My name is Tyler Radke. I co-head the software sector here at Citi. Thanks everyone for joining day two of our tech conference. Really excited to have Ansys here. We have Ansys' CFO, Nicole Anasenes, and then, Kelsey from Investor Relations. I'm gonna turn it over to Kelsey first to read the most exciting part of the presentation, the Safe Harbor- Yeah. Before we get into the discussion. So today's session may contain forward-looking information. Actual results and future events could differ, possibly materially, from those anticipated in our statements and from historical performance due to a variety of risks and other factors. Information about such risks and other factors, as well as GAAP reconciliations and other information on non-GAAP financial measures we may discuss, is included in our SEC filings and investor materials. These statements speak only as of today, and Ansys undertakes no obligation to update them. Back to you, Tyler. All right. So Nicole, thanks for joining us. It's great to see you. I thought we could just kick off... You know, everyone is always curious what you're seeing on the demand front. So at a high level, could you just talk about the conversations you're having with customers? How is demand trending, and what. It's still kind of an uncertain macro environment. If we could just kind of start off on the high level, and then we can kind of get into the specific end markets and businesses. Sure. Thanks, Tyler, and thanks for having me, and, appreciate everyone joining today and hosting us at the conference. So yeah, so first, maybe it'll help to contextualize my answer a bit and, and talk a little bit about, what aspect of customer demand we see, what drives our customer demand. And so what drives our customer demand is R&D, right? So the more, the more R&D that happens, the more designs that get generated, the more complex those designs, the more simulation that runs. And so in the context of the macro question, the macro backdrop, you know, what we tend to see is that R&D is, is one of those parts of the business that is the, the last to get turned off and the first to get turned back on when you have disruption, right? And so what we have seen more recently among the more recent macro dynamics that maybe other software companies have seen is that it, you know, it is not yet—there, there isn't that impact at the R&D function. Mm-hmm. And so what we have seen, you know, last year and going into the first half this year, is really broad-based growth across industries, geographies, customer types. And it really is driven by the need that our customers have to continue to deliver high-quality products on time with low warranty costs, in a time of increasing complexity. And so we've seen that kind of sustainability of that demand, even though in some of the other, there are other subsets of different areas of technology and software that might be a little bit more subject to the more overhead type of decisions that many companies are currently entertaining. Okay. So pretty consistent demand environment this year versus last year, and even second half versus- Yeah ... versus first half. Yeah, absolutely. Got it. Got it. Okay, and, and I guess on the, the most recent results, you know, you recently reported Q2 results, which beat the midpoint of your guidance, and, and you did raise revenue and ACV for the full year. I think the, the third quarter guidance kinda caught some investors by surprise. I know there's a lot of nuances in terms of 606 accounting and, and large deal lumpiness, but maybe just for, for investors who may have only seen the headline numbers, just kind of level set the puts and takes on the quarter and, and the assumptions in the outlook. Sure, sure. We probably don't have enough hours in the day to go through all of the machinations of what accounting does, but let me start at a high level. Let's just start with the key metric that we focus on, which is ACV. One of the reasons we focus on ACV, annual contract value, is because it is a more normalized metric that really looks at the contracted value that customers have on an annualized basis, which is highly connected to the cash we collect from them on that basis, right? The accounting dynamics around those transactions and the license types that make up that annual contract value very often creates volatility in the P&L, which I'll talk about in a second. That is the artifact of the Q3 P&L guidance that you're referring to. So if you just look at ACV, just to kind of level set, the ACV outlook has been very consistent. So we delivered 12% constant currency ACV growth in the first half. The implied guidance range for the second half is, you know, 11%-16%. It's 11%-15% in Q3. It's, you know, 11%-17% in Q4. That ACV number has been very consistent. Mm-hmm. The actual percentage of ACV from a dollar value standpoint that we expect to deliver in Q4, that percentage is very consistent with what we delivered in prior four quarters. So there's no leaning in, there's no anomalies there. The dynamics you're referring to, it really relates to the revenue guidance in the relationships between Q3 and Q4 implied revenue guidance that make up the second half outlook, and that is an artifact of accounting. And so what happens is, you can have similar types of ACV, but if the mix of license types that you have that make up that ACV in a given quarter has relatively different upfront recognition mix versus ratable mix, you can get dislocations in your revenue growth number. So let me use an example. So last year, we had in our prepared remarks, we had referenced several seven-figure deals. We referenced that one deal in particular was $59 million, right? So let's just take that example of that $59 million deal last year, and what that looks like last year and this year. So from an ACV standpoint, that $59 million deal would be roughly $20-ish million a year if it was a 3-year deal, right? 