Ladies and gentlemen, thank you for standing by, and welcome to the Synopsys Earnings Conference Call for the Second Quarter of Fiscal Year 2021. [Operator Instructions] Today's call will last 1 hour. 5 minutes prior to the end of the call, we will announce the amount of time remaining in the conference. As a reminder, today's call is being recorded. At this time, I'd like to turn the conference over to Lisa Ewbank, Vice President, Investor Relations. Please go ahead.

Thank you, Sean. Good afternoon, everyone. With us today are Aart de Geus, Chairman and Co-CEO of Synopsys; and Trac Pham, Chief Financial Officer. Before we begin, I'd like to remind everyone that during the course of this conference call, Synopsys will discuss forecasts, targets and other forward-looking statements regarding the company and its financial results. While these statements represent our best current judgment about future results and performance as of today, our actual results are subject to many risks and uncertainties that could cause actual results to differ materially from what we expect. In addition to any risks that we highlight during the call, important factors that may affect our future results are described in our most recent SEC reports and today's earnings press release. In addition, we will refer to non-GAAP financial measures during the discussion. Reconciliations to their most directly comparable GAAP financial measures and supplemental financial information can be found in the earnings press release, financial supplement and 8-K that we released earlier today. All of these items plus the most recent investor presentation are available on our website at synopsys.com. In addition, the prepared remarks will be posted on the site at the conclusion of the call. With that, I'll turn it over to Aart de Geus.

Good afternoon. I'm happy to report outstanding second quarter results, exceeding all of our key guidance metrics. We delivered revenue of $1.024 billion with GAAP earnings per share of $1.24 and non-GAAP earnings of $1.70. Business was strong across all product groups and geographies. We continue to make good progress on our margin expansion goal and generated record operating cash flow of $526 million. As a result of our first half strength and growing confidence in our year, we are raising guidance for revenue, non-GAAP ops margin, earnings and cash flow. Trac will discuss the financials in more detail. Before commenting on highlights, let me say a few words about the dire situation in South Asia. While parts of the world are progressing well with vaccination, we are seeing an enormous challenge for the people of South Asia. Our top priority is the well-being of our employees, and we have taken many steps to support them and their families. Ranging from orchestrating oxygen concentrators to teaming up with vaccination clinics to ambulance services, food delivery and family health, our objective is to maximally mitigate the impact of COVID and make sure that every employee can call on Synopsys as a beacon of care and solidarity. Despite the pandemic challenges, we are thankful that from a business perspective, we continue to ship our products and support our customers with no material disruptions, and our business is doing well. Looking at the overall market, demand for semiconductors is very strong. While some of the near-term demand can be attributed to segments such as automotive catching up after a year of COVID slowing, there is an undeniable new wave of growth on the horizon as every vertical market demands machine learning chips to harvest their big data for their specific needs. In other…

Thanks, Aart. Good afternoon, everyone. As we report another outstanding quarter, let me echo Aart's thanks to our team not only for their dedication but also for their unwavering focus on innovation to fuel the exciting opportunities we have ahead. We are in a great position as we set our sights on the next level financial ambitions. On top of a solid foundation of nearly 90% recurring revenue, a diverse and growing customer base and market and technology leadership, our track record of excellent execution continued in Q2. We are increasingly confident in our outlook and are raising our revenue, non-GAAP earnings, non-GAAP operating margin and cash flow guidance for the year. Now to our second quarter results. All comparisons are year-over-year unless otherwise stated. We generated total revenue of $1.024 billion, up 19% and above our target range, driven by broad-based strength across product groups and geographies. Semiconductor & System Design segment revenue was $930 million, with strong growth in both EDA, software and hardware and IP. Software Integrity segment revenue was $94 million. The positive orders momentum we saw in the quarter shows that the adjustments we've made in the business are taking hold. We are on track to meet our 2021 expectations of 15% to 20% orders growth and to exit the year with double-digit revenue growth in the fourth quarter. We're on a good path to accelerate revenue growth back to the 15% to 20% range long term. Moving on to expenses. Total GAAP costs and expenses were $830 million. Total non-GAAP costs and expenses were $707 million, resulting in a non-GAAP operating margin of 31%. We are on track to again deliver operating margin expansion for the year and are raising the bottom end of our guidance range. Adjusted operating margin for the Semiconductor &…

[Operator Instructions] Our first question is going to come from the line of Jackson Ader from JPMorgan.

