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

Good afternoon. Welcome to this session. Really delighted to have the team from Synopsys join us, Trey Campbell, Head of Investor Relations. I'll -- typical fireside format, I'll go through my questions, but please feel free to raise your hand. And before we get into the fireside, just a quick statement from Synopsys. Today's discussion may contain forward-looking statements related to their current outlook, expectations and beliefs, subject to certain risks and uncertainties that could cause actual results to differ. Please refer to Synopsys' most recent SEC filings for a discussion of risk factors that may materially affect these statements. So with that exciting thing out of the way, Trey, welcome.

Great to be here.

Yes, really appreciate you being here. I know a lot of macro cross currents. And I really wanted to spend the bulk of the call on kind of the longer-term drivers for what's a very high-quality industry and business. But just at the start, let's get some of the recent headlines, right, out of the way. So Synopsys, like all of your peers received some notification from BIS. Maybe give us a sense for what that letter says, right, to the extent you can and then what it means for Synopsys?

Yes. So thanks, everybody, for joining. Great to be here. So we had our Q2 quarterly earnings last Wednesday. It was the first time I've gotten to do earnings and then update the guidance the next day. But yes, so we had heard some rumblings in the industry. We had not received a letter from commerce, but we heard that there were and we saw some of the chatter on X and those kinds of things. So we delivered our quarter. And then the next morning, I guess, maybe they were listening to our call that we hadn't received a letter, we got one. And so if you -- we 8-Ked basically the paraphrase kind of comments of that letter. If you look at the cadence letter, it's pretty much straightforward. The letter is pretty as written in their 8-K. But it was something where there were -- it's a 2-informed letter, which typically is to stop doing something. There were a couple of broad codes through the period of regulatory restrictions. In China, there are a number of -- there's classification for software and hardware and chips and all manner of things put into different codes to check the effectiveness of those. We were informed to stop selling along this. So we basically worked through that day and have kind of stopped shipping to China. And one of the things it was a surprise in seeing that letter. Typically, when we get -- we'd worked through quite a few restrictions in the Biden administration and even previous to that. Typically, there's a anywhere from a 4-week to a 12-week comment period, which is valuable because you can imagine there are -- these are complicated technologies with a lot of interconnections, things like software versus firmware versus all of these things that matter when we're selling. And so we stopped shipping, but we've been having conversations, and I assume peers have been doing the same thing to kind of after the fact, try to create that comment period to understand what was the intent of the regulation, how do we follow that intent, but are there elements of our business that can still be sold that are not a part of the strategic intent of the regulation. So we've been working through that. And we pulled our guidance, which we thought was appropriate, given this is -- could be a material impact. And then we'll put new guidance up there as we work through this with the government and understand what is off limits and what could be sold. And so that's kind of what we're working through right now.

Got it. Is there a time line, Trey, to when we should -- we might hear of an update?

There's nothing I can give you as a point. I mean we're working vigorously on it.

Is it a weeks' thing? Is it a month?

Yes, I don't think -- I mean I don't think it's a long-term thing, but it's going to take -- it takes a little bit of time. I mean we have -- you think, oh, we only have a few products that we sell, but these are broken up into thousands of subpart numbers even in software of how things move together. So we have to be -- we want to be really solid in looking through that and making sure that we've really checked through it with our legal teams, work through with the government. Interestingly, we compete vigorously with our peers in this industry. But this is one space where we do work together as legal teams and as government relations teams because we don't have the scale of Apple and NVIDIA. We want to be a combined force. And so I know that the rest of the industry is also working through those.

Got it. Why would they single out the EDA industry versus the equipment industry, right, which has almost like 30% exposure? Like what is the difference between the enablement that as an industry, EDA is giving to China versus what the equipment companies are doing?

