Excellent! Well, thank you, everybody. We're about to get started. Welcome to the F5 Fireside Chat at Goldman Sachs Communacopia and Technology Conference. I have the privilege of introducing Frank Pelzer, CFO, Kara Sprague, and Chief Product Officer. Frank has been at F5 since 2018, joining from SAP, where he served as President and COO. Prior to that, Frank was CFO of Concur Technologies before it was acquired by SAP in 2014. As Chief Product Officer, Kara has responsibility for the company's entire portfolio of multi-cloud app security and delivery solutions. She joined F5 in 2017 to run the ADC business from McKinsey, where she held various leadership positions across their technology practice. My name is Michael Ng, and I cover F5 and CommTech here at Goldman Sachs. We have about 35 minutes today, inclusive of Q&A, so if you have any questions at any time during the session, just raise your hand and we'll get a mic runner over to you. So first, Kara, Frank, thank you so much for being available for us today. It's a privilege to be able to host you guys on stage. Thanks for having us. Excellent. So over the last several years, F5 has expanded beyond ADC into a portfolio of security and application security solutions. And security is about a third or a little bit over a third of revenue today. Could you just start by framing F5's history and leadership in ADCs, and what makes that a natural extension into app security, and the initiatives that F5 is pursuing to do more in app security? How much of this is an evolution driven by changing customer needs, or something else? Sure. And before I respond, I wanna get our safe harbor on record. So please note that our discussion today is going to contain, may contain forward-looking statements which involve uncertainties and risks. Our actual results may differ materially from those expressed or implied by these statements, and please see our SEC filings for more information on those risk factors. So thank you for that. So to your question, F5 rose to leadership in ADCs in the early 2000s, and that was a result of the enormous, performance, the capability, and the programmability that we built into our BIG-IP product family, which is F5's, historical product family, and was F5's only product family until we acquired NGINX in 2019. People tend to think of these ADCs as very focused on traffic management and more load balancing. But because of the role that an ADC plays in the line of traffic for a customer's application, it's also a really great place to put security capabilities. So think about entry into a bar. The place you put the security is at the door, right? That's where you put your bouncers. And so a similar sort, similar concept here in terms of ADCs, is you can tie security into the proxy. So we began adding this application security functionality, and we put that in more than a decade ago. This enabled customers to then start consolidating multiple functions. So instead of having multiple vendors and tools along the application data path, they were able to do their load balancing and their web app firewall and some of their DNS and some of their DDoS all in one box, and that was the F5 BIG-IP ADC. So, in the 2010s, as cloud started picking up speed and it became much more apparent that the future of the world was going to be more distributed, and that application components, because of the rise of containers and microservices, were gonna become a lot more fragmented, and hosted in a bunch of different places, we understood that that was gonna create a lot of important and significant challenges for customers in how they think about their applications. We committed ourselves to being able to secure and deliver and optimize any application, any API, no matter where those applications and APIs were hosted. So, with these distributed applications, I would also add that the threat surface area is dramatically expanded and different. No longer do you have basically this kind of walled area where all of your important applications are sitting, but now your applications are running in all of these different environments, spanning on-prem, multiple public clouds, and edge environments. The components themselves are disaggregated into smaller pieces, and so that naturally means that your threat surface area is much, much bigger. Given today that we sit in a data path, that application data path of about 40% of internet traffic, and we handle a large portion of that traffic daily in terms of 100+ exabytes, we think that we are uniquely positioned to offer increasing value propositions to customers as they continue to optimize the security and delivery of their applications. Great. Just sticking on the topic of ADCs for a moment, I was wondering if you could talk a little bit about, you know, what the typical refresh cycle might look like for ADC appliances? How often do they need to get replaced? And, you know, what are you bringing to market with things like the rSeries, that's helping to drive refresh upgrades- Mm-hmm ... and new features? Historically, F5 has released a new generation of our, our hardware appliances every five-ish years. Now, that might vary for different factors, and it's not a hard and fast rule. Prior to the most recent generation of our rSeries, the last appliance that was released in the 2016 and 2017 period. We're at about a five and a half to six-year cycle right now, and a lot has changed in the market and in our business. What we've seen over the last several quarters, there's been customer caution around the macroeconomic environment that has resulted in a lot of customers, what we call sweating their assets. They basically run their existing appliances