Okay. Good afternoon, everybody. Welcome to the Goldman Sachs Communication and Technology Conference. My name is Jim Schneider. I'm the semiconductor analyst here at Goldman Sachs. It's my pleasure to welcome Synopsys and CEO Sassine Ghazi with us today. Thank you very much, Sassine, for being here straight from your earnings report last night. Thank you. Before we get started, I need to read a disclosure. Today's discussion may contain forward-looking statements related to our current outlook, expectations, and beliefs, which are subject to certain risks and uncertainties that could cause actual results to differ. Please refer to Synopsys' most recent SEC filings for discussion of risk factors that may materially affect these statements. With that, let's get started. Sassine, I'd like to start with the earnings report last night. Very topical, significant stock reaction on the back of it. I think the quarter itself and certainly your guidance was below our expectations, mainly driven by weakness in your IP business, as you explained. Understandably, I've gotten many investor questions about the reasons for that. I want you to unpack those different reasons in detail. Before we get started with that, maybe just start by level-setting people about how these factors are likely to impact your IP revenue outlook for 2026. Yes. Thank you for the opportunity. I wish it was a different morning I'm sitting here in front of you. It's a great opportunity, actually, to explain our perspective and excitement about the business that we have. Our Q3 was a very busy quarter for us. We reported our Q2 number on May 28th. On May 29th, we got a BIS restriction that hindered our ability to sell into China. That lasted for about six weeks. We closed a transformative acquisition within the quarter, and there was a very short stub period that we needed to account for it inside the quarter, roughly about two weeks. Part of these massive disruptions, the BIS disruption, as well as the very exciting transformative deal that we got clearance inside the quarter to close, what we communicated yesterday was the following. We have three parts of our business. We have the design EDA, design automation. We have IP. Right now, the new business from Ansys, the simulation and analysis. In the IP business in particular, we saw two major headwinds. The first one was a disruption out of our control. We could not see it. We could not forecast for it. That's the China impact. When the BIS restriction got lifted, it did not mean that the customers in China quickly got back into normal. There's a lot of uncertainty in China from semiconductor companies or system companies designing or needing IP to design chips that changed the rhythm of selling. That had a meaningful impact in the quarter. The other factor was a large foundry customer of ours. We made an investment on our IP portfolio to support their roadmap and where they're going. We had assumptions in our forecast that there will be customers both internally to their own product and as a foundry customer for that IP investment that we made. That did not materialize. Based on these factors, we decided yesterday to guide with de-risking these two factors. What we communicated is that FY2026, we will be taking these factors into account. Therefore, for IP in particular, we will have a muted growth looking at FY2026, where the rest of the businesses are still operating and delivering a very healthy nice growth in line with expectations. That's the difference I want us to emphasize on, where the rest of the business remains very strong and in line with expectations, while the IP, we de-risk these two factors not only for Q4, but as we look into FY2026. Great. I want to unpack these different factors. Before we do, when you talk about de-risk, what does that mean in terms of the level of the conservatism you've applied going forward? On the first one related to China, we anticipate the China environment to remain tough. What does that mean? A company like Synopsys faces many restrictions to sell into China. The China market has great opportunities, but many of these opportunities we cannot sell to. For example, there are technology restrictions. If you are a hyperscaler in China or a semiconductor company trying to design on-gate all around or deliver on a high-performance chip, you cannot do it in China. That has a negative impact on Synopsys because we cannot sell our portfolio to that customer. The customer uncertainty, they're looking for, they're uncertain, our customers in China. If I start a design and a chip investment, can I finish it? There might be another disruption from the U.S. Bureau of Industry and Security or whomever for their inability to continue that investment. These are having a negative impact on our growth in China. Actually, China for Synopsys was growing in the range of 20% to 25% CAGR for about a four-year period. Last year, it slowed down to corporate average. This year is going to be significant reductions from where we ended up last year. That's what I mean by de-risking that factor. The same thing with our foundry customer. We're assuming there will be no customers for the IP that we built into 2026. Those are the two de-risk that we're