Conference Transcript - SNPS 2026-03-03

Speaker 3

Okay. Good morning, everyone. I think it's still morning. And welcome to San Francisco. This is the TMT Conference, Morgan Stanley, 2026. And I'm glad to welcome to the stage, Sasin Ghazi, CEO of Synopsis. Sasin, welcome. Thank you. Maybe just to help level set everyone, maybe it was only just last week we did the results, so maybe if you just help us summarize what it was you highlighted for Q1 results, what you said for Q2, and maybe a little bit about the 26 guide.

Sassine Ghazi, CEO

Before I go into the Q1 and the year, it's important to point out that FY26 will be the first year that we have the combined companies, Ansys plus Synopsys, coming together. And that provided a significant expansion for the opportunity that we have. Synopsys has gone from a silicon design solution to silicon to systems engineering solution. And the timing could not be better for these two companies to come together. As you look at the opportunities of physical AI, you look at the complexity of chips that needs to go into these systems, the holistic approach of designing these systems from silicon to system becomes essential. We delivered solid Q1. We said what we're going to do, and we did it. And we delivered on the revenue toward the top line of the guide. We beat on EPS. We reiterated the guide for the year. So it's a great start of the year and gives us confidence to the guidance that we provided.

Speaker 3

I made one mistake. I should have read the disclaimer at the top. Go ahead. Even if I just do that now, we all pretend we didn't do that. 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 Synopsis' most recent SEC filings for a discussion of risk factors and that may materially affect these statements. Sorry about that. No problem. Back on track. So maybe just with all that being said, now with the integration of Ansys, what is driving the growth as you look across the group? It looks to be from the outside as though IP and hardware are part of that growth, but how do you see things this year?

Sassine Ghazi, CEO

Yeah, again, if you zoom out and you look at where are the opportunities opportunities from a market point of view. The complexity of designing the more sophisticated chips for AI applications require deeper integration or what we call co-design, co-design between electronics and physics. That's one area. The second area is as you envision the world with many companies trying to invest and deliver to more intelligent systems, be it a car, a robot, a drone, etc. How do you design these intelligent systems by re-engineering the way that you engineer these systems, creating a digital twin, be able to reduce the cost, etc. So that's the second area. The third area is around agentic AI. It's such a great opportunity to tame the complexity of how these products are designed. So when you think of the electronics and physics co-design, Synopsys with the new portfolio will be leading with having the physics sign-off, be it thermal, structure, et cetera, come into the design phase of the silicon. Actually, next week we have our conference called Converge, where we'll be announcing a number of the joint solutions across these three vectors with the digital twin as long as the agentic effort that we're building.

Speaker 3

Gotcha. Maybe if we just stay with the agentic effort, as you've outlined here, it does look as though you are starting to see a little bit of a competitive advantage coming to yourselves relative to some of the peers. And I think on top of that, there's been some sort of suggestion that this will be value-based pricing. Maybe help us understand how could that be done in practice? Where are you seeing customer interest arising? And maybe the timeline as well for agenism.

Sassine Ghazi, CEO

The way we thought about AI is to tame complexity of the work our customers' R&D is dealing with. So we started an investment in 2017 around reinforcement learning, and we inserted reinforcement learning in every opportunity we have into our product portfolio. We started selling that solution around 2020 timeframe with a similar business model as we've had for EDA, which is license consumption-based. The next wave was generative AI. With generative AI, that really changed the user interface, the way the user deal with our technology. Because our customers are dealing with very sophisticated, complex problems they're trying to solve. How do we ramp up new engineers, or how do we make the existing engineer more efficient? So think of it as an assistant for the engineer with generative AI. A year ago, we announced our vision and roadmap for Agentic. We think of it as a series of task agents or agent engineers with a cognitive layer that you can orchestrate around these multiple agent engineers to change the workflow. The monetization opportunity and the new business model, we don't believe will happen unless the workflow will change. With agents, it will change the workflow. When the workflow changes, you're not counting anymore how many licenses am I consuming and am I signing a three-year agreement on-prem to get access for these licenses. is the delivery mechanism is very adaptive because those models are going to change at a much faster rhythm than the traditional software that we release every nine months. Most of our customers run that software on-prem today. Then there will be a layer of that, once you move into these agent engineers, that you need access to more compute, so therefore a cloud model. And the outcome is going to be significant improvement to the time to get to results and the quality of the results. So that's the opportunity where we see the workflow will change, then the monetization opportunity will be different.