beyond what they would otherwise operationally be comfortable with doing. And they also drive. They postpone their refresh, and they do that by driving higher renewals. As a result, what we end up seeing is that for our maintenance renewal business, especially for appliances that are four years and older, that has actually picked up and driving an increase in our services revenue. Whereas our product bookings business has seen declines because of that macroeconomic factor. Now, in our rSeries and VELOS, as you mentioned, that's our next generation systems, the ones released most recently, and we're still in the middle of that refresh cycle. They bring really strong price performance value to customers. So typically, when customers go out and they take the next generation of a physical system, they're expecting, you know, price, performance, impact. That's a very standard expectation, and vendors have to deliver on that. However, on top of that, and what customers and the market has not yet fully appreciated, is the amount of innovation that we have packed into these systems. And if you want me to go down a rabbit hole, I can talk about all of the cool things in here. But it includes elements of automation that make them much easier to operate and maintain. There's also elements that make it much easier for customers to upgrade, and migrate between different versions of BIG-IP and eventually other types of tenants. I'm also really excited to say that as a result of all of these dynamics and what we're seeing with the hardware business, the hardware business remains a strong part of F5, but we've been very successful in also expanding into software, and our software footprint as of last year is now more than 50% of our business. Great. Yeah. And that's a good segue. I do want to ask about the evolution of, you know, F5's portfolio of solutions, for both on-prem and, you know, cloud-based solutions like, BIG-IP and also NGINX. You know, how have you seen, workloads and customers go between, you know, NGINX and BIG-IP, or use both solutions within- Mm-hmm ... you know, their data centers? And maybe you can just talk a little bit also about the interoperability between BIG-IP and NGINX. So BIG-IP and NGINX serve different kinds of applications. BIG-IP, it was something that was architected and built in the early 2000s, and the kinds of applications that it has been serving tend to be applications that were built and architected from that time up until, call it, the mid-2015, mid-2010s. And those are all what we would describe as traditional architectures for applications. So monolithic applications, client-server applications, or client three-tier applications. Once you know, once containers evolved past about 2014, container-native applications and microservices are now the de facto development model for new application development. And so there's a whole bunch of new applications out in the market that are these container-native and microservices-based ones, and that's where NGINX fits. The reason why BIG-IP doesn't work there is literally BIG-IP does not fit in a container-native environment. BIG-IP has a footprint running of about 1 GB, sometimes bigger, and NGINX, on the other hand, can run on the footprint of less than a floppy disk. That's about one one-thousandth the size of a BIG-IP footprint. It's important to recognize that they're fundamentally different technologies that serve different application needs and have different footprints. An analogy you can think about, it's kind of the difference of what would you use an eighteen-wheeler for versus what would you use a sports car for. That should give you the intuition of how the two are different. So as I talked about it, just to summarize then, BIG-IP is for those traditionally architected applications that require a lot of protocol sophistication, a lot of programmability, very robust set of services. NGINX is lightweight, highly scalable, highly performant, with more straightforward, or less sophistication in its protocol coverage. And what we're seeing in the portfolio, because we get a lot of questions from analysts and from customers, and other stakeholders of how we are thinking about applications and migrating between these things? The reality is, we don't see a lot of applications migrating between our product families, because when you talk to customers, they are building a lot of new applications, but their work is less focused on the modernization of old applications. What we fully expect, if you look out, you know, five years from now, we expect the majority of our customers will be customers of many of our product families, including BIG-IP, including NGINX, and including Distributed Cloud, which is our SaaS and managed services capability. Great. I think that makes a lot of sense. And you know, you touched on this earlier in your comments about how, you know, F5 is in a natural place to address application security. Mm-hmm. I was just wondering if you could expand a little bit on that. How would you define what F5's app security portfolio looks like today, and how, you know, those security products and those delivery products really work together to create, you know, comprehensive solutions for your customers? Sure. So I talked about our roots as fundamentally an application delivery company. You know, we started out in the late nineties as a load balancer, you know, helping dot-com companies figure out how to not take down their servers when they got a lot of demand. So that was our roots. But I also talked about why once you get that critical space in the line of a customer's application traffic, it's also a great place to start putting on security. So by the