All right, thank you. Thank you. Welcome, everyone, to Cisco's 2024 Investor Day at Cisco Live. On behalf of Cisco's executive leadership team, I'd like to extend a warm welcome to today's event. I'm Sami Badri, Head of Investor Relations at Cisco, and today, you're gonna hear how Cisco is positioned to connect and protect customer networks. Before we get into the information and the content you came to watch, I just need to read off an important disclosure. I'd like to remind the audience that the matters we'll be discussing today contain projections and other forward-looking statements regarding future events or the future financial performance of Cisco.
Please refer to our most recent SEC filings, specifically Form 10-K and 10-Q, for a discussion of important risk factors that could cause actual events or results to differ materially from those in the projections or other forward-looking statements. In addition, we'll also be referencing financial measures that are non-GAAP in nature. We'll be providing a reconciliation of our GAAP to non-GAAP financial measures in the appendix of the appropriate presentation slide decks. These will be made available on the Cisco Investor Relations website within the next 24 hours. Next, for our agenda, we'll kick things off with our Chair and CEO, Chuck Robbins. He will walk you through Cisco's ability to connect and protect organizations for the AI era, how Cisco has transformed, and the opportunities ahead of us. Then, you're going to hear from Jonathan Davidson on our networking vision and strategy.
This will be followed by Jeetu Patel, who will speak on our revamped security vision. Then, you're going to hear from Gary Steele, our newly appointed President of Go-to-Market, who will provide an overview of Splunk and the opportunities Splunk brings to Cisco. We will then take a 20-minute break. Following the break, you will hear from our CFO, Scott Herren, who will walk you through how everything you've heard today boils down and drives shareholder value. Afterwards, we will transition and host a Q&A session, where you will have an opportunity to ask our executive leadership team questions. Then, we will conclude with some closing remarks. Before I turn it over to Chuck, I'd like to remind the audience that all the presentations you hear today will also be going on our Cisco Investor Relations website and will be made available within 24 hours after the event conclusion.
With that, it is my pleasure to introduce Chuck Robbins, our Chair and CEO.
Thank you, Sami. And, hello to all of you who are in the room. Thank you for being here today, and then thank you to all those who are tuning in and watching virtually. We're really excited that you're here, and for those of you in the room, hopefully you were able to attend this morning and got a sense of the pace of innovation that the teams are driving right now. So what I wanna talk to you about is how we do see the future rolling out and how we see this plan to win, how we believe we can get to growth.
But first, I think it's really important for me to reflect back on what we've accomplished since the last time we were together, which, as Scott and I were just discussing, it was September of 2021, the last time we did a Financial Day, a finance day, Analyst Day. So I wanted to recap what we committed then. What we talked about back then was 5%-7% top-line growth and 5%-7% EPS growth. We delivered an average of 4% growth during that timeframe from a top-line perspective, and we delivered at the high end of the EPS guide at 7% over that timeframe on average. We also said that we wanted to get to 50% of our revenue coming from subscriptions by 2025.
We actually hit that target a year early, and that was before the Splunk acquisition kicked in. Then from a capital return perspective, just a data point, from 2019 to today, we have returned 93% of free cash flow to our shareholders over that timeframe. So what I wanna talk about today is really about how we are absolutely laser-focused on growth. We're focused on consistent execution. I think it's fair to say that the last few years have delivered a lot of surprises to us and the entire industry.
Coming out of the pandemic, we went into the supply chain shock, and we've gone through all the various phases of that, and given that none of us have ever been through a supply chain shock like we dealt with over the last few years, we were learning as we went, and we obviously, this year, dealt with the final phase of that, what we believe is the final phase, which is the unloading of our backlog and the pressure it put on our customers to actually get all that equipment deployed. We're hopefully, as we've talked about, coming out of that as we exit this fiscal year, which I think will then put us in position for much more consistent execution.
And while the 50% plus of our revenues coming from subscription hadn't felt like it's helped predictability, I think now that we've gotten through this complex time, we should see that actually contribute to the predictability of our business and help our ability to drive more consistent execution as we go forward. I'm gonna talk a little bit about how we're going to invest and win in AI. We spent a lot of time talking about that today with our customers, as well as in cybersecurity and just in cloud in general. And so we'll go through where we're spending, how we're investing, and some of the results that we've seen and what we expect as we move forward.
But first of all, for those of you who are joining us, who don't know a great deal about our business, I wanted to try to simplify it for you. First and foremost, everything in the world, from a technology perspective, is being connected. We're connecting everything, and technology is really defining every aspect of everyone's lives, and we all know this. We're connecting factories. We're connecting traditional data centers. We're connecting cloud data centers. We're connecting electrical grids. We're connecting gaming applications. We're connecting branches, home offices, distributed remote users, automobiles. Everything is being connected. So our role, what Cisco does, is we fundamentally provide that underlying connectivity. Whether you're connecting a traditional branch, you're connecting a smart grid, you're connecting an electric vehicle, whatever it is, we provide the underlying connectivity for our customers.
Secondly, we protect those connections and the underlying technology architecture for our customers as well. And we'll talk a lot about the fact that the network and security are coming together in a way that will benefit Cisco as we look to the future. And the third thing that we do, which is what our customers really need in today's complicated world, is we provide insights about what's going on in your technology architecture, what's happening in your architecture, not only on-prem, but into the cloud, through the communication service provider networks, out to those connected vehicles, into the mining applications, or into the factories. We provide insights and help our customers understand what's causing application delays, perhaps, what's causing them not to be able to engage with their customers as effectively as they would like.
All of this is actually going to be accelerated through AI and data. With the Splunk acquisition, we have a massive differentiator with data. If you think about from a cybersecurity perspective, we see. Cisco sees 400 billion security events a day, 400 billion. Splunk is ingesting 4 PB of data into Splunk Cloud, Observability Cloud, on a daily basis. We see. We have visibility to 1 billion endpoints around the world from Cisco Security and ThousandEyes and the data that we get there. We have so much visibility, and as we apply AI and analytics to this data platform that we've brought in from Splunk, we are gonna be able to provide analytics, insights, and information about what's going on in our customers' infrastructure in a way that nobody else can.
There are companies out there who can help our customers connect, there are companies out there who can help them drive security, and there are companies out there who can provide insights, but no one is better positioned to do all three of those than Cisco as we move forward. All of that is in service of delivering digital resilience for our customers. Digital resilience means a lot of things. It means cybersecurity, actually focusing on helping them secure their infrastructure. But it also means avoiding business disruptions. It means avoiding IP theft. It means ensuring that you're in compliance. It means avoiding reputational risk. These are all things that every customer we have worries about and thinks about every single day.