20, 20, 20. That same deal from a revenue standpoint would have recognized $30 million last year, and this year in that same period, you would recognize the ratable component of that 36 months. So 1/36 of the remaining $30 million. So there you've got 20 and 20, same number, versus 30 and a percentage of 30, which I'm not doing the math effectively in my head to know. So that's how you can get dislocations. It doesn't happen often. Most of the time, ACV and revenue can be at the company level. It's highly correlated, but on occasion, you get these dislocations. Mm-hmm. Always and every quarter, when you get down to a country level, we can have these dislocations- Yeah ... because when you have those compares. So that's really. There's no underlying change in the business momentum. As I said, the business momentum is quite strong, ACV being the indicator of that, but the accounting dynamics, which kind of lead to the lumpiness of Q3 versus Q4, are really just an artifact of license mix. Okay, great. Well, hopefully that, that clears it up, and- Yeah. you did all the... I can just point people back to you when I get those. Okay, no problem. those questions. So I want- Maybe we're gonna train a GPT model- Yeah, yeah, exactly. ... to be able to answer that question. Exactly. Okay, good. That'll help Kelsey out, too. I wanted to dive in a little bit more into your end market. So, you know, we had Autodesk and PTC here yesterday. Those companies definitely have more concentrated end markets than Ansys, but high tech, aerospace and defense, and automotive are some of your biggest end markets. Could you just talk about all three of those? What are the key drivers that are causing customers to continue to increase their contract to spend more on simulation? Sure. So as you point out, we have a really... First, you know, we have a very diverse, diverse set of end markets. Those three that you named are, you know, they make roughly 60-70% of the total business. But we're also in, you know, consumer electronics, industrial equipment, healthcare. So we have a broad, diverse, exposure to a broad set of end markets. Now, I would say across those that you named and across those end markets, there are some common secular trends that are really driving disruption in everyone's R&D process. Things like electrification, things like autonomy, next-generation connectivity, Industrial Internet of Things, and sustainability. And all of those things drive a change in the way even simple products, you know, say, this bottle of water, and the bottle that contains this water and how it gets shipped, it is turned on its head, right? So every industry that we serve is getting disrupted. Now, when you look at those three, high tech and semiconductor, automotive, and aerospace and defense, they are all experiencing levels of opportunity, disruptive opportunity and disruptive change that are making their R&D decisions more and more complex and requiring a change. So if you take high tech and semiconductor, that is broader than just the semiconductor subset of actually making the chips. It's about making the chips and all of the handsets or the televisions or the consumer electronics around it. But if you just take the disruption on the chip side, things like moving from, you know, a two-dimensional chip design, where you're fitting all of the components on the chip, to a three-dimensional design, where you're stacking... You know, you've run out of boundary conditions to be able to make that compute efficiently. Now, you have things like 3D ICs, where you're stacking layers. Well, your typical validation problems in that old paradigm of design was really around the optimization of things like power and power integrity and those types of things. Well, now, if you're stacking things, you're introducing other problems. You're introducing things like electromagnetic frequency, which is where HFSS product plays. Airflow, because fluids and airflow are more of an introduction. So now what you're seeing is those same customers, they're solving more challenging problems with different architectures. They need to use more of our portfolio to do that. Similarly in aerospace and defense, and in autonomy, the disruption of things, trends like sustainability, needing to use more sustainable ways to move those transportation mechanisms. Electrification, you know, having a more computer-centric experience around how those things work, that is changing from a paradigm of a typical industrial, "I've built 100 types—I've built 100 models of airplanes. I've built thousands of models of cars. I'm gonna do the difference and do that Ansys." Now, you have an entirely new design, sets of designs, that have never been validated in the past, that are much more complex, that drive more demand. So that intersection of multiphysics in the component level, all the way to the system level, is more complex. And then the last piece is the mission level, right? So this is where things like Space 2.0 are highly disruptive. And something like space, it's whether it's commercial space exploration or commercial space travel or space exploration, like the James Webb Space Telescope, I mean, you can't do physical tests in space. Like, we're not gonna all, like, say, "Hey, let's see if it works," and go up on the next flight, right? This is Space 2.0 is an area where all of the, like, a significant amount of validation has to be done in simulation because you