time I joined F5 in 2017, our sales team was already adamant that the vast majority of their sales conversations were security-led. And so we have that distinctive value proposition in application security for a number of reasons. One is because of that control point in the line of application traffic, but also because we tend to be infrastructure agnostic. So what that means is that customers can deploy consistent security policy and consistent mechanisms in their security, regardless of where their application logic is hosted. That is not the case if they go with a public cloud native solution. And that's not the case if they stick with one of their on-prem security solutions. And so the fact that F5 capabilities, whether it's BIG-IP, NGINX, or Distributed Cloud, are consistent across their on-prem and public cloud deployments and edge deployments, is a real advantage and differentiation for the portfolio. And then at the same time, you know, I talked about our different deployment options. We are also offering customers a range of consumption options, too. And so that's another advantage that customers see in consuming from us, is that, you know, regardless of if they want a perpetual, purpose-built hardware thing, because that's what they're used to operating in their on-prem data center, or they want a consumption-based SaaS capability, they can still get that policy consistency across them. And so when you start looking at the reality that customers have today, which is they're no longer dealing with an application footprint that is limited to their data center, but in many cases, we're seeing a large number of customers... Our estimate from our latest research is 85% of customers have what I call a complex application portfolio that has both legacy and new components, as well as spans on-prem, public cloud, and edge, 85% of customers. Those customers are looking for a partner that can provide a thin layer of abstraction for certain services like their application security, and that's where we come in. Great. And as you think about the security portfolio and all the opportunities that you have in application security, do you see any gaps in the portfolio and/or other adjacencies that you might be able to expand to, that, you know, could make the whole portfolio of solutions more compelling? In terms of gaps, no. I'll say that because, you know... I'll just talk about what we have today. Yeah. So today we have the broadest and deepest portfolio of application security offerings, and we believe that we are the only solution provider in the market that can secure every app, every API, anywhere. I know that sounds like a marketing tagline, but I am saying we are the only solution provider that can secure every app, every API, anywhere, and that's a tremendous value proposition to customers. Just ask anybody to start talking about how they responded to Log4j. Second thing that our portfolio today has is we have a leading solution. We were literally just named as the leading solution by SC Magazine for our Web App and API Protection offering, which is a SaaS-based offering that includes a Web App Firewall, API security, anti-bot solution, and DDoS. We also have numerous industry analysts that make us a leading solution in our web app firewall solution on BIG-IP. We have a similar web app firewall capability on NGINX. And then when you start looking outside of the web app and API protection space, we also have a portfolio that contains rich solutions for anti-bot, Layer 4 and 7 DDoS. We have Identity and Access Management. We have traffic break and inspect, and so the portfolio is quite extensive. And that portfolio that I just described, which we have today, already competes in markets of tens of billions of dollars. So we're feeling very good about how we're positioned, and we're feeling very good about the chess pieces we have on the board. Now, we do also frequently and very regularly evaluate our market landscape changes and also look at, you know, what compelling adjacencies our customers are asking us about, and so that is a part of our ongoing evaluation process. But at the moment, there are no burning gaps. Great. You know, tying it all together, could you just talk about how this all rolls up into, you know, the Distributed Cloud Services platform to bring a unified offering to your customer? You know, what progress have you made in onboarding customers and adding potential new features to that platform? Yeah. So Distributed Cloud is really, really exciting. This is a offering that we launched about 18 months ago, and so it's very- Yeah ... you know, it's relatively new in the market, but we're seeing incredible, momentum and excitement, in our customer conversations. We primarily serve, with Distributed Cloud, two use cases. One of them is the Web App and API Protection use case, or WAAP, which includes, API security, which if you talk to any customers, if you, you know, frequent the halls of RSA or Black Hat, API security is the pain point that customers will talk about if you ask them about application security today. And so we have a very, very strong offering in that space. Also, on Distributed Cloud, the second major use case we service is something that we call secure multi-cloud networking. So this solution addresses customers that do have those infrastructures that span on-prem, multiple public clouds, as well as edge, and what they're looking for is an abstraction layer and simplification in how they operate across them. So there's a lot of complexity if you try to do this yourself in terms of the Layer 3, 4 through 7 networking across these different, these different environments, and that creates a lot of