Everything that I've talked about, the connectivity layer, the protection layer, the insights layer, the data layer, the data platform, all of that is ultimately in service of providing greater digital resilience for our customers, and that's what we're going to do. As we talk to our customers, we talk about what the three major areas are that they believe we can help them with. Number one, they all want to continue to modernize their infrastructure. I think we all learned as we entered COVID, how unprepared our customers were from a technology footprint perspective. They have all basically made a commitment that I will never find myself in that position again in the time of crisis, where I haven't kept my technology up to date, and it makes it very difficult for me to deal with whatever crisis is happening at that given time.
And boy, we certainly know that we have a lot of crises around the world right now. So modernization of infrastructure, and this is everything from networking to data center infrastructure, to our collaboration portfolio, helping them connect their employees more effectively for productivity, helping them communicate effectively with their customers. Just... And, and they actually say, "Just be great at networking. We need you to continue to be great at networking. Build the best silicon, build the best software, build the best platforms that help me automate and take the complexity out of this." And that's the number one thing they want us to do, and that's what we're going to do. At the same time, cybersecurity is very important. Obviously, the expanding threat footprint that everybody sees, the landscape is complicated. The bad actors are leveraging AI now to be more effective than they always have been.
Our customers are saying, "You have to help me." And the network and security are coming together in a way that we can help them like nobody else can. And then please help me figure out how to go on this AI journey and how to leverage AI to take advantage of all the data that exists inside of my technology infrastructure. And we'll talk about that later today when Gary and others come up to present. So how does this set us up to be positioned for growth? Everything that I just talked about sets us up with a TAM in our existing markets of $370 billion, and we believe that TAM is growing roughly 6%. We have expansion markets, which are effectively areas in the markets that we play that we can actually expand into very naturally.
They're very complementary to what we do today, and that's another 200+ billion. And then there are always adjacent markets that we could decide to enter, and that, that's represented by the 375 billion. So we are not TAM constrained. We have huge markets ahead of us, and we're gonna go execute and actually take advantage of these opportunities over the next few years. If you go back to September 2021, and we look at what we presented for our core markets versus our expansion markets, some of the stuff that was in expansion has moved into our core, and that's what we would expect to do over time as we go forward. So I want to talk a little bit about the growth drivers in several areas of our business. First being networking, the largest business we have.
This is the area where our customers want us to continue to innovate, deliver capabilities. We announced this morning, Jonathan's gonna talk about it, Digital Experience Assurance, which is a game changer for our customers, and we talked about a lot of other networking technology that Jonathan will cover when he comes up here. But fundamentally, as everything gets connected, the proliferation of devices, the proliferation of connectivity, as everything on the planet continues to be connected, this is good for us. It's a tailwind for us. Return to office. There was a recent survey done where 90% of the companies have said, "We are going to have a return-to-office policy." I think we're all coming to the conclusion that working from home every day, all week, isn't working. So we believe as the customers come back, they'll have to modernize their offices. We've already seen it happening.
We've seen strength in our devices business and the collaboration portfolio, which is indicative of our customers putting these conference rooms together, making sure that they have environments that people are happy to work in. So we think that will be a tailwind to networking as well. Obviously, security and networking coming together will create a tailwind opportunity for networking. Jeetu's talked a lot about Hypershield, about that technology, and how we can actually have distributed enforcement points throughout the network in every network device over time. So as we build out networking products with DPUs that can actually run Hypershield, that will create a tailwind for a refresh opportunity inside of our customers. Right, Jeetu? Thought you might mention that.
Absolutely.
Then overall, cyber and AI-driven refresh. Refreshing their technology, refreshing their networks to be better at defending against cyber, to make sure they have modern defenses, to be prepared for the AI revolution that we see coming into the enterprise. You heard. If you were at the event today, you heard Jonathan announce the first solution that we've launched from our partnership that we announced with NVIDIA, which is called HyperFabric, which includes Cisco networking. I know you're gonna cover this, Cisco networking security, NVIDIA technology, as well as storage. And we believe that that market opportunity over the next few years is certainly gonna be a tailwind for us as well, and we have a whole strategy to help our enterprise customers actually take advantage of that transition. We obviously believe that AI and cloud is a major driver for us over the next few years.
We have the traditional hyperscaler cloud infrastructure that we've talked about an awful lot, and I'll go into detail up here today because I know many of you are interested in that. Well, we have hyperscaler cloud infrastructure, the traditional, then we have the hyperscaler AI Infrastructure. We have enterprise AI Infrastructure, which I just talked about, that Jonathan talked about this morning and will cover in great, in more detail. We also have partnerships that are accelerating enterprise adoption of AI. We talked about a lot of them today. So we have an NVIDIA partnership. We have a Scale AI partnership. We have a partnership with Cohere. We have a partnership with Mistral. We have partnership with AMD. We have a partnership with Intel.
We're gonna partner with the ecosystem players to actually help our enterprise customers make sense of all this in the AI world and actually help them on their journey in building out their AI strategies as we go forward. Then finally, security for AI, which Jeetu will touch on today. As our customers continue to deploy more AI workloads in the enterprise or in the cloud, the security implications are huge, and we're gonna help our customers secure that AI Infrastructure to ensure that they're getting maximum value with minimal risk. As we talk about cybersecurity in general, I've had customers that have told me, "Look, I can't throw more people at this problem. I cannot. I need a platform approach. I have to have a platform that can do real-time threat correlation.
We have to have something that can actually ingest these threats, correlate them, and tell me what's going on in my infrastructure because I can't figure it out anymore." And this was the beauty of having Microsoft on stage with us this morning, talking about the need for the large platform players in cybersecurity to come together because the adversary's on the outside. We have to stop worrying about competing with each other as much as we have to focus on stopping the adversary on behalf of our customers, and that's gonna require a platform approach. There's gonna be a handful of platforms, and we're gonna work across those platforms to ensure that we're delivering the best security for our customers to help them fight these emerging threats. Our customers are facing increasing regulatory requirements. As we know, we see this a lot.
There's lots of things that are coming out of Washington relative to disclosure, relative to different industry regulations that all of our customers are gonna have to deal with, and my belief is that that's probably only going to get more complicated for them over the coming years. Security in this hyper-distributed world, as our customers are dealing with the fact that everything is distributed. If we go back 15 years, our customers had a private data center, and they had an enterprise, and all their technology assets sat within that footprint. That's when perimeter-based security defenses made sense. Today, everything's distributed. Think about it. Branches are distributed, obviously. Mobile workers, both at home or at Starbucks. We've got all of these IoT devices that are being added to the network everywhere. We've got data that is massively distributed.