I'd be speculating, I mean, in terms of the intent. But I mean, usually, there are usually a couple of things that come up with administrations in terms of -- when you think about restricting, there are 2 significant choke points. One is always EDA, one is cap equipment, as you mentioned. There -- it certainly can impact the ability to go get things done in China, but they're also one of the reasons you have to look at how you do these, these are high collateral damage areas, too, in terms of the restrictions that we've gotten historically have been targeted at AI. And so this all kind of started in 2019 when Huawei was banned, and then you had a number of other players in the space that were banned. And then you kind of went on to a technology restriction kind of schema where initially it was gate-all-around. So you basically were taking away the most advanced process nodes, which were highly used by AI and HPC chips. Then there was a restriction on the performance per square millimeter that you can drive. So hey, if you could manage to cobble it together without 2-nanometer or 3-nanometer, you could go do -- you'd still have this restriction. And then the final leg of this was in November when high-bandwidth memory was restricted. And so the combination of all of those technology restrictions have had an impact. And I just -- I mean, I'd point to pre-restrictions China geography was growing kind of 20%, 25% for Synopsys. In the most recent quarter, we were down 28% year-over-year, and that was about the same thing for the first half. A lot of the impact is those technology restrictions have meant that a lot of the AI and HPC chips that were getting developed may die on the vine before I just say, "Hey, I can't have a performance solution relative to what AMD or an NVIDIA can offer from a merchant." And so that means, hey, the IP doesn't get used. I don't need the hardware to verify it, even if we're continuing to sell EDAs on recurring contracts. So less kind of flexibility in that. But that's what we saw. And so we've been working through all those. As a company, we 100% are going to comply with everything that the government comes in with this. Same thing with our peers. And so -- but this one, it was atypical because there wasn't really a comment period, and now we're working that comment period after the fact.

Got it. Historically, what's sort of been the exposure across applications in China? Is it like half is "High-performance computing," the other half is mature node washing machines, trailing edge type stuff? Is it like half and half? Is it tilted more one way or another?

Yes. I don't -- I mean I don't have that kind of rough proxy, but I can tell you that there are a lot of chips on lagging edge, but I mean, we get paid a lot more in advanced node, right? Because these are the chips that are going to have the high-end EDA, use more IP and newer IP. Those designs are the ones that require hardware verification where you may be able to kind of do it without on lagging edge. So our kind of business was skewed definitely toward the advanced node in that space. But what's happened with the technology restrictions is we've -- as the AI and HPC TAM has compressed with the restrictions, we've moved more into automotive and IoT and industrial, and those have been really good markets. The challenge is they weren't as big from a revenue TAM perspective as the AI market.

So if we were to just do a very, very high-level breakout, is it like 1/3 high end, 2/3 trailing edge as it as...

Well, I mean, it really is whether you're looking at design starts or the revenue TAM out there. I mean there's still a lot of design starts at lagging edge because I think 25%, 30% of lagging edge is in China. And so there's a significant amount of kind of 16-, 14-nanometer and above chips that are made. However, our EDA IP intensity is much higher at advanced nodes.

Right. The reason, obviously, we are focused on that is because the investment community is trying to get some rough sizing of if there were to be, right, some kind of additional scrutiny, right, would it apply to a quarter or a half? Like what is the right way to kind of...

Yes. This is the one where right now, we're just really trying to work through a little bit of the clarity and to put more fine-grain understanding on what we can and can't sell. The other piece is it really gets down to duration. So -- and this is how long will these restrictions be in place? Will they be narrowed? What's the intent? How does this play? So it really -- a lot of the questions for us, we don't want to make dramatic changes to our business if something is very fluid. So I'm sure we'll get it to the Ansys question. But I mean, there's lots of knock-on implications of this of if this is a short-term thing, a lot of the question I've gotten today in one-on-ones and after the call, oh, well, maybe there's a glass half full, maybe you guys can close Ansys this week, right? And that -- it gets back to a duration question of would you do something like that if this is a short duration thing where we don't want to lose out on a market we really want to participate in, in China. I mean back to -- this was a 20%, 25% growth market before you had the restrictions. If you reach a trade negotiation or something that has the right controls but restores our ability to sell, we certainly don't want to walk out of a growth market like that.

Of course. Makes sense. And what's also interesting is that we are seeing all these headlines in the midst of just U.S. and China trade negotiations. It's hard to know because you also mentioned that historically, the industry has always been given notice. There's been a comment period, right, of several weeks, but this has just sort of come out of the blue right in the midst of these trade negotiations. Is there a chance that it is perhaps part of that bigger negotiation. It's not like singling out?

Yes. I mean I think there's a lot of conjecture on one way what it is. I certainly hope that it's a short duration kind of thing and that we can work our way through it. But I'm not on the right side of the government to answer that.