don't have the option for physical testing. And so, those are just some examples of the disruptive forces, and all of those things lend itself to the business model of, you know, you have more diverse sets of users solving those problems. You're asking different problems, which introduce more complexity in more areas of physics, and the interaction of more complex problems and more physics just drives more computation. Got it. Got it. And in terms of the, you know, you talked about all these areas of increasing complexity that's driving a lot of momentum. I guess as we translate that to the go-to-market, one of the things that I feel like we've heard more and more each quarter is just the size of some of these deals, these- Yeah ... these mega deals, these eight-figure type transactions. Can you just talk about, maybe from a go-to-market perspective, what you're doing to drive those, and do you think that's sustainable just given the underlying, you know, secular drivers that you mentioned? Sure. Yeah, so maybe I'll start with describing kind of what is our land and expand model, and then how have we evolved it over time- Mm-hmm ... and then how does it connect to it? So, starting with the first piece, so for some of you who may not be as familiar with Ansys, you know, a typical enterprise software model, let's say, I decide I'm going to replace our financial system in Ansys, right? So if, if we were to do that, that transaction for whoever that vendor would be, would be big land, right? The whole process gets instantiated in that system, all of the users move to that system, big land. And then the expand model is, I add to my finance team, I buy another module, little expand. Big land, little expand. Our model is different. What we are a tool that is part of an R&D process, that the land is... I, the land very commonly starts with, I'm a single analyst that's part of a team, that is building a project, that is part of a program, that is part of an R&D initiative, that has a specific question that I'm trying to answer, right? Small land. I'm gonna buy the mechanical suite to go figure out what the structural integrity of the plastic in this bottle needs to be to make sure it doesn't crush when it ships. But if you wanna understand, the broader but, but over time, now you're introducing the concept of, well, if it's an empty bottle, it is a different property than if there's fluids in the bottle and the shape of, and. Mm-hmm ... and now I need to look at vibration, I look at these other areas. You're asking more and more questions, which add to more and more complexity, that's more products, right? That's, that's more expand around it. But you're only asking the product in the complex- in the context of that one R&D thing. Now, I've got, you know, 700 products that I build, and I'm asking similar questions. Now, that's, you know, even more users asking more questions about more products, and now those products are more complex because they involve different types of questions that you're asking, right? So, the nature of expand is, in our business, is small land, big, big expand opportunity. Mm-hmm. So when you connect that back to our go-to-market motion. So our go-to-market motion before Ajei came in in 2017 was more along the lines of a departmental sale. I sit down with an analyst, I ask him which problems he's trying to solve, I sell him some software, I visit him next month, he's asking more questions. It was more, it's more departmental, transactional in nature. The motion that's changed over the past five years has really been moving from that departmental transactional sale to a strategic selling model, where we have accelerated the process of really understanding not just that analyst's problems, but all of the questions that are being asked, and what are the problems that are being solved, and what is the role of simulation to be able to reduce the cycle time of R&D, increase the amount of product complexity, while reducing the amount of failure and cost and time, while increasing quality. That strategic selling motion is really what has accelerated the rate of expand, with primarily the tip of the spear, the largest customers who have the most mature models. We've exported that motion down to the rest of the thousands of customers we have. When you ask about headroom, we feel really good about the headroom that is there. Mm-hmm. For our largest customers, it's because their complexity is growing faster than their ability to deploy simulation to solve those complex challenges. For the smaller customers, it's because we have not answered all of the questions for them yet. And so it really, it comes from both angles. Yeah. No, that's great framing, and I guess as we think about that strategic vision and again, evolving from just the departmental sale to that enterprise-wide sale, like, how far through your larger customers do you feel like you are with that approach? Is it... You know, obviously, Ajei's been at the company for a number of years now. Is this still early innings, or how should we just think about, you know, how far through we are on that? Yeah. I mean, I would say we have, we have a lot of headroom. Yeah. I mean, we have thousands and thousands of customers. Right. We have, you know, dozens to hundreds of those, those accounts, right? And so I, I'd say that there's, there's a lot of, there's, there's definitely headroom in that regard, and, and there's, there's a lot of headroom in just even those larger customers, that there, there's, we're not at the endpoint with those larger customers, right? Sure. 