Great. Aart, first one is for you. You talked about the chip differentiation by vertical as a nice tailwind for the company. I'm just curious, how do we scale the benefits of those -- I'm sorry, not scale -- square the benefits of those differentiations with the fact that there's -- there tends to be more increased IP usage from some of these newer entrants? And so I'm just curious, with differentiation seeming to lead to more custom design, what does that mean for those IP blocks that get designed once and used by many?

Okay. Well, there's 2 ways to look at this, from the perspective of the vertical or from the perspective of the pure semiconductor companies. From the vertical, the first thing that people need to choose is, will they design their own chips, yes or no. And if you take the example of the hyperscalers, some, not all, but some of the automotive companies, the hyperscalers clearly are doing more and more of their own chip design. And automotive are either sort of dabbling in it or looking at some of their suppliers, the Tier 1s. In all cases, they are doing more chip design. And so that is good news for us. And absolutely, you're right that a lot of that design is done by substantial IP reuse, and you saw that our IP business is strong. Now if you sit on the other side of the fence as a semiconductor provider, you look at these customers are -- each of them as opportunities to take sort of a core architecture and then say, "Well, how do I take my architecture and do different derivatives that are particularly good for different submarkets and essentially reuse some IP or add some that are just -- that is just necessary for that vertical?" And so I think there's no doubt that, therefore, we will see more chips of different type and more design. In all cases, consumption of IP will continue to grow.

Okay. All right. Great. That's helpful. And then, Trac, quick follow-up. Any -- I think the cash flow performance is great to see, but it was so much bigger than I think we expected. Was there anything that was pulled forward, either -- any deals pulled forward or collections pulled forward?

No, not at all, actually. It's -- the profile is good because the business is pretty healthy, and we're generating obviously very strong operating margins. The other part to keep in mind is Q2 over the last couple of years will be normally our biggest collections quarter given the profile of renewals and where we end up invoicing at the end of Q1. But it was a combination of both. It's definitely a very healthy business right now.

Next then, we're going to go to the line of Tom Diffely from D.A. Davidson.

Trac, first one is for you. When you look at your increased guidance, which was a pretty substantial increase, is this more a factor of passage of time with comfort in your backlog? Or did you actually see an acceleration of business trends during the quarter?

Tom, as to your question, there's a number of things going on. First part is that we're -- we've got half of the year behind us, right? And the visibility that we have after the business that we booked certainly improves the outlook. As Aart described, the overall markets are healthy, and we're executing well against that. So there's a number of different factors. The backlog -- I would translate the increase in backlog to whether the business is healthy or not. The backlog increased this quarter as a function of the renewals that we had planned. But overall, the business that we did book saw very good run rate growth. And the growth in that business gave us really strong outlook for the year and confidence in the year.

Okay. Great. And then, Aart, just a broad question for you. When you enter into a market like we have today where there's chip shortages, what are the impact of those shortages on design activity, either positive or negative?

Actually, very little impact. There are some people that do take existing chips and decide that they're going to do modification so that they can get more capacity with another vendor. People really hate to do that because it's a lot of work and no direct benefit, except that if you can ship to a customer, that's great. But it takes some time. And so I don't think that, that is going to be a particularly strong driver. But the shortages should not be just interpreted, I think, as a reaction to a market that has been partially asleep during the COVID time and now is obviously catching up. And automotive is the best example because the mistake that was made there is they stopped ordering. And in the past, they were very powerful, and everybody jumped when they needed something. Now there was just no capacity left. And the reason there's no capacity left is because all the capacity is used by huge demand period. And that's why I'm trying to differentiate a little bit with the temporary demand that comes out of sort of just this historical 1.5-year wave versus, I think, something that is much more profound, which is a whole new era of semiconductors impacting verticals. And I expect that to continue. And by the way, you can see how many, many places, including countries have decided to substantially increase the capacity. Those increases will take a couple of years to actually have impact. But they do illustrate the direction that our field is taking.