Understood. And then finally, in terms of the spending that is associated with this revenue, if we were to say hypothetically X percent of this revenue goes away, then is there some OpEx profile associated with it that can also go away? Or you think that OpEx is fairly fixed?

Yes. I mean there are certainly some variable elements of OpEx. I mean if you were in a situation and you said, "Hey, this is written in stone and it's permanent." Well, you'd look at the revenue driving kind of sales force and those things. But if you look at our business and just software in general, it takes a certain amount of resources to go put out Fusion Compiler in our key EDA tool or to develop IP blocks and these kinds of things. So regardless of how many geographies I can scale my revenue, it's a great thing for my P&L to be able to do that, but I still have to invest the right OpEx to go get these done. And so that's why -- no, I mean, we'll certainly look at all opportunities to be more efficient in OpEx and all those things. But it does take a certain amount of resources to go put this tool set out there.

Got it. And then finally, you mentioned that -- sorry, yes, Nick, please.

Yes. Just when you say you stop selling to China, does that mean you stop sending software updates? Or can they continue to operate on the existing software they have? And if they are operating on the existing software, how long can they operate...

Yes. I'll give you kind of the schema and the short answer is it depends a little bit. But I mean, the way we go to market, and this is -- I'll just do kind of a broad generic size because I mean it's different from customer to customer. But typically, we have 2.5- to 3.5-year EDA contracts. We basically chunk those -- let's say it's a 3-year contract. We basically chunk that up into yearly increments, and we provide a key for that year. And then if you continued using us, you get the next year of keys around your renewal date. So on an EDA side, customers are anywhere from the time last week that we got this note for the next 355 days, assuming this was written in stone, they'll go dark at some point during that period, right? It just depends on what their renewal dates were. We can no longer -- we're not maintaining or fixing bugs or doing updates in that, but they could continue to use the tool, which they'd have to kind of keep updated themselves. On IP, if that was something, we would -- if they've downloaded the IP from our databases before that date, they could use it, but there won't be new downloads until we create resolution here. And then with hardware, they could keep using that, but it would be the same thing. There wouldn't be updates or software bug fixes, those kinds of things.

Yes. I was just wondering like in conversations with the government, is the intent really like a restriction on the final owner or on the geography of sale, especially since like software design services can be moved around? Like how strict is the BIS like oversight in sales in the language sounds like sales direct to China outside of that military end user list where it's anywhere versus just Chinese subsidiaries that would open into Singapore, Middle East and moving around some of those designs. What is the intent behind that? And how do you see like reactions in talking to customers in China on how they are trying to like maintain that excess?

Yes. No, it's a great question. And it's -- these are some of the questions we're going back to commerce on to really get clarification on some of these things because it's a fairly generic and vague kind of -- and there you naturally will ask a lot of questions of would this apply to a U.S. multinational company with Chinese national employees? Is it an export if they look at it? These are the kinds of things that we would refine and we are refining as we go through it, but we need a little more time to kind of go through some of these things. But these are exactly the kind of questions that we want to get fine-grained answers on so we can basically follow the intent. Right now, as I said, we're -- we've stopped everything while we're working through this. But those are the kinds of questions we have to kind of get answers to so that we can make sure, hey, these are sales that we can make and these are allowable and supported. And these are things that we must keep off limits. So yes, it's a good question, but some of the things we're refining right now.

Any other questions? Is there a way, Trey, to ring-fence your product just by the kind of applications? Because the way I understand it, it's the same kind of basic tool, but it's the libraries, right, that go along with it that somewhat determine what range of applications it can get into.

And we've done this. I mean, so a good example is the gate-all-around, right? So we can sell Fusion Compiler, which is kind of a core digital implementation tool, place and route tool that we have. And in China, it just does not work on gate-all-around. So you can do some -- you can manage the -- what's okay by the tool when it calls back to us. And so we can do those kinds of things. But there wasn't -- at least in the initial note, there wasn't that kind of clarification.

Just a blanket. Okay. Understood. Let's get to kind of the core business. The AI exposure for the company, so the industry was involved, I think, right from the get-go, right? So the growth was above trend for a few years. Now it has kind of come back to the trend line. So where are you seeing the benefits of AI in your business? So first in terms of the features in the product itself? And then secondly, the ability to upsell versions of the product that have AI Copilot like enablement?