We're getting run over by data, and it's everywhere. Even the application architectures that we saw 20 years ago, now applications are being broken apart and, and run in a very distributed way. So when you think about security in this world, security in an environment that's so distributed, the entire architecture has to change, which is why we have to create enforcement points throughout the network, and that's what technologies like Hypershield are gonna do for us. So that's gonna be a, a tailwind for us as well. And then, as, as I said earlier, security for AI and AI for security, and this is really important because we need to help our customers. When they're building private AI applications in their private data center, we have to make sure those are secure.
When they're doing custom models in their private data center, we have to make sure those are very secure because you can rest assured that the bad actors are going to try to attack those models to change the results of what queries you might be placing against it. So there's a whole opportunity around helping our enterprise customers secure their AI assets, both in the private cloud and as they move into the public cloud. And then AI for security. Our customers are dealing with adversaries who are actually leveraging AI to actually be even better than they have been over the last decade. They're going to get better. The threats are gonna get better. We're gonna believe them.
Your employees or our customers' employees are going to believe that these, what turns out to be a threat, was actually a very legitimate email or a very legitimate communication, and we have to use AI to make sure that our customers have at least an equal ground in fighting the adversaries. And we believe, with some of the technology that we're building, we can actually put our customers in a better position than the adversaries, which will create a growth, a growth for us in the future. And then finally, this extraordinary breadth of data and intelligence that we talked about.
This is at the heart of the Splunk acquisition, and it really puts us in a position to be one of those platform players, to correlate those threats real-time, as I was talking about before, and really help our customers actually defend their enterprises against the bad actors. Now, we think about these areas. There was a lot of question about how much are we investing in some of these areas, like AI, cloud, and security, cybersecurity. So we want to share with you today that if you look across these three categories, today, we're literally spending 50% of our R&D in these three areas, and we think these three areas are such critical growth drivers for us in the future that this is appropriate, and it could go up and likely will go up as we continue to move into the future.
It's so important because of charts like this, and we're all familiar with it. This is the AI networking infrastructure in hyperscalers and other cloud infrastructure providers. 2027, it's almost $20 billion. We have to be investing. The cycles at which we have to move, the cycles at which we have to spin silicon, the new systems that we need to build, the advances and the speed at which they're moving with these new technologies are gonna require us to continue to invest more and more and more and more. We think this is—we think our investments are appropriate today, but again, we think they'll continue to go up as we move into the future. There's been a lot of interest in our success to date in the hyperscaler infrastructure relative to Ethernet.
I've had a lot of call, a lot of questions from a lot of you about the $1 billion that we've talked about in FY 2025. I think every call back I ever have, we talk about the $1 billion that we talked about in fiscal year 2025. So I thought I would give you a little more detail today than we've given you prior to today about our success and where we are with cloud, number of use cases, how much have we booked so far, and then just reiterate this $1 billion for FY 2025. So let me, let me take you through this. To date, and this is primarily in the last seven or eight quarters, we've booked $800 million, roughly $800 million, just over, in Ethernet infrastructure under AI backend models in the hyperscalers, okay?
We still have line of sight to the $1 billion in FY 2025 that we've talked about before, and we have a robust pipeline. In addition to that, we also are beginning to see the enterprise pipeline emerge, as we've talked about on some of the calls. This does not include any enterprise business. This is all hyperscaler. It's made up of 8,000s, Silicon One, and optics. That's what's in these numbers. If you look at the other part of the chart, what we wanted to give you visibility into is, you all have asked us, "How many, like, use cases have you won inside of the hyperscalers?" So if we, if we start on the top line, total, we've won 20 use cases across the hyperscaler community. Five of those are in AI, 20 that are in production.
That means we've taken business, and they're actually in production. 5 of those are in AI Infrastructure, okay? The second line represents 17 design wins, where we have been awarded the design win in the web- scale space, but we have not yet gone into production and most likely haven't taken any orders yet. It means that we've won the design win, we may be finalizing a chip, we may be finalizing some software, but these are wins that will pay off in the future.... And then the pipeline, obviously, are use cases that we're competing for today. So there are 12 that we're actively competing for, 5 of which are in AI. Hopefully, that makes sense and gives you a little more visibility into what's going on with Cisco in the web- scale space with our AI and cloud footprint.
However, so that you don't have to believe me, we invited a friend of ours to come up and chat with us today. So please join me in welcoming Dave Maltz, who's a technical fellow in CVP for Microsoft, and is responsible for Azure networking. Thanks, Dave.
Hey, Jonathan.
Thanks for coming.
Thank you for having me.
It's good to be here. So Dave, why don't you describe your role at Microsoft?
Sure. So when I introduce myself at parties, I say, "I'm the plumber for the cloud." What that really means on a more technical basis is I'm the engineering lead for Azure Networking, and we're responsible for providing connectivity for all of Microsoft's servers and services across all of our cloud platforms. That's from Office 365, Bing, Xbox, Commerce, our Power BI platforms. And we do that from every layer, going down from innovation in fiber, optical modules that we use to light that fiber, all of the router selections that we use to create our networks, the topologies, the design, build, and operation of our data center, our regional networks, our wide-area network, all the software it takes to manage those networks, and then everything built on top of that, the SDN virtualization layer, up to things like DNS and traffic manager and load balancers.
This is when you're not a ski instructor.
That's when I'm not out skiing.
So, so obviously, Dave, things have changed a lot over the last few years, relative to sort of the classic Azure capabilities, and now here comes AI.
Mm-hmm.
So how is that changing your business and changing the way you think about things?
It's really biggering it, and it's really increasing the scale. So certainly, you've all seen the rapid growth, and that's keeping my team very busy. One of the things that's great is that we've been able to take technologies we've developed together with Cisco from our web- scale, our cloud operations, and apply them to get some of the capabilities that we need to enable this AI scaling. Large language models are fundamentally a networking problem.
Yeah.
You have to move huge amounts of data between all these GPUs and endpoints, and so, getting the scale and the performance, really tight latency requirements, has been a big challenge.
So, thinking about open- source and SONiC, how does that play in this whole thing?