'Cause their problems are harder to solve and more com- Mm-hmm ... more complete. And so where we're focused with them is, all right, well, where are your constraints? Very often, the constraint is compute, right? Yeah. So compute capacity. Once your problem gets so big, you know, many larger customers have HPC environments, where you kind of burst off of your desktop to solve your problem. You could do it, but you have to go, like, take a trip to, you know, Hawaii and come back in two months, and the simulation will be done. Mm-hmm. Or you can burst to an HPC environment, and it'll solve it faster, right? Now, there's limitations to that, that on an on-premise way, and as the cost of compute has come down, in public cloud providers, the more and more customers are looking towards a more diversified strategy around where that HPC is, right? Mm-hmm. So, so that's an example where, okay, when you run into, when you run into constraints with a customer, like, why are we not using more? Well, it's because we, we have this really giant fluid, sets of fluid and aerodynamics problems we're trying to solve. We run against compute constraints. That's where our innovation on HPC and using GPUs versus CPUs to be able to break that barrier, or the innovation in the cloud marketplace offerings around being able to port your licenses to public cloud to be able to unconstrain compute, right? Yeah. So, you're solving different problems and unlocking different things for those customers, but there are many, still many, many, many, many places to go with that. Got it. Just to follow up on that, the HPC point, 'cause that kind of came up in our conversation with PTC yesterday, just as they're thinking about moving their core Creo base to the cloud. What, how are you seeing customers approach that? I think you said a hybrid strategy. Are they leveraging a cloud provider, or are they doing it on-premise, or how do you kind of think about how that changes, to your point, given all the, you know, intensity around GPUs that we're seeing? Yeah, I mean, I'd say that. So there's two aspects. So the GPU use case is still evolving and narrow. And while the price performance is quite good, and we've actually made some breakthroughs in certain areas of our products where we've been able to reduce cycle time of solves dramatically through the use of that technology, it's still a relatively large upfront investment, particularly in the on-prem environment, to do it. So the access to GPUs in the cloud are a good experimentation place for things like that. I would say, as it relates to customers' HPC environments, it varies quite a bit, right? So if you are a more of a mid-market customer who has more episodic needs for HPC, the cloud is a great solution because you don't have to make that capital commitment, you don't have to maintain that environment, you can burst into the cloud. It's an OpEx thing- Mm ... that you just kind of move in and out of. And so that's why the marketplace helps, you know, our marketplace offerings, like Ansys Gateway, allows you to be able to more seamlessly optimize that experience and port your licenses. A larger customer will tend, who has more consistency and high levels of utilization around HPC, will still maintain large on-premise environments because it is, you know, there's efficiencies of scale that you can get if you have, if you have a very predictable HPC workload, which sounds like an oxymoron. Mm. But you can, but if you have really, really long queues of things to- Mm. of things to run Right ... you can have predictable high levels of utilization. If you have small or batchy or, you know, not very predictable queues, then the cloud might be. And the reality is all, like, even large customers have both of those. Mm. So there's this coexistence that you're seeing of, of the two models- Got it ... that is happening. Okay, that, that's helpful framing. So as we think about your smaller customers, could you just talk about bringing simulation, you know, you've talked about pervasive simulation for a while, but kind of- Yeah ... extending it beyond validation to earlier in the design cycle. Just update us on kind of the progress there and your pervasive simulation strategy. Sure. So I think what you're referring to, and if PTC was here, they'd probably talk to you a little bit about this too, and kind of the work that we do together. So the idea behind this is really, you know, typical. Just to give some context to the industrial R&D process, so industrial R&D processes tend to be quite linear in nature, right? You're designing- you have a designer who designs the product, you design, design, design, design, time's up. Now you take that whole design, and you got to go validate it. And then, because it's a complex product, you validate that product, and then you send it back to the designers, and they've got to redesign, right? It's a very, very V-shaped process, right? You, you design it, and then you got to go redesign it, and then you go through those Cs. So, the notion of embedding simulation earlier or later, upstream and downstream to that traditional validation process is aimed at helping customers be able to validate elements of the design that reduce the cycle time of that, right? So that's the goal of that. And our approach has been, this is - and this is very long-term in nature, right? Because industrial R&D processes are not like, say, building software, where you just train 1,000 engineers on how to use agile techniques versus waterfall techniques - Mm ... and you had to deal with that change management. That's hard, by the way. But you have entire supply chains and ecosystems around industrial companies. You have sunk cost in manufacturing, which assumes that you're going to build products and ship products in