assuming in our forecast. Understand. Relative to China, do you think that is primarily a timing thing, or do you believe your customers in China are seeking alternative solutions, competitive solutions, or internal domestic solutions? Today, the domestic solution in China for EDA and IP is not mature. Our position in China compared to our competitors is significantly larger in terms of overall revenue and% of revenue in China. The distribution of our business in China is heavily weighted on IP. We feel it more than our competitors, given our IP strength and our IP portfolio. When customers decide to build less or invest less on semiconductor development for high-performance compute, we feel it more given the profile that we have in IP. Yeah, that's tied to a number of design starts and tape-outs also in China because it's more tied to IP pools. Yes, but it depends on markets. For example, if you're an automotive company in China, we're selling our IP just fine. If you're doing anything related to edge AI or data center, the lack of access to, say, technology like gate all around or how much performance or power per millimeter square you can design a chip, the customer is deciding, is it worth it for me to make that investment and get a chip that's not competitive or buy the chip from another supplier? That has an impact on our selling rhythm. Given they're not making the investment in IP or EDA, it's not going to yield to a competitive product for our customer. Relative to the foundry side of things, that's a large customer for you in general. I think many people in this room understand the challenges that customer is going through with respect to that business. I guess people would like to know, first of all, why did you forecast it to begin with in fiscal 2025? Should we just assume that it doesn't recur? It really doesn't kind of come back at any point in the future? Yes. Actually, the question that's coming up in the last instance, our earnings, is why did you even invest to build your IP there? I want to say, even though we have the choice to decide where to invest, when you are the leader in IP, there's an expectation from our customers and our customer's customer that we're the leader in that ecosystem. We need to invest based on various markets. For example, in our IP portfolio, we have a leadership position for our customers that they're planning to do automotive, consumer, mobile, edge AI, data center for multiple foundries because customers are looking for optionality and optimality across the portfolio. When we made the decision to invest with that foundry partner, there was an expectation that there will be usage of the IP that we built for either internal consumption or foundry customers. That did not materialize for all the reasons that most of you know. The part of your question is, why didn't we see it? What were the signals that we missed? Yeah. These were signals. It's not that we missed it. We missed the magnitude of the cumulative effect of China and the lack of foundry customers for this particular investment we made happening all at the same time. That was the magnitude that we dealt with inside the quarter. That is why, as we look ahead, we decided to de-risk both of them as we look into Q4 and FY2026. OK. For IP in particular, I emphasize for the rest of the business we're still forecasting in line with expectations. Yeah. Last issue, just on the IP sort of portfolio and roadmap and emphasis and revenue model, talk about what that is because I think many investors are probably more concerned about that than they are either of the first two issues, which is just, is this a, tell us maybe a little bit about what it is. Is it a product focus? Is it a delivery focus in terms of integrated versus customized? Is it off-the-shelf versus customized? Maybe just kind of talk about how long it takes to fix. The IP business, we've been in the IP business for 26 years. We know the IP market very, very well. We know the customer trends and the customer roadmap very well because the first engagements with our customer start with IP. The market over the last four or five years has been, I want to say, pivoting from an individual discrete IP delivery or off-the-shelf IP to a subsystem, to more customization around the IP. That's primarily for hyperscale, edge AI. To some extent, the automotive market is starting to demand something similar. What do I mean by subsystem? They come to Synopsys and say, I need the following list of IP from you, but I need you, Synopsys, to do the work, to put some logic, some test, and integrate the set of IP and deliver it to me as a subsystem or potentially some discussions as a chiplet. That's a great opportunity for us. We've been investing in that opportunity. We've been selling to these opportunities. Each one of these opportunities is very unique and customized, so it's very resource intense. What we are doing in order to increase our capacity is, from an architecture point of view, making the engagements more efficient from one engagement to another, even though they're customized. From an architecture point of view, we are looking at how to leverage AI everywhere in our workflow to improve the efficiency, the quality, the speed of