Great. And just a quick clarification. Did you say you had virtually no impact in India with your...

Well, we're dealing, like everybody else, with a humanitarian situation that is very demanding. We've been able already for a while to rebalance the activities of our employees in such a fashion that to date, we have 0 material impact or delays in shipping anything or supporting anything. But I expect that for a number of months, India will still be, from a humanitarian point of view, a point of focus where we'll give a lot of support to our team.

Next, we're going to go to the line of Joe Vruwink from Baird.

Great. I maybe wanted to start. New technologies like DSO.ai, SLM, these have been getting called out more regularly over recent quarters. Is there a way to characterize or maybe compare to products in your past and the consequence of these new technologies? Is this just the natural evolution of Synopsys? Or is there something different? And perhaps it's specifically about AI adoption in the industry, but is there something different about these technologies where the ramification later on could be more consequential?

That's a really, really interesting question actually. Let me start with SLM because what's interesting about SLM is this word life cycle. Because if you just take a chip and you say, put it in a phone or put it in a car, we all know which one has the longer life cycle. And so -- and in the case of the car, you have, of course, safety that is part of it. And so suddenly, the ability to put inside of the chip sensors and the diagnostic system that, by the way, gets trained by AI so that a chip can self-diagnose as, well, I'm not feeling so good. You better start stopping the car, so to speak, is going to be a very high value. Maybe even more practical immediately is in cloud centers, where people run compute at the max speed and they want to know is a certain set of processors going to go down so that I replace them before it happens. So in essence, preventive maintenance. And so in that sense, SLM is interesting because we touch these chips literally at the early days of even what are the type of transistors. So very minute physics, but now we also have very meaningful interaction with very large companies that are exactly in those verticals I described. Now DSO.ai, I think, is breakthrough technology. And it's always difficult to compare something that we did over 30 years ago. But the early days of Synthesis had something similar, which is it took a set of human tasks where complexity just was outrunning the human and automated it. And overnight, we could do circuits that were better than what a human could do in a fraction of the time, and were faster and smaller. Now we're talking of entire chip pieces, very large designs with many, many different constraints. And the fact that we can take tasks that take people multiple months and bring them down literally to a few weeks with fewer people and, in the last few quarters, even better results certainly sounds very similar. But to me, it's sort of essentially 30 years later, many orders of magnitude more complexity. And I think it fits well the very moment where the semiconductor industry will want to do many more chips for all these verticals. And so we use sometimes the front line of using AI to design AI chips, but that is exactly what this is. And it's exciting. We're just at the beginning of that, but the impact is already economically felt by the users.

Okay. That's really interesting color. Second question, is there anything about the sequencing by quarter of this particular fiscal year that maybe is a bit different than you originally expected? I'm thinking about things like it was another very strong quarter in China. You occasionally hear about maybe pulling forward some future business. And then the way the margin guidance appears to sequence this year, it looks like perhaps 4Q has a bit more incremental cost. Maybe that's just hiring related. So I suppose is there anything that is maybe different timing-wise or just the sequencing of your quarters?

Joe, this is Trac. Overall, the profiling of the quarters is very much close to what we had planned around. I'm actually really happy with the profile this year given how back-end loaded last year was. Most of the things you described, the revenue, pretty linear this year. As far as the margin profile, that's just a function of hiring -- largely a function of hiring throughout the year. But we're pretty pleased with how it's shaping up relative to the plan that we had at the beginning of the year.

[Operator Instructions] Next, we're going to go to the line of Gary Mobley from Wells Fargo Securities.

Congrats on a strong first half of the fiscal year. I see that China was up strong again, and I wanted to ask about -- I wanted to ask your personal view, Aart, and perhaps to get some color from you on the changing geopolitical dynamic between the U.S. and China seemingly now more so influenced by the broad supply shortage we have in the semiconductor side and this renewed focus on onshoring chip production to the U.S. And I realize that you're not so much aligned to expansion in chip production. But I think a lot of these measures that are proposed, these bills and whatnot are focused on growing R&D investments as well. So I'm wondering perhaps if you're starting to see any sort of influence from that in your licensing or what may be in order for you guys looking down the road.