certain ways, right? And so that is a lot harder to change. And so it's gonna be more measured in nature, but the manner in which you attempt to change that is to make- is to provide the simulation capability in context of that other user's experience, right? So we've done this in two ways. One is through partnerships. So partnerships like PTC, Autodesk, Rockwell Automation, are all very different in the use cases that they address, but the common aspect to what all of those partnerships do is they take the power of simulation, and they embed it into the context of the workflow that those companies manage and the end user's experience, right? So in Ansys Creo Simulation Live, it's about embedding simulation into the element of design phase, so the engineer has the ability to control the ranges of variation that the designer designs so that the validation process is faster, and they're gonna design something that works, right? Mm. So that's one example of techniques that we've used to propagate that, you know, kind of workload by workload and one at a time. The other is through the portfolio, where there's kind of two examples I'll give. This isn't the only examples, but one is Discovery Live, where, you know, we have taken a product approach to say, "All right, what is the paradigm of those users?" So today, the paradigm that our core customer has in when they use simulation is the question they're saying, the question they go into it is: I'm a mechanical engineer, and I'm going to evaluate structural integrity and solve that problem. So you go into our mechanical suite, and you're all about answering questions around structural integrity, right? But if you're a designer, you're designing this water bottle, there's a bunch of aspects of it. It's structural, there's fluids, there's a bunch of aspects to what you're designing. And so your paradigm is, I'm designing this bottle, or I'm designing a box, or I'm designing, pick your, you know, an airplane. So the Discovery Live product is aimed at starting in the paradigm of the designer and then invoking the simulation capability in context of the questions they're asking. So I'm building an airplane, I wanna know what the fluid flow of the aerodynamics and the fluid flow of that. I wanna know what the turbulent force of the engine is. I wanna know what materials are going to be the most effective. So that's one technique to propagate that. The second is really around, in order to make that really happen, you, you don't just need to affect the designer and the engineer, you need that whole ecosystem to work together. There's a lot of complexity that gets introduced into products that only engineers have the ability to understand, and that's where things like model-based engineering and the MBSE advancements that we've made, those techniques allow you to be able to take complex simulation use cases that tend to be interactive in nature at, say, a system level, and create reduced order models, which are understandable to different engineers in the process or even potentially designers in the process, so that you can do some of that higher level of validation before you get to the complex end case. Those are just- Mm-hmm. A couple of examples of what we've seen. And we see, and we see good progress, and we see good uptake on it, but again, going back to the kind of description of rate and pace of change, rate and pace of change in these R&D cycles is more evolutionary than revolutionary, although it's definitely happening. Yeah. Great. Well, it's been a great discussion so far, but we haven't talked about generative AI. Yeah. So, I figure we should probably- What's a conversation without AI? I know. Exactly. So obviously, you know, this is—we're at a tech conference, we're talking about Generative AI- Yeah ... in every session. But I guess let's talk specifically about the strategy for Ansys. You know, we talked a little bit about how the GPUs are used to kinda accelerate some of the simulation, but how do you see the Ansys simulation portfolio incorporating AI? Is this something that can open up the addressable market? Is it something you're gonna be able to incrementally monetize, you know, have an AI SKU at maybe a price point? Just talk through how you're thinking about that. Sure. So, there's three kind of dimensions to kind of breaking down the impact of things like Generative AI on the business. So, one is, of course, how we run our business ourselves, and that's not gonna be-- that's not too dissimilar than every company who's asking themselves this question: If you have capabilities like ChatGPT, which accelerate your ability to produce and optimize content creation, where can you do it? Yeah. So things like sales enablement, we talked about AnsysGPT, which kind of takes all of the brains of the PhDs in the business and puts it into one mega, mega brain, through AnsysGPT. So there's those types of techniques which we will do through the normal course with the right eye towards security, IP protection- Mm-hmm ... all of those things. The second one is, how can Generative AI change our products? Now, the Generative AI is really mostly suited towards answering questions of optimization, right? Mm-hmm. So it's less about invention in its pure sense, but more about, like, the optimization problem. And so there's certainly corners of the portfolio which these techniques have been around for a very long time, so this is not new to us, and that are already components of the portfolio. But in the areas of the simulation process where it is fundamentally an optimization problem, these techniques can help that user experience, right? So, one example we've given in the past is the question