delivering that IP. About three months ago, we merged a couple of divisions inside Synopsys to accelerate and improve our capacity to deliver to these customized subsystems. The other part that we're having the discussion with our customer is the business model. The way we sold IP traditionally, you build it once, you sell it many times. As it started moving to customization, we started charging for an NRE for the customization and a use fee. Now, given we're delivering a big part of the chip as a subsystem, the discussion with our customer is an NRE, a use fee, and some sort of a royalty attached to it, given the investment and the commitment we need to make for each one of those engagements. I see. Does the shortfall you're seeing now reflect something in terms of your ability to deliver that customized, subsystem-based solution with that number of broad engagements, or is it more tied to the pricing model you're trying to achieve? It's truly the headwind came from the first two factors. It was the China disruption and the discontinuity with our foundry customer. The third reason, which is more prioritization on resources, how do we scale our workforce to be able to deliver to all the opportunities out there? Actually, the one example I gave earlier today, UCIE, just as a simple example, it's one of those IP titles that is needed to connect chiplets and dies together for the advanced package. Today, we have more than three dozen committed engagements on UCIE with our customers. That's business that we booked and we're delivering to. There's another dozen opportunities out there that we can go capture, but we don't have the ability to capture because we're tapped out on the resources. It's the ability to make fundamental methodology, architecture, evolution to be able to become more efficient, to capture more, and priority and choices that we need to make for these opportunities. OK. Great. Thanks. Maybe just relative to Ansys, because I got that question as well. Is it correct to say that the Ansys revenue base is coming in as you expected? Can we talk about the seasonality in their business versus the seasonality in yours and how that kind of impacts what you'll report in the next few quarters? Yes. Thank you, Jim, for that question. Because yesterday, what was lost in our earnings call, there were assumptions made because we, again, had a very short stub period in Q3 for Ansys revenue reporting. We reported it as just one line item. The expectation and the feedback was, can you report as Ansys, not only the simulation and analysis? Because Ansys had two parts of their business, simulation and analysis and semiconductor. Can we continue on reporting for transparency and visibility on the health of that business as a separate reporting moving into FY2026, into Q1, Q2, etc.? The answer is yes, because our core business is doing so well. I don't want the assumptions to be that the core business may be in bad shape. That's why Synopsys is not providing that transparency. Given it was, as I said, less than two weeks stub period, we just decided to only report the simulation and analysis. Ansys is performing in line with the expectations. So was or is the rest of the business at Synopsys. OK. Finally, relative to cost, you implemented a 10% headcount reduction by the end of fiscal 2026 as part of your cost synergy program. How broad is that reduction across what functions? Was that already planned as part of your Ansys acquisition? How should we think about comparing that to the $400 million cost synergies you outlined a year and a half ago plus? Yes. During the regulatory process, it's very difficult for Synopsys Classic, with or without Ansys, to make big decisions on portfolio or make big decisions on headcount. Starting in Q1 of FY2025, as the Synopsys Classic, we went through the exercise of both the portfolio as well as resources. What is it we can do once the Ansys acquisition is complete to accelerate the synergies? The $400 million run rate was planned to be achieved by year three. The 10% that we communicated yesterday is how do we bring it earlier and give Synopsys the opportunity to achieve the $400 million, as well as reinvest in the right places in the portfolio to accelerate our opportunity. is still $400 million. You're just sort of bringing it forward. Bringing it forward because we want to reinvest in the business in many places, as I said, where we have the opportunities in front of us. It is about further scaling to capture the opportunities. In line with that, what we communicated as well is our operating margin in the mid-40% as our long-term target, which we still feel very strongly we can achieve. OK. Very good. OK. Enough with earnings. Let's talk about some longer-term issues for a moment. We've heard a lot about artificial intelligence this week from many quarters. I think the EDA space is a very interesting crossroads with respect to AI. First of all, I think you have a great and perhaps unique view into your customers' roadmaps by proxy of their design starts. What are you seeing with respect to customer design starts? What's the level of chip activity in terms of starts or tape-outs? Do you see it slowing down any time soon or accelerating? The way we look at