Yes. No, it's a great question. I mean there's a -- we kind of started -- it was really Sassine, who's our CEO, who started this journey when he ran our EDA group kind of 2018, that kind of time frame. We brought out the first optimization engine in 2020. Samsung -- this was around hot chips. Samsung read an article about it. And we started to see this was -- we added it to a variety of our tools to have traditional reinforcement learning optimization engines, hugely valuable to our customers. Also, then we -- this has kind of been a journey we've developed along with a lot of people in the industry, knowledge assistants or copilots that kind of get the first blushes of generative AI and get engineers into a prompting space. And we've seen huge benefit. I mean we've run kind of sample cohorts at some of our big customers. And the nice thing is they broke it up into the maturity level of the engineers. And it's not probably unexpected that really senior engineers don't get a ton of bang from a copilot because they've done it so much. But junior engineers, mid-level engineers get a ton of efficiency, like 30%, 40% because you instantly find things that may be in a 1,200-page manual, right? And so a huge learning curve benefit. Those things have been beneficial and are adding to the P&L, but they aren't that kind of inflection in terms of the P&L and the cash flow. Where we see that really coming is with Agentic. And so for -- as you kind of look through AI, a lot of these early stages are table stakes to really get to Agentic because as we get into actually having agents that we sell to customers that can actually do design engineering workloads, they're going to pull on those optimization engines. They're going to use the copilots, all those things. And we see a ton of interest from our customers. We're already doing some beta work inside. We want to get it out there as fast, but this is probably a few years away. The challenge in our space relative to maybe someone in a lower risk environment in a call center, things like that, the impacts of making an error are so huge, right? You -- if it's a $1 billion tape-out, that's not the place you want to make the mistake. And so we need to operate in that kind of 5 to 6.9 space. So where we'll bring this in is we'll start to do things that are more risk tolerant, things like checking test benches, checking the work of engineers. But this is going to progress. And I encourage you if you didn't get to go to our CR user group in March, Sassine spent a lot of time on this, and he had a pretty good parallel analogy of the autonomous driving kind of Level 1 through Level 5 and how that would look in Agentic. It starts with early kind of work in the background, but it will progress to where you have agents working with other agents, an orchestration level that's Agentic where -- and then it gets to kind of more autonomous mode. And so there's a huge amount of promise in this, and it comes into a market where we have a significant resource scarcity issue. And so the industry overall, our customers and ourselves, about 20% short the number of design engineers we need to go make all the products on the road map that people have for the next 5 years. This is one of those where you can say, hey, this is an excellent way to kind of be a scale multiplier in terms of how our engineers work. And we have a lot -- there's a lot of work to do technically. There's also a lot of work to do on business model because this is a pretty big change. We've been selling 3-year subscriptions. As you get into this, you certainly would have something more metered or consumption-driven, kind of usage driven. And so there's a lot of business model update. But at a high level, kind of cocktail napkin math, this hunts. I mean it's -- we -- our customers pay us about 15% of the R&D money for our IP and EDA. The other really goes to their headcount. And so it's one of those arguments where you can say, hey, would you pay 50% more on 15% to get a 30% reduction on 80%. Very ServiceNow-like argument. But a lot of work to go do to get there, but it's compelling.

A few quick things. So Ansys, from what has been said, I think let's set the BIS, right, implications aside, I think you guys have kept this June closing date. Are you getting enough kind of warm and fuzzy from the final jurisdiction that matters to maintain that level of confidence?

Yes. I mean I started when I was talking to everybody today, I was like I'm not sure I could construct an argument with all of the things that are -- that you see in Bloomberg and the New York Times and everything to get everybody here to give me a high five on the conviction level. But what I can tell you is we're having really good dialogues. And so...

Is there any sticky point still, you think?

They're the same kind of good dialogues that happened. And we've gotten through the -- last week, we got the final consent decree from the FTC. So we've cleared all the other geographies regulatory. The discussions have been really good. There is significant interest in keeping Synopsys and Ansys technologies in country and having those capabilities. Ansys only has 5% of their revenue there, but it's important revenue that touches a lot of important infrastructure and other capabilities. We're used prolifically in the country. And we're -- the one thing I would say, too, as we had in other geographies, we're getting a ton of customer support. These are Chinese customers who want to see the combination and be able to use simulation and EDA together. And yes, there's tons of noise and issues at a top level working through trade issues. But there's also a realization that there's a long term, right? These are capabilities that the country and companies want. So good dialogues. We still are very optimistic, and we want to get this through the typical approval.