of turbulent flow is actually a very complex set of variables that is very difficult to constrain, that can create a lot of complexity in solving that problem. So through using some of these techniques, you can kind of reduce the variability of the amount of optimizations you have to think about and accelerate the time of that process. The question of, you know, we have running a mechanical simulation, how much compute is it going to generate? That's an example of, like, that's an optimization problem- Mm-hmm ... depending on how many inputs you have, you know. So those are examples of ways in which these techniques affect the product. They're less so. It is more difficult because the actual solvers themselves and the software themselves are not really optimization problems. It is really the process of running many, many, you know, partial differential equations to actually approximate how the physical world behaves. That's not as amenable to some of those things, so it's a little bit harder to affect that part. The last piece that Generative AI in terms of impacts that the business can have, really comes down to: How do customers change the way they build products? So the most obvious place I think that many people ask about and many people talk about, whether it's with us or others, is: Well, will this change design? Does this change how design works, whether you're, you know, a creative building content or whether you're an engineer building a product? And of course, if this creates more designs or more mass customization of designs, it means more simulation. Mm-hmm. So, people, the more designs you have, the more designs you have to validate, the more simulation. The other dimension of how it could impact a business and over the longer term is, it is likely that if a human being who has generated prior designs is not the author of those future designs, that the nature of those designs are going to be fundamentally different than anything else that's built before. So the zero basing of validation means it's a more complex thing. You have to have more variables that you're testing to validate the design, so that's more computation. So you've got, you know, more users, more products, and more computation, all in that. Now, the challenge is, what are the tipping points and what is, how is that really going to happen? We're quite confident that this is a tailwind to the business overall, over the longer term. The more difficult aspect to predict is what is the shape and curve of that- Mm-hmm. How does that go over time? Yeah. Now, that, that's a great way of framing it in kind of those, those three buckets. It, I guess on the last point, just in terms of your, what you're seeing already from your customers, and I often get the question about, you know, the semiconductor end market. Clearly, massive shortage of GPUs. Yeah. I know Ansys isn't, like, tied to semiconductor output, which makes the business very- Yeah ... very steady, but it does feel like there is kind of this race in the semiconductor industry to, to design the next generation of high-performance computing chips, 'cause obviously, there's a huge shortage and- Yeah ... and a lot of budgets for that. So I guess, have you seen signs of this kind of being an accelerant, this gen AI wave kind of being an accelerant for increased R&D budgets, specifically within semiconductor simulation, or- Yeah ... it's still kinda too early on that side? I mean, I think it's a little too early, but I would say that your kind of thesis and framework of how to think about it is right, right? Mm-hmm. So, so the companies that are more closely connected in a first order sense to these trends, like, like, you know, some of these other companies, NVIDIA or whoever, are obviously going to, going to. Are, are moving the quickest and have, have, you know, are the closest to the, the, the changes in the design, the design, dynamics around their business that are, that, that could accelerate this trend. Then, but there's. But it's really the, as we said at the beginning, the, the diversity of our product set, I'm sorry, of our, of our industry coverage is, is really across all industries. Yeah ... that you could imagine, and the rate and pace of how generative AI affects those customers is still a little bit more, more measured to see. Mm-hmm. Right? Yeah. And so it's, there's no doubt that this is going to change the way design and content and things happen over the long term. Yeah. Right? It's, it's not really an if, it's a when, and it's a what does the shape look like? Yeah. Yeah. Great. Well, that, that's a great perspective. I wanted to... You mentioned this a little bit earlier, but just on the subject of cloud adoption, we often get the question from investors, you know, if you look at PTC or Autodesk, they've made these subscription transitions. Yep. PTC's talking about the SaaS transition. How are you thinking about that at, at Ansys, especially given what continues to be declining cloud computing costs? Sounds like customers are at least thinking about running some of these simulations in, in a hybrid fashion. What, what does that eventual transition and, and shift to cloud look like, for Analsiys? Yeah. Well, I'd say that there's, there's two aspects to our cloud strategy, right? There's, there's what we call cloud marketplace and cloud native. I talked about cloud marketplace before. Mm-hmm. But before going into that, I think it's important to kind of contextualize the nature of our workload and how it relates to that cloud shift versus a typical enterprise software, right? So I'll go back to my financial system update example, because I guess that's the one stuck in my head