Well, because we're trying to hit this new level of scale and a new level of performance, a new level of exactness in performance, we really needed to create an open platform out of our network so that we could get the telemetry that we need out of it to understand what's happening in these AI jobs, so that we can manage it and make sure that it's working at the quality that our customers expect. And that's where things like Software for Open Networking in the Cloud, which is an open-source operating system that we've developed as part of the Linux Foundation-
Mm-hmm
... together with big players like Cisco, to get that platform running across all of our network devices, from our very smallest switches to our biggest chassis switches. So again, we can then manage and operate the network in a way that'll enable these AI use cases.
In talking with you and Jonathan earlier, we're a big contributor to this with you, so-
Absolutely.
Okay, so how critical is Cisco's technology to what you're doing?
Cisco is a critical partner for Microsoft. I will say that we believe in maintaining diverse supply chains.
Of course, you do.
But Cisco is a critical portion of that. We love the Silicon One ASICs. Again, that's a diverse set of switching ASICs than we can get in other parts of the network. Cisco's been a great partner in technology with us because they operate at all of those levels. They're innovating with us at the optical layer. They're innovating with us at the routing layer to really produce these kinds of scalable solutions that we need.
That's great. Now, you heard me talk about design wins, and while I know you won't go into any specifics, what's your approach and sort of talk about the perspective and how you guys think about that?
Okay. It's really, again, understanding what that customer need is, and then translating that into how are we gonna enable that using the technology that's coming out of our partners like Cisco. With the suite of applications and devices that are available, we've been able to put together some really compelling solutions.
Okay, and I know one other thing, or two other things they'd probably like to hear. Number one, we gotta talk about InfiniBand versus Ethernet.
Yep.
'Cause that's what everybody wants to understand. How do you think about that?
So InfiniBand has been a great technology. I think it'll always be present in our high-performance computing clusters. At the same time, I have 100% confidence that we're gonna be able to meet these AI use cases using Ethernet. It's been the dominant technology. It has better cost, better power performance, than we've seen with some of the InfiniBand work. And Microsoft, together with many industry partners, is working to make that happen. And as I say, I'm confident we'll do that.
Thank you. Finally, working with us today versus a decade ago, is it a little different?
It's very different.
Okay.
If I could be blunt, 10 years ago, I think Cisco missed some of the cloud revolution.
I've been blunt with them already, so they know that.
Okay. Because Cisco was such a dominant network player-
Yeah
... that they sort of thought they knew what we needed to do in the cloud.
All right.
I see a totally different Cisco today. We have roadmap conversations about what the needs we see, Microsoft coming multiple years out, with Cisco, so that they can be producing the technology that we'll need to consume, not just today, but multiple generations into the future.
That's great. I can't thank you enough for coming down here just to do this for us, and I know you've got a plane to catch-
My pleasure.
So I'm gonna let you go. Dave, thank you so much for coming. Dave Maltz.
Thank you. Cheers.
So hopefully that gives you some confidence that we actually are talking to these customers that we talk about on a regular basis. So as it relates to cybersecurity, moving into the next big opportunity, we think that, you know, our core markets are growing 8%. Obviously, we continue to see a lot of, of innovation and, and new product delivery from G2 and the team, and there's 14% CAGR in expansion markets. If you'd gone back and looked at this in 2021, Splunk would have existed in that expansion market area, and now it's gonna be part of our core market. So that's sort of how we think about this as we go forward.
As I think about the momentum we have in cybersecurity, which has been an area that we haven't performed as well as we should have over the last few years, first of all, the firewall is returning to growth. We had a great launch that Jeetu will talk about, I'm sure, about the new firewall we launched today. We're having great traction with the new products that we have put out there, and we had just a massive number of new announcements today. The market is recognizing, and when I say the market, our customers are recognizing it, our industry analysts are recognizing it, and we're beginning to see it in the uptake of the products as we put them out in the marketplace. We're obviously seeing order acceleration. We talked about that on our last call.
The Cisco Splunk story, I think you're gonna really appreciate, and Gary is gonna go through that in just a little while to tell you how that's going and how we're delivering incremental value for customers, bringing Cisco and Splunk together. As I think about our differentiation, as I wrap this up and get our teams up here to talk about how they see growth in their areas, I think, first of all, the increased pace of innovation and the breadth of our portfolio is a big differentiator for us. As I've said before, the amount of data and the insights that we can provide our customers are gonna be very difficult for our competition to actually deliver on relative to the scale that we can.
We have global scale, and this is one of the big things that we talked about with Gary and Splunk, about being able to help them expand on a global basis, particularly through our Go-to-Market and partner ecosystem. As we talk to a lot of these AI players, including NVIDIA, in the AI ecosystem, one of the big advantages they see in partnering with us to build solutions to take to the enterprise is that our enterprise Go-to-Market and our partner ecosystem can help deliver those into the enterprise customer, which would have been difficult for them without the partnership with Cisco.
Then finally, I would say the trust that we've built up with our customers over the last 40 years, we think that's a big deal, and you can see it with the customers here this week and the energy and the excitement with all the announcements that we've made. I really believe that that is a big differentiator for us as we go forward. Let me finish the same way I opened. Our commitment is that we are absolutely laser-focused on growth. After the break today, Scott's gonna come back, and he's gonna share a framework for our long-term model. What I will tell you now is that I believe that our aspiration is to actually exceed that over the next few years, and that's what we'll be focused on as a team.
That's what we're gonna be laser-focused on as we leave Las Vegas, and we wanna do that in a way that's delivering consistent execution. The last four years, as I said, have been really complicated in dealing with all the dynamics that have been in the supply chain and inflation and all those things that we're all very familiar with. And then finally, hopefully, you feel a little better about what we talk about with cloud after having Dave up here. We are going to invest, and we're gonna win in AI, cloud, and cybersecurity. And with that, I wanna thank you all for being here in the room. Thank all of you who are watching out there virtually. And with that, I think I'm gonna bring Jonathan Davidson to stage.
Thank you, sir. All right, well, it is a pleasure to be here with all of you today. Thank you for taking the time. See some familiar faces, and thanks to those on video as well. Thank you, Chuck, for teeing that up. I really appreciate the opportunity to share with all of you what we see happening in the market in a little bit more detail, how we are collectively responding to the opportunities in the market, and how large that opportunity is for us ahead. So with that, we see, as Chuck mentioned, a continued need for secure connectivity. We know that customers of all sizes are going to continue to need to modernize their infrastructure to support their own digitization, as well as the current wave of AI that is sweeping through all industries.