Speaker 3

Okay. So the workflow here, as far as, you know, when you say workflow changes, do we see steps being shortened materially? Is it optimization on certain, like place and route, for instance? How does it all work?

Sassine Ghazi, CEO

So let me describe it at the semiconductor chip design as well as the simulation and analysis because they're different user persona and workflow that you need to think through. At the chip design, the chip design is multiple parts of that workflow. You start with the requirements, and you start really the chip design very much as a software entry point. Then you go all the way to the last phase, which is process technology manufacturing physics-based. Because if you're going to manufacture, say, a TSMC or Samsung or Intel or GF or whomever, you need to make sure that you have the physics representation of that process technology taken into account during the first phase of the chip design. So the workflow has multiple steps. Some part of the workflow, the agents can do a great job to augment the human engineer and take on certain tasks to make the engineer more effective and efficient. Other part of the workflow is better quality of outcome, of results, faster. Some part of the workflow is very difficult to bring in an agent to take on the task. Other part of the workflow and agent will be a perfect fit. So the way we mapped it out last year, they're going to be L1 through L5 in terms of levels of maturity for agents, from a task agent to the orchestration, et cetera. Today we are delivering number of tasks agents or agent engineers with some sort of a cognitive layer to orchestrate these agents, and we're in early adoption with customers to see what is the impact overall on the time to design the chip and the quality of the chip because with the objective to improve the power or the performance or the cost of the chip. So when we think of a workflow in the chip design, it's a series of tasks and which tasks can be at this stage delivered to an agent to deliver to it. In simulation and analysis, it's slightly different. In simulation and verification in general, it's all about how much verification can I get done in the fastest and shortest time possible. Because there the bottleneck is time. AI is a great opportunity to accelerate. And we're working on a number of other investments beside AI. There, for example, compute acceleration, like GPU, is another opportunity where, for example, a fluid dynamic simulation can be accelerated by 100x, 300x. That's a significant improvement on a task that may take weeks that you can reduce to days or hours.

Speaker 3

Yeah. Incredible. A hundredfold increase. So maybe if I just try and just quickly summarize there, you're talking about various optimal points across the entire flow. The flow itself could introduce, you know, different agents, maybe multiple agents across the flow. And then there's a difference between the EDA flow, obviously, and what you're doing in simulation analysis. But therein, you did talk about verification. And there is maybe a concern in the market that some of the coding that's done, pre-verifications would go to RTL, for instance, could be done utilizing models at customers. And yet, you know, you have tools that have got decades of data algorithms behind them and data layers. So how do you deal with that competition, that potential risk in the market?

Sassine Ghazi, CEO

Yes, there are part of the workflow, they're very similar to a software development. You write code. That part of the business is very, very small because our customer owns writing that code, not Synopsys is trying to write the code for the customer. The tool we provide, once you write that code, is it going to be functioning? And as you go deeper into implementing it into physics, are you able to manufacture it? So it's not a code that you write and you say, okay, that performed that task, I'm done. There's a physics element of it. The code that is written needs to be physically implemented and that physical implementation is still using software. Then once you go into the manufacturing step, that you need to make sure you have the physics of manufacturing taken into account when you're implementing that physical implementation. Those are series of complex solvers that we have built over decades in collaboration with the manufacturing technology and at the architecture level. We are, when we talk about agent engineers and tasks agents, we are leading with the disruptions that we say with AI, there are certain tasks that how do we leverage the solvers that we have with the cognitive layer or any foundation LLM that is out there and provide a new, better solution for our customers. So I don't want to take the thunder out of next week as we're at Converge, but when we talk next week about the series of advancements we've made with these agent engineers is quite remarkable. And the reason we have the advantage is because we own the solvers. And our customer, from a data point of view, et cetera, we have that entire visibility to build these agents.

Speaker 3

And so, intimately, the agent engineer will call the solvers, the underlying software tools, so you get a much more deterministic outcome for the client.

Sassine Ghazi, CEO

Exactly. And then the next phase is when you have an orchestration of these multiple agents. So that's where the workflow will change.

Speaker 3

Maybe just changing to the new NVIDIA relationship that you guys talked about a month or so back and trying to understand the sort of commercial traction for Synopsys here, timelines to use, the work that you're going to do around Omniverse. Maybe just help us understand how does that work? How do you monetize this? And when does it all happen?