today. So if we move from an on-premise environment in our financial system to a cloud environment, the nature of that workload is quite straightforward, right? The whole process gets instantiated in that application- Mm-hmm ... and the nature of the interaction in that application is really a shared data and change and collaboration, which is a database lookup and a database update. So the nature of the data transfer, the compute intensity, and the pattern of that is relatively low and relatively predictable. So the performance I get as an end user is probably, in some cases, depending on how you invest in your on-premise infrastructure, is probably better in the cloud, more portable, there's more benefits to that, from a user experience standpoint than on-prem. That's why with compute costs going down- Mm-hmm ... you can see the acceleration. Now, our workload is quite different. It is a scientific workload, and really, what the nature of how you use it is you use the desktop application to build a model to determine how you are going to simulate the problem, and then you have to run the solution to the problem and interpret the results. Now, the act of building the model, in and of itself, can be very compute-intensive, right? If you're modeling turbulent flow across, you know, a gas turbine, like, that's a very big problem with a very big model to solve. That can use many cores of compute, and, you know, it's not uncommon to have terabytes of data. The solve itself can be even bigger, right? And so if you think about that end user experience, it's a different workload, and you can be more compromised in those examples in the end user experience. So that's why it's different. So that's why the customer use cases are what inform our cloud strategy. Mm-hmm. So the customer use cases are really around that compute constraint around the marketplace offering, right? So I have my... I need to burst to more compute to solve this bigger problem. Our marketplace strategy allows you to be able to port your license and optimize your use of simulation in that compute environment in a seamless way. The second one is cloud native, and that's the one that when we spoke about Investor Day, and then we speak about now, that is more of a long-term Horizon Three use case, and that is really about addressing new users and new use cases that tend to be more narrow in the in the problem set you're trying to solve, right? So how it shows up in the more recent part of the business is, you know, in order to enable this, we have been on a journey to componentize our portfolio, to expose elements of our portfolio with standard APIs and make that accessible on a platform that we call PyAnsys, which really allows you to be able to use standard Python programming languages with which anyone under the age of 20 can actually do. I'm not under the age of 20, so I don't fall into that category. And you can write a simple application through the Python language, and invoke the element of Ansys, whether it's a solver or whatever, as a component of that, right? And so those are examples where you're really starting going back to the design example, right? You're trying to answer a question, and you're using the capability of the solver to invoke it. So how it shows up today is, I may be building a product which is largely a structures problem, but I have a very specific question about electromagnetic interference of that structure. So a cell phone charger that's sitting on here and the electromagnetic interference. The use case today in the traditional paradigm is I'm in one product to answer one question, I'm in the other product to answer the other question. The opportunity that PyAnsys gives you is to say: Well, I'm really answering a structures problem, but I really have this very specific thing in the workflow of answering that problem I want to answer. So you could build a very specific, workflow, a very specific set of workflow to answer that specific question in context of it and not change the user experience, right? So that's the more common near-term use case. Over the long term, it's things like, the optimized example that we gave, that we gave in our investor day, right? So this is an extreme example where, Optimeyes is a company that builds, software that is, assistive technology for cataract surgeons, right? So the end user is a surgeon. They use the software by, you know, they map the eye, and then they look at the map of the eye, and they're planning their surgery in the software. And as they're planning the surgery, the input of the surgeon is: I'm cutting here, I'm replacing this lens, I'm putting this one on. But what's happening in the background on the validation side is it is invoking the Ansys structural solver to say, what is the outcome that that's gonna create? Is that gonna create the best structural outcome and integrity in the eye? There's no expertise. That doctor is just a very smart person, not a PhD in mechanical engineering- Right. But they're using the power of the solver. Mm. So that is an example of the very long-term outcomes, but it really starts with these near-term use cases and kind of the activity that has gone around PyAnsys and the experimentation that is happening today, which has gotten great upticks. I mean, the libraries that these have been published in have had tremendous download activity. They've had tremendous amounts of engagement. They've had great positive feedback. But it's still really early days- Yeah. And it's a more long term. Yeah. Awesome. Well, I think we are out of time. It's been great discussion. Okay. Nicole, thanks for joining us, and thanks everyone for listening to the chat. Thank you so much.