You know, we have seen the transformation of every single business due to cloud. Software has eaten every company, and now we know that AI is going to transform every business as well. With each and every single one of these transformations that have happened, the network has become more important. With cloud, the network became more important. With the transition of every company becoming a software company, the network became more important. With every company digitizing, the network became more important. And now with AI, we think the same thing is going to happen. And Cisco is the preferred networking vendor for almost every single organization around the entire world, and we are number one in nine different networking franchises. So the opportunity ahead for us is clear.
The opportunity itself and the TAM is massive, $221 billion by calendar 2027. And we see four key secular drivers that are going to continue to drive this forward, and Chuck talked about them this morning. I'm gonna give you a little bit more context and a little bit more detail around AI, cloud, digital transformation, and the broad category of access. So let me share a bit about what we see in each one of these various categories. First, in AI, we are deployed in the backend AI Infrastructure of three of the top four hyperscalers.
We are taking that expertise that we have built up over many, many years with those hyperscalers, and we are simplifying it, we're wrapping it with orchestration software, and we're simplifying it so that enterprises can deploy this on-prem, either in their own data centers or out into their edge. In fact, we launched that solution today. We call it HyperFabric, and we see some clear uplifts with where it's going there, and I'll talk a bit more as we go forward. Cloud. We know that both public cloud and private cloud are tailwinds for us, and we know that hybrid is the final state. There will always be private clouds on-premise, always be multiple public clouds, and so we are selling infrastructure for both public and private clouds.
I think importantly, we're helping enterprises securely connect from their on-premise private cloud infrastructure to those public clouds, and we're actually helping enterprises connect from public cloud to public cloud because they need diversity. Third, digital transformation. Every business is now a digital business, a software business, a cloud business, and every business will soon be an AI business. This means factories themselves are digitizing, and we know that all enterprises and all governments have a distinct need to invest in their critical infrastructure, of which networking is a key and critical component. Access. Access is a big number. You need more connectivity and more secure connectivity, whether this is in your campus, whether this is in your branch, or your cloud, your hyperscalers, your digital factories. In fact, this morning, we had a very large Royal Caribbean. We had a nice, beautiful video from them.
If you go back just a few years, each ship had about 2 Mbps going to it, and now every ship is multiple gigs because we always wanna stay connected. We also see continued growth in the communication service provider market, although we know that has been challenged recently. But there is a gentle tailwind of funding globally of $several hundred billion, of which $several billion of those will be applicable to us over the next 8-10 years. So we view that as a gentle tailwind for us. So just like Chuck talked about three things, I'm going to talk about three things with you today.
We are going to go through and talk about these three distinct areas, about how we are helping you, or how we're helping grow the business by focusing on AI, by focusing on hyperscalers, both with AI and outside of AI, and of course, there is a very large growth outside of those areas that we're just calling rest of cloud, which is inclusive of enterprise. Second, we are going to talk about how our differentiated strategy includes taking security and networking and bringing them even more closely together, and how we can overlay observability on top of that. We know from what our customers are telling us, and we see it ourselves, that because of how applications are written today, the network and infrastructure and applications, everything is becoming more complex, and that is driving the need for greater visibility.
And not only visibility, but the desire to have automated action when we know that there is a problem, so that we can minimize downtime. And finally, we have been going through and accelerating our customers' model to solve their greatest pain points by focusing on a platform-based approach, and I'll talk a bit about where we're going with that. If we first start talking about how we plan to accelerate the growth of the company, we'll first talk about the TAM that we see in the hyperscaler segment, giving a little bit more detail about what Chuck shared. This is obviously a large and growing TAM. This is a critical space where we believe that we can be very successful, and we have shown that we can be very successful over the last several years with the data that Chuck just shared with you.
As I mentioned earlier, 3 out of the top 4 hyperscalers are using Cisco in their backend AI Infrastructure. I just wanna call out here the delta, because I think some of you are probably looking at the numbers going, "Okay, hyperscalers, $20.5 billion with a CAGR of 28%. AI and hyperscalers, $8.8 billion with 18% CAGR." This is data that's come out of 650 Group, based upon really moves around where you see and how quickly you see InfiniBand switching over to Ethernet, and that's really the biggest driver. And we are more bullish on how quickly InfiniBand is going to move to Ethernet, especially in the hyperscaler segment, because they love diversity. Now, Chuck gave you the numbers, but I think it's worth calling out again.
We have 20 use cases in production, 25% of those are in AI, 17 design wins, 65% of them are in AI, and we have 12 different use cases in production. Now, Dave Maltz from Microsoft touched upon this a bit, but I'll put a little bit more context around it. It takes a very deep engagement, sometimes for years, to get to the point of a design win. Post the design win, there's a lot of work, there's additional testing that happens to get ready for production. Once we're ready for production, that's when we start receiving orders, and that's when things start flowing through to revenue.
So this could be a multi-year process for every single one of these use cases, and we are staying laser-focused, and we are close to all of these hyperscalers and the large other web -scale providers to understand all the use cases that they have now and all the use cases they want us to participate in in the future, and we are investing heavily in this space. I also want to call out that the way in which we engage with these hyperscalers is very unique because we do a lot for them. We sell them silicon because some of them are going and building their own white boxes. So we have that opportunity. We build optical capabilities with them, and I'll talk more about what that means. We build optics, and sometimes we sell them components for both optics, and sometimes we sell them components for optical.
We do gray boxes. What is a gray box? A gray box is a white box that they asked us to build for them. That's called a gray box, which means that we build a custom box for those hyperscalers. And then, of course, many of them are using fully integrated systems that include our silicon, our hardware, as well as our own software that sit on top of that. So we have lots of different ways in which we can engage with these hyperscalers, and they love the depth and breadth and having a single vendor to be able to go and have consistent technology experience across multiple of their own domains. Now, if we then talk a little bit about the enterprises and where we're going. Talked about HyperFabric for AI, and I'll talk a little bit more about what that is in a minute.
But also, you're gonna hear a term from me called Digital Experience Assurance, and it's how we can help all of our enterprise customers understand what is happening, not only in the infrastructure that they own, but also in the infrastructure that they do not own. So let that percolate in the back of your head, and I will share with you some data that I think you will find very interesting a little bit later in this session. Now, as I mentioned, every single business is a hybrid business, and we need to make sure that we have not only connectivity that's physical, physical connectivity from our wide area network, but also we're looking at virtual connectivity now, 'cause I'm connecting from my private data center, my private cloud, to the public cloud, and I'm doing that with virtual connectivity.