Well, the first comment is when a market is strong, it typically tends to be strong for everybody. And the very fact that even politicians know -- not only know what a chip is but may have seen one is certainly encouraging. And the fact that nations want to invest more because they think it's strategically important to be close to this whole next age of Smart Everything and AI, I think this is all very encouraging for us. And so while there may be tension between different countries of who does what, the race is on, and the fact that there's shortage is just accentuated by some of those tensions. But I think the most interesting part of all of this is that there is gradually now a broader understanding that the whole next wave of human products impact and so on is very, very much linked to the notion of big data intersecting with AI, i.e., smart results. And it's interesting that now even electronics are referred to as infrastructure in a country. Well, all of these words are encouraging because I certainly believe that while it's not a panacea to all human problems by any means, it has enormous power to evolve all vertical fields. And so I'm not surprised really that the degree of attention has gone up. But right now, the fact that people want to spend more on R&D or manufacturing capacity, it's all good news for semiconductors.

Appreciate that, Aart. Trac, you mentioned that backlog was up sequentially. I know you haven't filed your Q yet, but specifically, what was -- what were the remaining performance obligations for the end of the quarter? And related to that, would you expect revenue growth and backlog to trend sort of in line with each other? Or would you expect over time to generate a larger percentage of revenue from turns business like Emulation or whatnot?

We've made a change -- or you saw a change in the turns mix largely at the beginning of FY '19 as a result of the 606 transition. And I would say for the most part, the business has been relatively stable, and it'll be -- that percentage might move from quarter-to-quarter depending on their hardware IP deliveries. But I think we're in a pretty good stable level right now. The backlog is up, and we should be filing our Q next week. So you should be able to see the actual amount. But I think we're going to disclose that it's about $4.8 billion or north of $4.8 billion.

Next, we're going to go to the line of Gal Munda from Berenberg.

The first one is just when I look at your performance in H1 actually in total, like you said, it's really the opposite of what we had last year when it was very, very back-end loaded. So nice recovery there. But then also looking at the Q3 guide, which is very solid, I'm thinking when you look at for the rest of the year and you have a pretty good visibility now, especially into Q3, is it fair to say that if -- especially some of the hardware orders, some other stuff comes in, you could consider the guidance as fairly conservative still at this stage? Or do you think that's kind of a fair representation of what you're seeing?

Yes. It's a good question. The profile we laid out, the quarterly profile we laid out for Q3 and Q4 is largely a reflection of the revenue recognition profile of IP and hardware. We actually had very good visibility in the second half, and that's why we raised the guidance for the full year. At this point, keep in mind, given the new revenue rules, you're going to see some variability from quarter-to-quarter depending on when hardware IP is delivered. But there's nothing unusual in the profile in the second half other than that. And actually, the business is really doing really well.

Right. Yes, that makes sense. And yes, that's what I was thinking. And then the second one, you talked again about double-digit growth for this year now implied in the new guidance. Last year, we were virtually at that level as well. How do we think about Synopsys as an effectively sustainable double-digit growth company now, especially if I'm thinking potentially Software Integrity ramping up growth a little bit and starts contributing incremental -- maybe a few bps of growth? Like is that a profile that you're happy with when you're kind of thinking about the midterm planning?

I want to be cautious about getting too specific about the numbers looking forward. But overall, what you're seeing in the results for this year and last year really is why we feel really optimistic about the future and why we have communicated our confidence in being able to drive the business towards a Rule of 45. And that's really going to come through a combination of really strong revenue growth and margin expansion. And so you've touched on it a little bit. You're seeing some really good acceleration in the business. And as Software Integrity ramps up and gets back to where we believe it's capable of operating at, that should help with the overall mix. But keep in mind, it's only 10% of the business. The overall growth rate is a reflection of what we're -- how well we're doing in EDA and IP as well.