the market, and we started communicating this a few quarters ago, is like there's a tale of two markets in semiconductor. There are the set of customers that they're serving the high-performance compute, data center build-out, and infrastructure. There, it's a different world than the rest of the semiconductor industry that is serving automotive, industrial, et cetera. The speed in which the roadmap is accelerating is very different between these two cohorts of customers. Now, you zoom out, and for Synopsys, we have many new entries, customers entering either designing semiconductor directly or working in an ASIC type of a model where they do some architecture definition, early software development. Then they hand over to a partner to design the chip for them. That has been driving a very nice opportunity for Synopsys. You can name any hyperscaler that falls into that category or automotive. Increasingly, we're seeing other categories of customers trying to not necessarily design the chip themselves, but provide the spec for semiconductor companies to design the chip for them. That's the trend that we're seeing at the high level. OK. I think relative to your use of AI in your tools, maybe talk about what benefits that provides to your customers and design engineers, whether in terms of reduction in design hours, productivity, time to market, whatever. If you can give any kind of metrics around that. More importantly, how can you charge for it? How can you charge more for it, drive value? At the highest level, we're thinking of AI as three categories in our offering. The first one is reinforcement learning that we're embedding inside our existing products. We started doing this around the 2017 time frame, and we released the first wave of those products in 2020. We started deploying at customers, selling this technology. That's, think of it, any part of the chip design flow that you can use reinforcement learning to optimize and get results better, faster, we have it. The monetization there is similar to EDA. Customers, when they buy the software from us, then it's becoming table stake. I want to get the AI-powered product, and there's an uplift for that technology. The second one we started about two years ago is consolidating the knowledge base of our know-how of our product to make sure it can be modernized under the generative AI umbrella for an LLM-based interaction with our product, as well as can you create part of the design. For example, can you create a test bench? Can you create a documentation for an RTL? Can you finish writing up the Verilog for me? That's the generative AI bucket. The third one is Agentic. You cannot get to Agentic without having the middle piece, which is the knowledge base. In Agentic, we classified it as three levels, L1 through L5, very similar to autonomous driving. L5, you can think of the full autonomy chip design for a particular part of the workflow. L1, L2 are specific tasks that you can assign agents to do the work, and level three is orchestration between the agents. Today, we are between L2 and L3, where we have a number of task agents, and we're in the early phase of orchestration of these agents. The monetization for the middle layer, generative, we're not expecting monetization in it because it's a necessary step for phase three. With agents, the moment the workflow will change, it's a good opportunity for us to demonstrate that value to our customer and see what is the monetization opportunity for the Agentic. Interesting. OK. Maybe talk a little bit about, you know, I think just kind of continuing on AI, talk a little about the rationale for combining with Ansys to begin with. I mean, I think if you think about, I mean, I'll just articulate it in the way I think about it, which is a lot of these AI-constrained designs, whether it's thermal or EM interference or what have you, they all sort of kind of go to the combination of the chip design and the physical design. Maybe talk about what level of kind of synergistic opportunity there is from a product standpoint. More importantly, how long does it take you to actually co-develop products and kind of get there to where it actually makes sense and you can sell it to customers? Yes. That's the part where we're incredibly enthusiastic, excited about delivering the joint solutions. If you look at it at the macro level, most advanced chips that are today delivering to either high-performance compute or any type of acceleration, they're turning into 2.5 to 3D type of chips. The biggest challenge that our customers are facing are physical challenges. The electronics design of these chips, I want to say the methodology is well understood. Customers can plan and design for timing, for power. As they go into the packaging and in-field use of these chips, based on a workload, there are many failures that are happening in the field, be it thermal. In many cases, when you're stacking these dies on top of each other, they may be warping or cracking. There are mechanical failures that are happening in the advanced package world. Ansys is the leader in physics simulation. Today, the way the workflow works, you put a lot of margins in your design in anticipation of these issues. It's costly and it's