Cisco is in the secure connectivity business, whether that means physical security or whether that means physical connectivity or whether that means virtual secure connectivity. All right. Now, there's a lot of technology spaces that we're in, and so I'm going to talk to you about the sustainable technology leadership that we have in these various areas. And so, you know, we're in Las Vegas, and I thought a little show-and-tell might be appropriate. So I'm gonna start with optical. So in the past, this was a WDM transponder. This is what shoots multicolored light out onto fiber, okay? Out of those folks right there. All right. This is really the legacy way of doing it.
You had to buy a chassis, which locked you into a specific vendor, and then they would sell you the chassis at a low cost, and then they would sell you a lot of these at a very high cost and a very high margin. It's pretty heavy, I'm just telling you right now. It's ... I'm not gonna be able to hold this for much longer. All right? The market has now moved to this. This is a pluggable ZR optic. So what you used to do with this transponder, you are now doing the exact same thing, but in this size. So you can imagine the power consumption of this device versus this device, the heat dissipation, i.e., cooling, which is a big cost inside of the data center, and then also just overall cost. This uses a standards-based interface. All right?
So you can use this from Cisco, and on the other side of the link, it can have another vendor. This requires the same vendor on both sides. And as I mentioned to you, hyperscalers, carriers, enterprises, they all want the flexibility of having multiple different providers of technology. So that was my show and tell for that. One more piece here. This beautiful module here, this is a 1.2 terabit per second per wavelength, wavelength device. This is the highest performance optical pluggable on the planet. So we continue to lead in that space, and it enables significant differentiation for us. We have a fabulous engineering team who's been at the cutting edge of these pluggable, coherent technologies for many, many years. That's just the optical side. If we move into optics, we have a leadership position in silicon photonics.
What that means is that not only can we do things at a superior cost and quality, but we also can participate in additional TAMs, like liquid immersion. And so if you're following the latest in hyperscaler deployments, liquid immersion technology is the way to cool these big GPUs the most cost-effectively. And you need to have a pluggable that you can stick into liquid, and we are leaders in that space. But we're also leaders in other optics capabilities, like co-packaged optics and linear drive pluggable optics, and we are gonna continue to innovate in that space. Third, the innovation wheel of silicon is spinning faster and faster and faster. What we used to do in 2, 2.5, 3 years per silicon generation is now happening in 12 months or less.
What we've seen is, over the last 4.5 years, we have put out 14 different variants of Silicon One. This is a lightning-paced speed of innovation and something that the team is very proud of. And then we also have a blog out there explaining how you can utilize Silicon One to have 2x better job completion time for AI workloads, and looks like the link is not on here. I think actually it is on there. You can get the slides and click on it when you get the slides. Finally, systems.
There's a tremendous amount of work happening in the systems space, and what we've seen is that areas that used to be battlegrounds for software and systems, a lot of that is moving to optical, optics, and silicon, specifically in this hyperscaler and rest of cloud space, which is really good for us because we are leaders in all those spaces and may not be all that great for those in other spaces. Now, if we move forward to systems, what we have seen is you need to be able to take all of these pieces together, the silicon, the optics, and go and differentiate. It's a little bit harder to differentiate in these system spaces, but what we know is that circular designs are critically important for all of our customers, but specifically for hyperscalers, because the speed of innovation happens so quickly for them.
This was a big reason why we went through and disaggregated our portfolio four and a half years ago so that we could sell silicon, we could sell those gray boxes, but we could also go and build and fastly build these custom designs for them because we have scale that they need to be able to meet the needs that they have. And then finally, as it gets to software, and Dave Maltz from Microsoft touched upon this, the open-source networking OS is a critical differentiator. We have leaned into this. We are a top three code contributor to SONiC. We are a member of the governing board. This is not by friends and buddies, how you get voted on, of course. There's an algorithm that was developed that's based upon code contributions and other hard, tangible factors in order to become part of the board.
So we, along with the other board members, to which Dave Maltz is the chair, actually can help steer the future of SONiC. We're big fans of SONiC, where it's going and, and where we see it go, so we are absolutely committed to the future of this, and we actually are in production in two different hyperscalers with SONiC production code. So we're really excited about what is happening in that space. I mentioned the partnership between Cisco and NVIDIA, and the simple way to think about this is the company that brought you cloud networking, so using cloud technologies to manage your campus and your branch, we've just done the exact same thing for the data center, and we have focused it on the AI enterprise on-premise data center. 'Cause I can tell you, building out an AI data center is more complicated.
You have to make sure that everything is working cohesively as a system. The NIC has to be working in conjunction with the networking interfaces. You need to make sure that you've got job completion times as the highest priority, so the application needs to be tightly tied into what's happening across the entire system of infrastructure, much more so than any other type of data center, and it's complicated. And so we wanted to go and simplify this task for enterprises who want to go and deploy on-prem and keep their data on-premise. And this has been the very technical partnership that we have had with NVIDIA for quite some time. We have been selling NVIDIA GPUs as part of our compute systems for over a decade. Obviously, there's a lot more interest now than there was a decade ago, but we have a long-term relationship with them.
So this cloud software wraps all the automation, the orchestration, and cloud management for those on-premise data centers, and we view this, and we see it clearly as a very differentiated offering, which gives us a huge tailwind effect and will help us grow our overall data center business. All right. I'm gonna change gears a bit and talk about how we're deeply tying together our networking portfolio and our leadership in those areas with our security portfolio and all the great innovation that Jeetu and team have done. You're also gonna see a tight integration with the overall observability portfolio. Now, bringing these two things together, the Cisco Networking Cloud, which we launched 12 months ago, and the Cisco Security Cloud, which Jeetu launched 2 years ago, that was planned from the beginning to have consistent and seamless experiences....
The world needs more secure connectivity, and we are the ones who are going to be able to give it to them because we are infusing security directly into the networking fabric itself. G2 launched this a few months ago, and we're gonna have consistent rolling launches as time goes on. I'd love to give you an example, starting with a hospital. A hospital was having problems, and they upgraded their entire networking infrastructure to something that we call Software-Defined Access. What we found was that 70% of the devices attached to the network, they were unknown, whether they were cameras or thermostats or heart monitors or laptops or tablets. What we were able to do is we were able to use our AI-native technology, define what those things are. Are they cameras? Are they thermostats? Are they other things that are connected?
And we were able to automatically put them into what I would call virtual boxes or segments. And we can then go and build out a baseline using this AI-native technology of, is the camera acting like a camera? Is the thermostat acting like a thermostat? And then we can take automated action if those devices or elements are not acting the way they're supposed to be acting. A simple would be sending alert to Splunk. A more, perhaps aggressive tactic might be auto-quarantine and cut them off from any network infrastructure or put them in a walled garden. This is a critical differentiator. This enables segmentation across campus, data center, and public cloud that is all consistent.