Right. No, that's why I was thinking, right? It's 10%. So it's growing 5 to 10 points faster. Obviously, that's a nice contribution. Congrats again.

Next, we're going to go to the line of Jay Vleeschhouwer from Griffin Securities.

Aart, let me ask you 2 related questions regarding the evolution of EDA and your markets. Tonight, you referred to a new era of EDA. We've heard similar remarks from you and others in the industry for some time now or silicon renaissance and so forth. The question is how that affects your business profile specifically with respect to services. That is to say as you move into this new era, does this tend to increase the kind of services and AE support that you necessarily have to provide to encompass or support this new era? And if so, what could be the margin implications of having to provide that incrementally higher degree of support? Relatedly with respect to the next generation of chips, the domain-specific chips that you talked about now for actually 2 or 3 years, what does that mean during the design process in terms of license consumption? If we think about your model as now prospectively a kind of consumption model, as is often the case in simulation, do you think that the consumption or utilization intensity per design, per run, however you want to think about it, necessarily goes up?

Well, I think your question brings sort of together everything that we would refer to as systemic complexity, meaning that, of course, chips continue to be more complex and bigger and more transistors. But been there, done that. And by the way, that will continue for a while. What makes systemic complexity interesting is that you get multiple players intersecting. Suddenly you have companies that are sitting on these verticals that are highly interested in knowing how the chips will actually work because they write their own software, for example. And vice versa, the people providing chips are really interested. So what software does this car manufacturer really want to run on it that has to be really fast so that I can change the architecture of my chip to accommodate that? And our role is interestingly broadened because we're sitting at the intersection of all of that. And you may recall that a number of years ago, under our logo, we literally put silicon to software. Well, that was essentially the summary at that time of this vision that there's a continuum that ultimately brings the power of chips to end users that have -- that are not really interested in chips, except that they need them to get smart outcomes. And so when you move into that space, it's not so much that support is increasing but that there are new service opportunities as a number of new players initially may not know so much about these domains but they know a lot about their opportunity space and helping them connect is an opportunity that will continue to grow for us. And we will certainly be able to manage it so that, including with the tools, the profitability will be very reasonable or even good. The second part…

Next, we'll go to the line of John Pitzer from Credit Suisse.

Congratulations on the solid results. Trac, this question was asked a little bit earlier, but maybe I can ask it in a less politically correct way. I mean despite the beat and the raise, this is the quarter where the full year guide gives us some insight into the last quarter as well. And relative to that full year guide, you are kind of embedding a pretty -- a meaningful deceleration, if not in top line, in EPS. Is this nothing more than normal conservatism? Or are you really trying to signal that there's something bottoms-up that just makes Q4 a little bit softer this year than it might have been in other years?

No, not all. We're -- we don't see any deceleration in the business at all. In fact, it's quite the reverse. We feel really good about the momentum we have in the business. And that's reflected in the full year guide. We've always cautioned that the quarter-on-quarter profile is going to move depending on what the profile of revenue is. And from a revenue perspective, I'm actually really, really pleased with how linear it is this year and the fact that we were able to get half -- almost half the business booked in the first half of the year. So it's a great profile with really good visibility. We continue to ramp up hiring in the second half. And so you're going to see the expense profile go up. But as we're ramping up hiring in the second half, we're mindful of the trajectory and what it implies for our ability to continue to drive margin expansion over time. So we're very cognizant of that. But no, there's nothing unusual to the profile. And I wouldn't characterize it as deceleration at all. I think right now, when you look at the year-over-year comparison, just keep in mind that last year was unusually back-end loaded. So it's going to skew the comparisons, particularly in the second half.

That's helpful. And then, Aart, I want to go back to an earlier question about sort of the regionalization or domestication of semiconductor production. It's a clear benefit to the equipment ecosystem. But I'm trying to get a better understanding of what it means to the EDA ecosystem. As existing foundries kind of move from region to region, is there sort of redundant or duplicative spend on the EDA, kind of question number one? And question number two, there's one big guy out there that's trying to reemerge as a foundry business from just an IDM. How does the EDA spend conceptually trend as they try to do that?