Sassine Ghazi, CEO

So we've been a partner with NVIDIA for decades, and what triggered that deeper partnership is really the expanded portfolio. When you look at the Synopsys portfolio, historically has been targeted and focused to semiconductor companies, the companies that they're designing chips. As these chips are sitting in a very complex system, be it a data center or a car or a humanoid, etc., in order to achieve the best total cost of ownership of these end-intelligent systems, a lot of customization is happening. That's where you see a plethora of architecture, ASIC, customization, COT happening with these end-system companies. The complexity of thinking of the system as a stack from silicon to software and to the system level, the end product level, is a complex engineering task. And today, the way our customers are trying to differentiate is how do they reduce margin and improve the cost across the stack. With the ANSYS portfolio, we're very uniquely differentiated because we own the physics simulation with the design of the silicon. NVIDIA could see that. They could see that as they're designing themselves a system and as they're trying to capture the opportunity of the future, we have a massive workload for engineering. So today you cannot think of any type of engineering, be it mechanical, electrical, electronics, et cetera, that is not using synopsis one way or another. Most of these workloads are running on CPUs. Most of these workloads need to be accelerated. So the closest opportunity to accelerate is a GPU. We have a roadmap that we are committed to deliver with joint R&D with NVIDIA by the end of this year. And we have some proof points of number of technology that we have invested prior to that announcement with a range of anywhere between 10x to 20x, 25x improvement to a CPU. That's a great monetization opportunity. So what the customer will see is from many weeks to fewer weeks or days that they're willing to pay extra money and value for that acceleration. That's one layer of the collaboration, is the GPU acceleration. The next layer of collaboration is the digital twin of these intelligent systems. With Omniverse, think of it as the cockpit to bring in an end product requirement. And before you go into the manufacturing of that product, you need to validate, will it function in that simulation world, in the physical world, in the real world? So you need an accurate physics simulation in order to go from the digital design to the physical design. And again, this is where the ANSYS portfolio today, if you think of a car or an airplane engine or any applications, a drone, a robot, there are mechanical aspects that are driven by electronics. that you need to make sure you have the right either gas or electric battery mileage, you have the right fluid dynamics to improve the efficiency of the product, et cetera, et cetera. And this is where the Omniverse partnership is about.

Speaker 3

Perfect. Sounds very exciting, actually. So we look forward to March 11th was the event. Superb. Maybe if I take you back to, however, Q3 last year, We did see a couple of issues come up, which do look like they're now going into the rearview mirror, one of which was in China. Maybe you could help us understand again. It did seem as though IP in China was hitting something of a roadblock and there were some roadmap issues going forward. Can you just maybe outline what was happening there and if indeed China as a growth story in IP comes back for synopsis?

Sassine Ghazi, CEO

Yeah, so let me describe China and then IP as both separate and how did they impact one another. China for a number of years was a very fast-growing market for synopsis, primarily driven by the number of startups. If you go back to 2019, 2020, 2021 timeframe, the number of startups in China doing chip design were at a pace much faster than any other region in the world. That was a fantastic opportunity for Synopsys, not only driven by IP, driven by IP and the rest of the EDA portfolio. Then what we've seen over that period of time since then, the number of startups are shrinking. And there's a headwind coming from the cumulative impact of restrictions on technology in China. The impact on Synopsys was more significant than our competitors because our position in IP in China was much bigger than any peer that we compete with in China. So that's the China dynamic. So what we communicated for FY26, we are de-risking China, meaning we're assuming the environment in China will not change. So that's the China aspect. Now, from an IP portfolio point of view, we have the largest in terms of breadth of an IP portfolio that serves multiple markets. So today, if you are a semiconductor company designing for automotive, for industrial, for mobile, for HPC, Synopsys is your partner of choice. If you're designing on TSMC, but the next chip is on Samsung, Intel, GF, we have our IP available on all foundries for various markets. That has been our business for about 25 years on IP. So you build it once, you sell it many times, and you provide the IP before the customer needs the IP. The trend over the last three, four years, especially with AI-driven semiconductor chips, is a lot of customization of that IP. And that's where you see companies like Broadcom have been benefited greatly from customization in ASIC. And then you see many of these hyperscalers designing their own chips in order to improve their TCO for specific workload. That's a fantastic opportunity for Synopsys. But it's a change in opportunity, meaning we still have to deliver on that broad portfolio that you build it once, sell it many times. And we need to build a new lane of customization for these opportunities. So what we communicated is for that new lane of opportunities, we have to approach it with a different business model and with a different approach with these customers. Because if you take the top one or two ASIC companies out that they have their own IP, the rest of the ASIC companies rely heavily on synopsis for their business. They cannot have an ASIC business if we don't have our IP business. So they're very open to adapt to a new business model. And same thing with hyperscalers. As they're building COT, they're not building an IP group. They need the IP to come from Synopsys, but we need to customize it. So when we talked about FY26 will be a transition year for IP, it's for us to continue on serving the current business model and market, and as we adapt and allocate resources and investment to serve that new opportunity.