opportunity for error and also a lot of security holes. But the fact that we have a capability that offers multi-cloud networking at the Layer 4 level, and it goes all the way up to securing that multi-cloud network, because we cover Layer 7 as well, is a differentiating point for us. And we also offer that capability as a SaaS-based model, as opposed to most of the other solutions today are packaged. So those are the two big use cases on Distributed Cloud. In addition to those two primary use cases, we also have some supplemental capabilities and services. So we've built out a DNS offering on it. We have a CDN capability on it that has been supplemented by a January acquisition of a company called Lilac Cloud. And we are continuing to extend the portfolio of services. Great. I was wondering if you could talk to some of the changes in the F5 consumption model, right? You're doing more software, more subscription, that improves visibility into revenue and revenue growth. You know, how much of the business is underpinned by, you know, multi-year subscriptions today? What are the typical terms of those subscriptions, and, you know, which services kind of lend themselves to more of a subscription model? Sure, sure. So, let me take that one, Mike. When Kara came into the business, we were sort of a single product company that offered perpetual licenses, and that was it. And, what we have found is that as we've evolved our portfolio of solutions, our customers really value the flexibility that we offer them in terms of, they can either consume perpetually, they consume through a subscription, or they can, frankly, consume through a consumption or utility model. So we offer all of that flexibility to our customers. Not every product fits in each one of those buckets, but our goal over time is to normalize our product portfolio so that customers really can select from that basket of choices. That lends itself then to what we have as our Flexible Consumption Program, as a master agreement, if you will, for how customers may want to consume those solutions. That's typically a three-year term-based model. There are some of our solutions that are offered only through a SaaS solution, and that's gonna be a ratable revenue recognition. Now, that flexibility does come with complexity, particularly, in the way we model the business, the way investors have to model the business. But we're willing to make that trade because our customers truly, truly value that. What we've seen in terms of visibility, though, over that period of time, you know, again, flashback to 2017, 2018, where 90%+ of our revenue was through perpetual hardware. Now, today, with, you know, over 50% of our product revenue coming from software, and in the last quarter, 87% of that was through subscription, we do get better and better visibility into what that's gonna mean. With our ratable business, it's obviously very easy to model. For our term-based business, every three years or so, we get a new revenue recognition under 606, and so you will see some fluctuations in quarter-to-quarter, what that revenue is going to look like. But as we get to the second term or the third term or the X N term, you know, beyond, more and more of that software revenue is gonna come through a renewal cycle as opposed to a selling a new business cycle, which will give us a lot more visibility, in the out years. Great. That's really helpful. You know, one thing that's obviously been top of mind for the market has been trends in enterprise spending. You know, as enterprises more closely scrutinize budgets, F5 has noted that some customers have delayed or canceled new projects. Could you just update us on the latest that you're seeing here and when you potentially expect these trends to start to reverse? And then, on a related topic, you know, we talked about refreshes. Like, how long can customers actually delay refreshes- Mm-hmm. on things like ADC and security before performance starts to get impacted? Sure. So, I'm gonna take that question as of what we talked about in July. I'm not gonna update anything- Sure ... you know, inter-quarter. But, what we talked about in July is that, we saw a stabilization of demand. Now, at a much lower level than what we would have expected going into the year, but it was an improvement from the declines that we had seen in, in Q1 and Q2 of our fiscal year, and so, that was a good, that was good from our, our perspective. If we take a look at the dynamics, I think they were similar between hardware and software, but for different reasons. On the software side, we noted at the beginning of this fiscal year that, you know, our software growth rate depended on a little more than 50% coming from new business activity. That isn't necessarily new logos, but, you know, customers expanding their footprint with us or new customers coming to us. And then a little less than half of that was gonna come from the renewals. On the renewals side, we've actually seen that perform quite as we expected. On the new business side, particularly on the transformational deals that can be rather large, that's where we saw a tremendous amount of scrutiny, particularly in Q1 and Q2. A little bit of that got freed up in Q3, but, you know, again, at a lower level than what we would expected coming into the year. On the system side, I think, you know, in hindsight, we're probably, you know, what we thought was just pure natural demand in COVID in 2021 and 2022 could have been some pre-buying into 2023, and you're seeing customers work through the consumption of that, especially as supply chains have released, and we are able to actually work down the backlog that we built in FY 2021 and 2022. And