Good question. So we touched manufacturing a little bit because we have a number of tools that are designed specifically under the topic of silicon engineering to help optimize circuitry and chips for manufacturing efficiency and yield. But you're absolutely right that most of the investments on sheer capacity don't touch us so much, except if people want to enter the business such as be in a foundry or go after specific segments of the market that they didn't go before. There, certainly our tools matter a great deal to help them get there. In general though, when volume increases, with it, also the number of designs increases. And so the advances in sheer silicon technology and the number of designs are actually very positive at this point in time. And the fact that more people want to be in the manufacturing means that more people are also into the investment of R&D around the field. So at this point in time, it's all positive.

And then, Aart, if I could just take one more in. One thing that's kind of unique about this current semiconductor cycle is how tight trailing capacity is right now. And I'm just kind of curious, are you seeing any evidence that perhaps design activity on the trailing edge is picking up as customers are using some of this tightness in the near term to try to rethink about moving down node widths -- new line widths at maybe a faster rate than they've historically seen? Or how do you see kind of the TAM for that trailing edge market over the next several years?

Well, it's an excellent question because some of the trailing edge manufacturing equipment has already gone up in terms of pricing as people try to get capacity wherever they can. But the other comment would be different foundries are sort of also focusing on different type nodes. Some focus mostly on the most advanced nodes. Other would be sort of in the middle field. And then the truly older trailing edge nodes, they're typically limited by the amount of capacity at 200-millimeter wafers. What is interesting in the nodes that are maybe not the leading, leading edge but, let's say, 3, 4 years behind that, there, the application of the newer tools that we have actually has a lot of impact on those, too, because newer tools for older nodes still means a lot better design out of these older nodes. And so it actually gives them a bit of a second life from an efficiency point of view. And as you said, if people can do really well with nodes that have been well honed where the yield is high, the cost equation is very attractive. And we see, for example, one of our most advanced tools, Fusion Compiler, going back to older nodes with some of our customers with great delight.

Next, we'll go to the line of Jason Celino from KeyBanc Capital Markets.

Great. Aart, Trac, it seems like a pretty good demand environment for Emulation and Prototyping just across the board. But in the past, we've seen some customers gravitate towards the latest and greatest products here. One, is that still the case? And then two, if that is still the case, how does the new improvements to the ZeBu EP1 compare to some of the other announcements in the market?

You're welcome. So clearly, Emulation and Prototyping is increasing in value and importance in some parts of the tasks and especially for us in the task that has to do at this intersection of hardware and software. And there's no question whatsoever that, that is an area that will continue to grow. And we had mentioned in earlier conversation specialties for certain verticals. It's interesting that these are capabilities, Prototyping, that are of high interest in the automotive space, for example, because they start working on the software many, many years before a car is even fully conceived. And so being able to accelerate all of that is of high importance. Now underneath that, there is invariably always the same demand, which is make it faster, make it faster. And that is what Emulation and Prototyping is all about. But it's also give it a larger capacity. And it's also are there certain tasks that you would like to accelerate such as the question of, "Well, if I write my software this way, how much power is it going to consume versus if I write it differently, will I be able to do better?" Well, those are specialty questions that we now are increasingly answering using Emulation and Prototyping. On top of that, we're on the most advanced chips inside of our machines that are available. And so this ability to specialize and optimize for specific applications turns out to be extremely valuable, and we're doing very, very well with that.

Next, we'll go to the line of Vivek Arya from Bank of America Securities.

I had 2 as well. The first one also about the growth in the business. So Aart, you mentioned a fair bit of improving engagement verticals such as autos coming back up. But when I look at your full year growth outlook of about 10%, it's about the same growth that you did in the last fiscal year. So my question is a more conceptual one. Why aren't your sales accelerating when semiconductor designs are getting more complex? And when I look at the implied Q4 sales, they are basically flat year-on-year. So even if I ignore the quarter-to-quarter visibility -- or volatility, the full year sales are about the same level of growth as last year. Why aren't you seeing an acceleration in your sales growth?