Speaker 3

And that adaption of an IP business model, if I remember right, you used the word pivoting to subsystems as a design, which seemed like as though you were dancing close to the idea of making chiplets or custom designing chiplets for perhaps ASIC players or hyperscalers. Is that the right way to see this going forward? And do you see this as a bigger time of dollar capture for you guys?

Sassine Ghazi, CEO

Our customers are pulling us in that direction. And by the way, when I say customers, in this case, are the HPC customers, the hyperscalers and the semiconductor companies building to serve that cohort of customers. We've been evolving from delivering an IP as a standalone small block to a bunch of IP working together that sits on a bigger part of the chip. We have been down that journey for a number of years now. But what we're trying to change is the monetization of that work. because the way we've sold IP is you buy the IP for one use on a chip. The next chip, you buy the IP again. The next chip, you buy the IP again. It's been very healthy and a great business model. Now with the customization, you have a use fee plus an NRE. But given the demand on the resources, and these are scarce resources, We need to have a use fee in NRE plus some sort of a share. Think of it as a royalty for these engagements. And that's the period we're in right now. We're in a number of discussions with those customers to pivot to a new layer of monetization for that resource investment that we're making. But we're not going all the way to build a chip. And the reason we're not, because we are an ecosystem. The moment we decide to go all the way to build a chip, then we're competing with the whole slew of our customers. I would rather sell to many, many, many of the customers and monetize better versus building and competing with the customer.

Speaker 3

Okay, so there's still value to be had coming in and enabling subsystems for customers, utilizing resource to get there, and then taking a royalty in the market afterwards.

Sassine Ghazi, CEO

Exactly, because in this model, we can engage as an ecosystem to many of either the ASIC or chip companies that are serving that market or the hyperscalers themselves that they're building that chip.

Speaker 3

Gotcha. Okay. Maybe if we turn to foundational IP, and I think you're a number one customer, there wasn't a pull-down on their customers at a certain process node. And the suggestion, I think, from yourselves had been and maybe just look at that as potentially a zero for this year, the foundational IP from the number one customer. Is that still the right way to look at this? And how about foundational IP as a general opportunity for the next two, three years?

Sassine Ghazi, CEO

So, by the way, from a terminology point of view, there is the foundation. We have two categories of IP. We have a foundation IP and an interface IP. So think of them as both are needed to work with a foundry in order to on-ramp customers on their node and technology. What we said is for that particular foundry customer, the way we're approaching FY26 from a guide point of view, we're assuming that there will be no new design start with that foundry customer in our guide. If there are, great, that will be an upside. But we're de-risking it from a guide point of view. Now, we can argue you're being pragmatic with that approach. Because we have the transformation of the IP opportunity, as I just described, that we need to make sure we capture it as well. From the long term, the opportunity for IP is massive. For all the reasons I mentioned earlier, If you're building a chip and you don't want to just buy a general purpose from whomever you're buying the general purpose chip for your data center, and ASIC, you're limited to very few players, you want to build either your own chip or work with a broader set of ASIC companies to customize the chip for that third cohort. You cannot get to it without the Synopsys IP. And that's the opportunity that we want to capture and make sure we have the resources and the investment allocation for that market.

Speaker 3

Gotcha, that makes sense. One more for me, and then I'll open it up to the floor. Just on Ansys and the integration now, it seemed as though on the call last week that you pulled forward some of the synergies, the cost synergies, maybe to next year, but possibly even this year. Is that the case? And can you maybe help put some numbers around that for us?