so, that is sort of a natural demand cycle that we do expect to come back more fully, particularly in 2024 and beyond, from depressed levels of what we've seen in 2023. Great. I did wanna just follow up on that a little bit. Just given where you know, expect to exit the year. Yeah. Could you just expand a little bit more on, you know, the setup going into fiscal 2024 for- Sure. - System software services? Yeah. Yeah. Yeah. And what we talked about at the beginning of 2023 is we, you know, had a much larger backlog than we are used to running as a business. Our typical backlog is, you know, $15 million-$40 million at the end of a fiscal year. You never know, we have two-week lead cycles in good times. We never know what we need to build, and so you're always gonna have something that falls into the next quarter. But we were seeing that, you know, at $231 million, so about a quarter and a half of what we generally would see in hardware revenue in our backlog. Not as big as some vendors out there, but still sizable in relation to us, and we just didn't know what the buying patterns were going to be. As we've worked through the year, we have seen more and more of that backlog being consumed because we are able to improve our supply chain and our ability to deliver that product. But that's gonna create probably a 6%-8% total headwind for us in revenue going into FY 2024, and we've talked about that. And so that will be, you know, a quote, unquote, you know, revenue headwind from the backlog reduction that we've been able to see, but for all good reasons. Right. While certainly appreciating the comps from the backlog recognition in 2023, you know, F5 has said that they expect to deliver double-digit EPS growth- Yes ... in 2024. Maybe you can just reconcile that. Yeah. You know, obviously, some of it is cost management. Yeah. Yeah, most of it is cost management, Michael. So we aren't dependent on, you know, much of any revenue growth in 2024 to achieve our goal. Now, we did have to pull back a little bit from the statement of double-digit EPS growth in 2024 to pre-tax because of some of the benefits that we've seen in our tax rate from R&D tax credits and other foreign countries making determinations on tax plans that have been a benefit to us in 2023. But on a pre-tax EPS basis, we absolutely expect to have double-digit EPS growth. You know, $130 million of that was coming from just the salary reduction of a 9% workforce reduction, and we'll get the full benefit of that rate, that run rate in twenty... in 2024. We had about five and a half months in, in 2023 of that. We have also looked at some of our IT spend rationalization. So if you've got three collaboration platforms, can you do that in one? If you've got cloud spending, you know, for workloads that have been built up in development but aren't still being used, but the meter's still running, can we get more efficient looking at those type of areas? We've looked at, you know, headcount. I'm sorry, we've looked at facility reductions based off our headcount reductions and our return-to-office policies. We've taken a look at event spending. We've taken a look at T&E. So we've uncovered all of those rocks and really focused on that. We do have a little bit of a headwind because in 2023, we had an artificial reduction on all of our corporate bonus plans to make sure that that was in alignment to the revenue loss that we saw. And so in 2024, that will be a headwind going in, but we've taken all that into account when we've talked about, you know, a target of 33% operating margins for FY 2024. Great. Why don't I squeeze one more in before I see if there are any questions from the audience? But just as a follow-up- Yeah ... to that point about at least 33% EBIT margins for 2024, maybe you can just talk a little bit more about the puts and takes. Yeah. Obviously, you have the $130 million of savings, but, you know, how are you thinking about productivity, pricing, and, and product mix as well? Yeah. So, all of those factors that I mentioned, pricing is not gonna be a big dynamic actually. You know, pricing is probably a little bit of a headwind, particularly in our services business, where we had an uplift to our services revenue this year from some of the price increases that we put in in FY 2022, which took effect on the renewal cycle. But really what we are seeing are the cost reductions that we already put in place in April plus productivity improvements. We are specifically using AI in some of our service and product development areas to get more efficient, but we're looking at standardization and automation in other areas of the business. There's gonna be no stone unturned to look at for that efficiency, but all of that is sort of baked into that 33% operating margin guidance. Great. Any questions from the audience? Maybe I'll just ask about the backlog. Yeah. You mentioned the $15 million-$40 million as your typical backlog. Yeah. I think on your last earnings call, F5 also said that they expect to substantially work down the system's backlog, through the end of End of fiscal 2023. End of fiscal 2023, right? Yeah. So could you just remind us, are, are the supply chains, you know, completely behind us at this point? Will we go back to typical starting, you know, 1Q 2024? Knock on wood. But you know, I never underestimate what a macro environment can do- Yeah ... and other policy decisions. But, you know, we've got. I I think we are back to where we were pre-COVID of you're always gonna have, with 2000 components in a box, some issues somewhere. But, you know, we're not seeing the same level of dependency across all of our hardware platforms on, you know, a handful of vendors, that