Well, the acceleration tends to come after the orders play themselves out through a multiyear ratable revenue recognition, for one. The other thing is actually, right now, I think we just increased our projection for you for the -- or guidance, I should say, towards the end of the year with a high degree of confidence that it is well on track. So we'll have to see where we ultimately land. But there's no doubt that our objective is to move beyond the single digits to higher. We have just not changed our guidance at this point in time, and we will do that as we enter next year. But the message you should take away from everything we're saying is that right now, we feel that we have great opportunity to grow very well.

Vivek, I also -- this is Trac. I really want to caution you about looking at Q4, okay? Because remember, last year, Q4 was extraordinarily high because of the schedule of hardware and IP in Q4. And so that comparison is not going to be -- that comparison in Q4 is not a good indicator of the momentum of the business.

Got it. And the next one is I'm curious, what's your level of exposure to all the AI start-ups? And do you see any kind of rationalization or shakeout in the number of start-ups? Because, right, in previous calls, Aart, you have mentioned the systems companies doing a fair bit of their own AI work, especially the hyperscalers. And then we have the large incumbents such as NVIDIA and others. Do you think that the industry can really afford to have so many smaller players going after this market when the cost of staying in semis is so high? So the basic question is that -- what is sort of your exposure to all the work that's being done by these AI start-ups? And do you see any kind of rationalization in that?

Well, we're exposed extremely broadly to many, many, many AI companies. And yes, it is true that maybe as a little bit of a caricature, every single one of them is designing the best-ever AI chip. And on one hand, you can say, well, you know that will consolidate over time. On the other hand, you can say, no, this is exactly the behavior that you see in a very high promise, early phase of a product or technology development. And there's no doubt that a number of the more successful AI companies have already been acquired by larger companies. And lo and behold, 2 years later, you have someone -- the same people doing the next one. And so I think we're in an extremely active phase of invention, development, finding end markets. And then over time, we'll also see that the AI itself is going to become more and more specialized to the verticals. So yes, there will be consolidation at some point in time. But the number of designers remain -- is certainly not declining when that happens. And we have actually done very well through exactly these type of phases where there's a lot of activity.

And then our final question is going to come from the line of Pradeep Ramani from UBS.

I had a couple. First, on China. Is the right way sort of to think about China as being -- contributing to roughly 12% to 13% of your revenues in the back half? Or do you see any sort of deceleration in the back half? I mean the comps are probably getting harder, but I just want a clarification around how investors can think about China. And I have a follow-up after that.

Pradeep, this is Trac. China continues to do very well for us. And we -- while we disclose China as a separate country, we don't guide to that or comment on the outlook by country.

Okay. And for my follow-up, I guess, with respect to the Rule of 45, I mean, this year, you're going to be very close to 40. And your sort of semis margins are sort of already -- I mean this quarter were 33%. How should we think about sort of just the sustainability of margins in semis not just for in the back half but longer term within the 45% -- or the Rule of 45 kind of framework?

Okay. Let me try to explain it this way. The -- overall, we do see an opportunity to improve margins across the entire business. Really for us to drive to Rule of 45, it really is going to require everyone to contribute. And keep in mind that the semi business does represent 90% of the overall mix. And so that's going to contribute. But I would circle back to the fact that it starts with growth. The reason why we do feel good about our ability to drive margins up in all areas of the business is that we're seeing really strong growth in the business. And that's going to help us effectively get more operating leverage and therefore drive margins up.

If I can add to that. Fundamentally, we set our objective on the Rule of 40. We're obviously in striking distance to meet that. We even whispered to you Rule of 45, and that's obviously because that's what's coming next. We're -- I think we are reasonably well disciplined to make sure that, as Trac said, we focus on growth while increasing ops margin. And the 2 support each other. So if nothing else out of this earnings release, you should take away that I think we're well on track with our own plans and that everything we've communicated to you for the last few years, we're executing on. I assume that this means that the meeting is over. In any case, thank you so much for participating today. And we hope that you and your family stay safe as hopefully the world is moving to rapid vaccination. Be well.