And this particular network architecture that I'm talking to you about, that gives you the AI-native endpoint identification, the behavioral analysis, and the quarantining, this is what we call end-to-end micro and macro segmentation. And this architecture is not only used by Cisco, but there are nine other top security players who are also using our architecture. So you can get consistency across networking and security. All right, if we move on to the next one, we're building phenomenal SD-WAN technology, and G2 is building phenomenal security technology, specifically around SSE or Cisco Secure Access. And what that gives you is market-leading SASE solution embedded with Digital Experience Assurance. Now, I think this might be the third time I've said it, and Chuck mentioned it, and you're all going, "What is that?" Stay tuned. I'll tell you in a little bit. Stay tuned.
But what we know is that security plus networking, we know that it's already working. How do we know? Well, I can tell you that 90% of our customers who are using our enterprise routing SD-WAN technology and also who send us telemetry, and that's a lot, use our next-gen firewall features that are embedded into those routers that I get from G2. So this is the power of security and networking coming together for generating great outcomes and additional franchise sales to our customers. All right. To continue on, our innovation engine, as Chuck mentioned, is spinning very, very quickly, and many, many years ago, we became API first. But it's hard to understand, like, what does that mean for our investors? Well, for our customers, it makes them very happy.
We are seeing every single day, 500 million API calls just into the Cisco Networking Cloud. What's the relevance? Well, the relevance for us is that it makes our customers very sticky because they're building business process workflows that connect down into their infrastructure, and when they're using our APIs, it becomes embedded into what they're doing on a daily basis. And so what it's driven is for customers who are using our APIs, a 98% renewal rate, and also translates into 2.4 times more spend with Cisco. APIs are very sticky, and so we will continue to remain an API-first focused company. And we are continuing to invent the critical technology that all of our customers are needing, not just the hyperscalers, but our enterprises, government agencies, and all organizations around the world.
Over 3,000 just networking patents in the last 3 years, and these are issued patents. So our innovation wheel is continuing to spin quickly. Now, I have said AI- native a few times today, and some of you might be thinking, "Well, what does that even mean?" AI- native means that we have baked AI- natively into every single thing that we do. The systems that we build, the management software that we build, the cloud management software that we build, how we're integrating between security, networking, and observability is all based on a foundation of AI. And having AI- native means that you are a step closer to a resilient infrastructure. All right, I'm not gonna talk about all of these innovations on the right side, and I just really picked five of my favorite ones.
So I'm just gonna talk about 2 of them with you today, but you can Google them, and you can see more about what they are. So RRM, I know it rolls off the tongue, Radio Resource Management... and why is it useful? Well, all of you are sitting in front of me, and I'm sure people who are on the webcast, you're all using, likely using Wi-Fi, and you're in a different position now than you were a couple of hours ago. And so what we are able to do, where you might have tuned your radio network once a month, once every 2 months, or every 6 months, or maybe once a year, maybe never, we are able to do this on a continuous basis because the rooms and the dynamics in the room, and where the user happens to be, is constantly evolving.
And so we are using AI-native technologies to modify the RF signals to make sure all of you have a great experience every single time that you're connected. So that's one thing. The second thing I want to talk to you about is WAN Insights. We deployed this over two years ago. This is all about using AI-native technologies for forecasting what you might need, to see what you might need in the future, to give you path recommendations, to give you proactive policy changes in order to give every user and every application the best possible outcome that they do. And this is another huge step towards that digital resiliency. All right. That was the first two. I'm gonna talk about the third one now, and how we are fundamentally transforming the customer model. We're evolving the customer model, and we're focusing on turning networking into a platform.
Just like G2 has done in security and turned security into a platform-based approach, we actually have been doing the exact same thing, we just haven't been talking about it, of turning networking into that as well, so that we could go from selling individual franchises, data center switching, data center compute, routing, to tying these franchises together to enable a greater land and expand motion. Now, I think all of you are thinking, "Well, how many franchises are there?" Well, there are 12 networking franchises that we are laser-focused on. Let me explain the journey that we have been on across each of these. All of you today, who are not technologists, this is what you think the network looks like. The digital experience goes from wherever you happen to be to your application, everything else, magic.
I see a few people going, "Yep, that's what I think." All right. The reality of it, though, is it looks a little bit more like this, and this is a significantly high-level cartoon slide. So for the next 4 hours, I'm gonna walk through each of these... I'm just kidding. This is just to give you a little bit of a view of the reality. It is complicated to get all of the peoples, places, and things to the applications that they need to achieve the outcome that they're looking for. And guess what? A lot can go wrong when you're going down this chain. And we have wanted to solve this particular problem for decades. We are in our 40th year in existence as a company, and we've wanted to solve it for the entire time. The challenge is this: infrastructure's gotten more complicated over time.
There is infrastructure that your companies own, like your data center, your branch, your campus, when you're working from home, and then there's a bunch of infrastructure that you do not own. You have your last mile to your home, you have your service provider, there might be a transit provider, of course, there's lots of different clouds. So how can you assure digital resiliency when you don't own the infrastructure and you don't see it? But with networking, security, and observability, we're able to federate the data and deliver digital resilience, which means that we can deliver a good user experience in the face of any disruption. So Digital Experience Assurance, which we actually announced today for the first time, is what enables us to deliver that digital resilience, and it is enabled by ThousandEyes.
But it all starts with data, and Chuck gave you some of these data points. One, 1 billion endpoints we have visibility into. We have visibility into 650 billion daily points of measurement, and we're doing over 5 billion synthetic measurements every single day. We have the data. This is also a huge sustainable differentiator for us. In fact, for every new ThousandEyes vantage point that we deploy, which more happen every day, we get another new 5,000 pieces of data every day for each vantage point. And then we use that data to continuously create intelligence to understand your individual digital experience. We can pinpoint things at the application layer, the endpoint, the network, and everything in between. And then we can help you and your teams remediate in real time or near real time. So what's the outcome that we're driving for?
You get a customized view of what you own and what you don't, but you only see the parts of what you don't own that are relevant to you. You don't wanna see the whole Internet. You just wanna see the path from where you are, or the application you care about, to where it resides. That's all you wanna see. What we've seen is the outcome of this is the difference between a multi-hour outage and a few minor minute disruption. Okay, but there's more than simply data. We know that our products are being used. I talked about the 500 million different API calls. We also see 260 million mobile IoT endpoints, of which 100 million of them are connected cars....