Got it. Next important topic is your largest customer, right? And they have a new CEO, right, who comes from Cadence. So talk to us about what that means from a competitive landscape, your market share? Is there an opportunity that maybe just the spend there expands and we are just worried about a fixed pie, whereas maybe the pie there can expand, right? How do you think about market share and their spending levels?

Yes. We've -- Intel has been a great customer. We exist there with others. And certainly, Lip-Bu has ties -- deep ties to Cadence from his past. But we've worked with Lip-Bu also, and Sassine has a great relationship with him. In fact, there are a number of companies in the Walden portfolio that just use Synopsys, right? So before you even got to Intel, I mean, there's a lot of companies that were exposed through his investment portfolios. Our goal is to make them as successful as possible. I mean it's a critical company for the world, and we want to do everything we can to make them successful. I would say EDA is pretty sticky in terms of -- you have users that have used it for a long time. Certainly, people move back and forth, and they may use both tools. But that's not one of our major concerns. We have a lot of other things that we're working on. But the main one is just to make them super successful.

Got it. Do you think there is a possibility that maybe just their spending because they have -- I mean, you have direct experience there that their spending has not been at par with the percentage of R&D, as you mentioned, that's the industry norm. Do you think there is a chance that maybe there is a larger addressable market?

Well, I mean, it -- a part of this gets back to -- and I know I was at Intel for 22, 23 years. And the most important thing for any company is leadership products, right? I mean -- and EDA is a piece of that, but all the other aspects that come into that. I mean -- and I think Lip-Bu was here, he'd probably say, "Hey, we're focused on building leadership products however you get that done." So yes, I hope that is the opportunity that they operate in a great way, and it creates more TAM for everybody.

But you haven't seen any slippage or share or anything along?

No. I mean we -- so we have long-term EDA contracts with them. We have another contract where recently, over the last couple of years, they used us. They had done a lot of internal IP and chose us as we kind of get into 18A and some of the more advanced nodes. And then they've been a great hardware customer, too. So yes, all 3 of those are separate. And so no, we've continued to see good support.

Got it. And finally, just from your experience now at all these organizations, one question that always comes up, people will rightly or wrongly contrast Synopsys, right, and trading multiples with like cadence and trading multiples and say, well, how come there is such a wide gap, right, in the multiples? Historically, I think that existed because of the difference in profit margins, but I think Synopsys has done a good job, right, since Shelagh has come, right, to kind of help close some of that gap. Why do you think there is still this really large delta?

Yes. I mean I'm sitting with a group of experienced investment professionals, and I'm going to tell them why I think the stock is trading. So take this with a grain of salt. So yes, I used to.

Rumored by greatness I greatly exaggerated.

Well, it's a quick aside, but when I was at Intel and IR and Stacy Smith was the CFO. And every IR professional has this experience where it's like what happened to the stock today, right? And I'm sure you all get this. And after talking to multiple trading desks and hearing 7 different stories of why the stock moved, I didn't -- Stacy and I didn't -- it looks like more sellers than buyers. Yes, that's what I'm discerning. But Yes. No, I think it's a couple of things. One, I think, was what you hit. When I came in about 2 years ago, Shelagh's had a tremendous focus on scaling the company at an improved operating margin. We've been doing about 150, 200 basis points a year of operating margin accretion, which is fantastic. It's not over. We're going to keep doing that. But we're still at a delta relative to cadence. But as we did and we gave pro forma when we did the Ansys acquisition, we expect the combined company to be mid-40s, which would be market-leading kind of operating margin. We need to continue to work on that front. I do think -- and we can all compare notes at the watering hole afterward. But I mean, certainly, the deal overhang has meant that when I came in, I looked at the stock, a lot of investors own us and Cadence, right? So I mean, you're kind of playing the entire market for this sector, which is very rich, a lot of high growth. But Cadence is a pure play that isn't wrapped up in a $35 billion acquisition, where a lot of ours is how is the deal closing, how are these things. So I think that's created a little bit of a disconnect. But the margin one is real, and we intend to close it.

Terrific. Thank you so much, Trey. Really appreciate it.

Yes. Thanks for joining.

All the best.