Sassine Ghazi, CEO

Yeah, actually, it was, by the way, thank you for bringing this point because it does require some attention to share with you the strength of our balance sheet. We had short-term debt to be paid off over a three-year period. We ended up accelerating it and paid it off in six months. So that tells you the strength of the balance sheet we have, of the position that we have. So, yes, that was paid off from a short-term debt point of view. And actually, just Monday, yesterday, we have announced 250 million buybacks with, again, that we believe it's a great return given where we are from a market point of view. The other priority that we're driving, we said there will be $400 million in cost synergy that we'll achieve in three years post the ANSYS acquisition. That is being accelerated as well. We have confidence that we'll be able to accelerate and achieve the $400 million in cost synergy much earlier than the three-year point that we set initially. Perfect.

Speaker 3

Maybe I will open up the floor at this point, see if there's any burning questions. I've got one down here.

Speaker 5

Thank you.

Speaker 1

Cecilia, more of a near-term question, but one of the concerns that analysts are expressing is around just the core EDA growth slowing from the high single-digit range, even while we're seeing all this strength in AI compute demand. And so I guess my question is, why would the core EDA growth be slowing in this environment?

Sassine Ghazi, CEO

Yes, so thank you for the question. How do we monetize core EDA? Either our customers are investing more in R&D, and we get a percentage of that increased investment, or they have a need to use more of our new technology. New technology, be it the GPU acceleration, the joint solution with Ansys, or the AI technology, etc. We have a great monetization happening today in EDA for that cohort of customers, or we call it the tail of two markets today that we're selling into, that are investing more in their R&D, and they are adopting the latest technology to tame the complexity they have. But if you look at the broader semiconductor market, there is still a big chunk of companies that they're not investing more in R&D, and they're not racing forward in their roadmap. Net-net, when you aggregate these two, this year we're going to be delivering close to a double-digit for EDA. But the long-term confidence in EDA as double-digit, same thing with simulation and analysis as double-digit, as IP in the mid-teens remain, given the need for that sophisticated silicon that our customers are building.

Speaker 1

Thank you.

Speaker 3

Thanks for that question. Maybe just staying on that. There's a question there. Was there another one?

Speaker 2

Just following on the same question, I was curious. Given the market structure that exists in EDA, why do you limit yourself to the R&D budgets of your customers? Why not think about their revenue, their demand equation, and try to tap into that value creation versus say, okay, if I'm customer A and I set my R&D budget at X, well, I hope to get a percentage of that. When clearly those customer companies are tapping into a demand equation that is massively accelerating.

Sassine Ghazi, CEO

On the EDA side, the percentage of R&D dollar that's coming to Synopsys and our industry has been increasing due to complexity, due to primarily complexity of what our customer is doing. Where we're seeing the opportunity to capture exactly what you're saying is in IP. and IP is exactly what we're positioning that change in the business model from the historical 25 years of just a use fee to capture that dollar amount. If you look at percentage of R&D that the EDA used to capture about 10 years ago, it was in the single digit, 7-ish percent or so, where right now it's close to 12%, and that budget of R&D is increasing. The change in the workflow is the new opportunity to change the way we capture value for the impact we're delivering to the customer.

Speaker 5

Maybe we've got one more down front. Thank you.

Sassine Ghazi, CEO

I guess your EDA business is fundamentally a SaaS model. So the question is, are you yourself vulnerable to AI disruption? It's not really a SaaS model. Today, it's license-based that is on-prem primarily. So the way our customers today, they assess how much money do they spend with us is based on their number of engineers and the tasks they're running. So in many cases, you have an engineer running many, many licenses. In other parts of the workflow, one engineer running very few just because of the interaction and part of the workflow. As far as the AI impact, AI is a great opportunity for us. It's actually driving the opportunity to monetize more. I mentioned earlier reinforcement learning, et cetera. Those are techniques that we have product and we're monetizing. And remember, what we deliver, it has a software layer that is driving deep physics solvers and engines that we do. Therefore, the AI opportunity to improve the user interface with generative AI, etc., we're leading with that innovation to our customers. With the next opportunity of agents, we are leading in building these agents with the solvers that we have. So we see it as absolutely a tailwind to simplify what our customers are trying to do, which is very complex already both at the silicon level as well as the system level. I see we're out of time Sasin thank you very much thank you so much thank you