not competitive. In the first half of 2026, we committed to our customer the first wave or the first horizon of delivery of the physics simulation integrated inside the Synopsys design platform. That's a great opportunity that our customers are seeing in the combination of Ansys and Synopsys. That's an immediate opportunity we see. We know exactly what we need to do, what we need to build, and drive it. If you're a system company and you're thinking about physical AI, how do you create a digital twin or a virtualization of your system and taking into account the physical effect of the real-world physical environment? I believe, yeah, it was in Q3, we announced one of the partnerships, and there are many partnerships, but with NVIDIA, where Synopsys go-to-market, and that's the team responsible for the Ansys go-to-market through our channels, we will be selling Omniverse, where Omniverse will be doing the visualization of the system. The Ansys solution will be doing the actual simulation of the real world and physics. There's both the advanced package, which is an immediate opportunity. As you look ahead with the AI accelerating into a physical world and application, that's the other opportunity we see with the acquisition. Maybe just kind of speak to structurally for the core EDA business, what opportunity do you see to actually accelerate the growth rate in core EDA, either via the use of AI tools or the synergistic effect of the products with Ansys as you already outlined? The key driver for our business is complexity. If our customers are sticking to the same technology node, same chip, then our business will be muted. When our customers are racing to move forward to the next technology node, next complexity of chip design, it gives us the opportunity to sell our latest product, including AI, the latest product, including the Ansys integration. Complexity is what drives our opportunity. The second driver of our opportunity is the silicon design is becoming pervasive in terms of the need to optimize and customize that silicon for a specific workload. That's the other opportunity that drives our business growth, not only on the EDA software, but the hardware that we have, which is emulation, prototyping to build the software before the silicon is available. We're seeing a great momentum in that part of the business. On your IP business, I'm sort of curious as to if you think about either the hyperscale companies or the system companies who are taking your solutions now, do you feel that they are going to be bigger consumers of IP relative to your traditional chip design customers in the future? They have been over the last seven years or so a big driver of our growth. Even in some cases, if a hyperscaler is not designing their own chip but needing to have early visibility of the system performance through virtualization of chips or through emulation of the chip so they can start their software development early, that part of the business has been a great driver from the hyperscalers. The other part of the business with hyperscalers, most of them are looking at three options: buying off-the-shelf merchant, doing some sort of customization with ASIC partners. A lot of them are actually going deeper into the chip design process, which is a great opportunity for IP, EDA hardware that we offer. OK. Great. Maybe I'll just kind of ask one more question, which is, you know, we talked about earnings results. The stock has dislocated quite a bit. In your conversations with investors last night and today, what is the one thing that you think is maybe misunderstood about the Synopsys story and the thing that you think you would encourage investors to keep in mind relative to their thesis on the stock? Yeah. Thank you for asking this question. Actually, one of the investors made the comment, we don't think Synopsys today is worth a third less than what it was yesterday. That's how big the drop happened today. Of course, you can always look back and manage how much transparency and the effectiveness of the transparency that we communicated. We wanted to communicate upfront that there were two factors, the China and the foundry customer investment, that had an impact in particular for our IP business. We don't want the expectation to be that that was just a Q3 expectation. We believe these two factors will remain for the foreseeable future into 2026. We wanted to de-risk for IP in particular these two factors. When you look and you're modeling your 2026, model it with a muted growth, not the same level of expectation we had before, which is mid-teens growth. We will get back to the mid-teens for IP in particular into 2027. We'll accelerate our growth, but we needed to provide that visibility and transparency. As far as the rest of the business, it is very healthy, it's performing exactly in line with the expectations. Lastly is the Ansys segmentation. They wanted to make sure, given it's a $35 billion acquisition, how well is it performing as a standalone. We have no problem providing that information starting in Q4. Great to hear. Thank you very much, Sassine, for being here with us. We really hope you appreciate it. Thank you for providing this notification today. Thank you so much. Thank you.