that is only a positive alignment that has worked well for us. But the notion of everything will be modelable and stimulatable, I think, is certainly part of our forward looking horizon?

Can you talk about the Rule of 40. Can you look at product level or solution level for a company?

Well, formally, we measure only at 2 major levels in the company. At the same time, we try to apply the concept wherever we can without having sort of the fiscal rigor, that one would have by doing a complete P&L because the P&L, there's only one P&L for the company, but the notion is really understandable by everybody in the company. And so part of saying, how do you commit to making a company more efficient, it is by having everybody understand what that means and not everybody meaning, oh efficient means you're going to cut my job. Instead, it's the opposite. No, no. If we can grow faster than we layer in expenses, we improve both the revenue growth and the ops margin. And as people started to understand that, it's become a much more participatory and constructive effort. And for sure, I would say, the top 3 layers of Synopsys are completely involved in this.

You can -- maybe one more, and then we'll have to wrap it up.

Yes. I am going to try and make this quick, but you know the old days, we used to talk about design from CAD simulation and when we were talking about that being involved in it knowing things earlier and adding internet safety and all of that separate. It almost sounds to me like you're becoming more of a partner to a lot of these start-ups and then trying to bring new concepts and new ideas to the market. What are the margin implications of that? Will we calculate more like partners? How do you make sure that this does not negatively impact the margins?

Well, for starters, I would say, many of these partners, and I like that word because I think it's imminently important in systemic complexity that you partner with others. Many of these partners are not young companies. They're old companies. They're well established. So the whole functional safety to take that. The entire automotive industry has that need and has actually a pedigree in that, but it's just now learning how to apply to these new hardware, software systems. And that's where we are actually -- thinking co-thinking leader on how to do that, but rely on their experience. And I think that will be the case for all the verticals. There's going to be specific rules for health that are particular to that industry. We will have to design those into the systems. Now there are a lot of new startups, specifically on the architectural side. And those people actually, interestingly enough, they're narrowly focused because what they're trying to do is say, here's my AI architecture for this type of algorithm because I only do this type of algorithm, I'm 1,000x faster. We want something more general to only be 100x faster, 10x faster. So in general, I would say there are more partners that need to work together. And I would argue for the entire world right now. Codependency is actually both the necessity and opportunity for the future.

That's great.

[indiscernible] do you then end up with modules that you can reuse...

Inside of Synopsys, we absolutely do that. Yes.

Thanks. With that, we're going to wrap up.

Thanks for the questions.

Thanks very much, I really appreciate it.