really, you know, were going through some serious straining and impacting our ability to ship product. And so, that's the good news. I think on the demand side, that's still what we're trying to work through, of when that will sort of change. I did fail to mention, I think, in the other question... When we've seen sort of a slowdown in demand in the past, when people sweat assets, as Kara was mentioning- Mm. You know, for us, when we look back at, you know, late 2000s, early 2010s, two cycles that we went through, it was generally a four to six quarters of slowdown before customers did snap back. Now, I think a little bit of the difference here is, the reason was the supply chain crunch. Now, as some of that product has been shipped, I'm not sure the snap back is gonna be making up for all the demand that we didn't see, but I think we will be in a better place in 2024 than we were in 2023, from some of those factors. I don't think supply chain is something that we are as concerned about now as we certainly were, you know, for the last couple of years. Great. How is F5 positioned to participate in all this investment around generative AI infrastructure? And then, you know, how are you thinking about implementing AI into your products or into your internal processes? Yeah, I think, I think we're positioned quite well. So if, if you think on an overall basis and, and look at what F5 is most, F5's revenue and business performance is most attuned to, we are fundamentally tied to the growth in applications and, and APIs, APIs being a sub- subclass of app- applications, and the traffic flowing through those. That has driven all of our business dynamics for the 25 years that we've, we've had history. And, and when you look at, the, broad adoption of, of AI technologies, not just generative AI, but broad AI technologies, that is gonna fundamentally be another accelerant on the number of apps and APIs that are out there in the world. Those apps and APIs struggle from a lot of similar security challenges that regular apps and APIs do, and so there's an existing set of opportunities for F5 to attach our existing product base into those new workloads. In addition to the fact that there are some unique threats and risks that these AI-based workloads have. So folks are probably, or maybe you they are familiar with, there's an OWASP organization that traditionally, it's a public organization that categorizes the top ten threats to different kinds of applications and workloads. So there's an OWASP Top 10 that WAF vendors or web application firewall vendors deliver against. There's another one for API security that vendors work against, and there's equally one for large language models, as well as one for machine learning. And in fact, F5 team members are leading some of the efforts to define and classify what the top ten are for those last categories. And so that is a new, a net new opportunity for products and capabilities that F5 is very well-positioned with our capabilities and expertise to serve. So that's one dimension of it. Specifically in our products, much like you see these generative AI tools being used to enhance usability of other kind of products as kind of co-pilots, a lot of opportunity in our product set for doing that as well. In addition to using those technologies to surface insights into the, you know, hundreds of exabytes of data that I said flow through our stuff already. And then you heard Frank mention AI and AI's potential in terms of transforming the productivity of several of our functions, right? So, you know, our employee base is about 6,500. We have a very large group of engineers. We have a large group of customer support personnel. And so if you think about anything from code support to case customer support, case deflection, those are great opportunities for us to implement some of these tools and technologies. Great. Why don't I see if the audience has any other questions? Great. Frank, maybe I'll kick a last question over to you. Sure. Sorry, was there a question there? No. Okay. Just on capital allocation- Yep ... you know, F5 committed to returning at least 50% of free cash flow to shareholders via buybacks. Maybe you could just expand on capital allocation priorities, how you're thinking about buy-balancing buybacks with investing in the business and more M&A. Yeah, you know, frankly, our attitude towards this has really not changed since we talked about it in November of 2020 at our last Analyst and Investor Day presentation. And so, at the time, we had stopped our share repurchase program. There was a lot of tech debt that we settled through some inorganic means. And so we were in the process of integrating those technologies. We had a sense that we were gonna be acquiring one additional company in that, and Volterra has become, you know, a critical glue, for our Distributed Cloud and the platform that we've been able to design around that. But at that point, we felt pretty good that we've got now the portfolio in place. Kara mentioned that earlier, and it was wonderful. But, we... At that stage, we said, "Okay, we're gonna go back to at least $500 million for the next two years, then 50%, at least 50% of our free cash flow going towards share repurchase." The rest of the balance of what we could do is higher than that for share repurchase, or we could use that for inorganic means, and, and that's the balance that we've kept since. So no change in that capital allocation policy going forward. Great. It's a good way to cap off the session. Kara, Frank, thank you so much for joining us on stage today. Thanks. It was a privilege to be able to host you. Thank you. Great. Thanks so much. All right. Appreciate it.