Less than 10% of all cars around the world are connected, but by 2030, analysts are saying that close to 100% of all cars will be connected. Depending upon who you're looking at, that's 60-80 billion cars a year. But there's a lot more than just cars as part of this number. We have 29 million networking devices sending us telemetry every single day, routers, switches, and more. Now, that's a lot of data. It's hard to comprehend 650 billion points of measurement. That would be like 100 times more people than are on the planet today. So everybody you've ever met in your entire life, imagine there's another 100 of them. That's a lot of people. All right. Data enables you to change outcomes.
So I wanna share some exciting data with all of you, and also a thank you. So to those of you who registered for this event, about 30 days or so, over the last 2 weeks, what we have done is we've looked at your companies using our ThousandEyes capability to understand you and see how your experience has been. All right? So this is publicly available data. We don't have any magic sauce to look inside your infrastructure, so nobody freak out. All right? But we looked for events that impacted your digital resilience, and this is what we found out. And this first one's not surprising. 100% of you depend upon cloud services for video collaboration. That one's not a shocker, but 64% of you depend upon SharePoint, and 41% of you depend upon Salesforce. Here's a fun one.
Nine of you had a CRM outage in the last two weeks that lasted four hours, and five of you were impacted by a DDoS outage last week. In fact, one of you, because of that DDoS outage, your website was having problems for seven straight hours. Now, the impact of that outage was probably a struggle for your IT teams to understand, and it probably took quite a time to try to remediate or what could they even do. This is what we often have called the mean time to innocence, because one of these big problems happens, what happens is you get 30, 50, 80 people on a bridge coming from all the different teams, and they're all trying to figure out, is it the application? Is it the cloud? Is it the optical team? Is the data center team? Is the campus team? What's going on?
All right, everything's green, but it's not working. What's going on? It's not fun. If you were a ThousandEyes customer, we could look inside your infrastructure and give you even more data and give you the potential for real-time fixes as well. That's the power of data, that's the power of being AI native, and this is what ThousandEyes is enabling us to deliver for all of you. All right. Hopefully you've seen a little bit about our sustainable differentiation and what the power of data is across our networking platforms and how we're tying to security and observability. Because whether it's enterprise connectivity, data center connectivity, or cloud connectivity, we are the right people to partner with, and enterprises continue to vote that direction.
Our networking platform is tightly integrated with our security cloud platform, and with Splunk, we will continue to land and expand additionally even more. So I'd like to give you a few pieces of data that I think might be relevant to you. We start with the networking platform. We have 12 franchises that exist there. We can then upsell, cross-sell with security franchises. We can then upsell with observability. I just talked about digital experience assurance. That will be tied into Splunk as well in the future. But we also can now help people go AI native in the campus, make their campus a smart campus. We can take your heating, ventilation system and tie it to our Webex system, so that if someone's in a room, we can turn the cooling on or the heating on. This is all done through the cloud.
We actually have more square footage under management than any other company in the world. From an understanding, just like Google and Waze understand the physical world, going from place to place, we understand the mapping world inside of campuses who are our customers. And this is really the power of platforming, which we do in the campus, we do it in the branch, and we're doing it in the digitization of other industries and factories, and I talked about where we're going with, with connected cars as well. Now, this strategy, the power of platforming, is working. And one of the proof points that we have is that we are estimating that we will exit this fiscal year with a $6 billion annual recurring revenue software networking business.
We never shared that number with you before, but Scott felt it was important to let you know the scale, and the magnitude, and the size of, of that business. Also, proof points that the platforming approach is working in networking. Across those 12 franchises, 90% of our customers have-- of our top 2,000 customers, I should say, have 5+ franchises, and 20% of them have 9+ franchises. Now, for each additional franchise, the ARR spend goes up 1.3x with Cisco. Okay, those are also great match. So there's a potential for 70% or more of customers to have 4 or more franchises added. Now, I'd like to give you a few examples just to bring it home for you as well. We have a really large retail gas company here in the United States.
They have deployed Cisco routing, they've deployed Cisco switching, and they've deployed Cisco wireless in every single one of their gas stations, and there are many. But they've spent more with us on cameras than the other three franchises combined because they need 25-40 cameras at each and every single one of their gas stations. But we have a common platform, the Cisco Networking Cloud, that enables you to easily cross-sell and upsell across these different franchises. Now, these used to be individually one, but now there is more stickiness with this overall platform approach, and with our networking cloud, we're gonna keep driving that land-and-expand motion even more through the Digital Experience Assurance, through what we call Cisco Spaces, tying the physical space to the virtual connectivity, and of course, there's gonna be more software security and observability.
Now, I wanna land with you with finally with three customers that I think really help explain this land-and-expand motion. The first of them is Self Esteem. If you're not familiar with them, they own thousands of stores and, thousands of gyms, not stores, thousands of gyms, and they are a Cisco customer. They have Meraki at the core. They have 6,000 gyms total. 350 gyms are equipped with our technology, and we have an enterprise agreement with them for another 1,800 gyms that are going to be deployed. Now, we have been able to add franchises, both for this physical world, but also for the cybersecurity world. And I'll give you another quick camera use case, like why are cameras important? Well, this gym is open 24/7, but they don't have an employee there 24/7.
We were actually able to, using computer vision and the cameras, detect when there was a tailgater. So I badge in, walk in to do a little bit of weights, and somebody jumps in behind me. That actually automatically triggers an API and alerts the owner of that site. This is a custom application that was relatively easy to go and build. Hilton: 500,000 networking devices at 3,500 hotels. This is massive scale. But what's most exciting to me is, once they deployed Cisco infrastructure, 75% fewer complaints to the front desk and 90% drop in the number of tickets being opened because of infrastructure.
Of course, it's good for anyone who goes and stays at Hilton because you have a 10 times better Wi-Fi experience, 10 times better performance than you had before you deployed Cisco, and this is a, a great outcome. Now, Bosch, this is not only technology innovation, but it is also around business model innovation as well. Bosch signed a 5-year whole portfolio agreement. That gives them access to the entire Cisco portfolio of networking and security and, and observability as well. We turned 2,000 contracts into one contract, and they already had many of our franchises. They had switching, routing, data center, assurance, security, collaboration, services. It was all rolled in there. But once we signed this WPA, they got access to all of our SD-WAN software, and so they've expanded into that franchise as well.
That's the power of having this networking platform and the ability to go and cross-sell and upsell on top of that. So to summarize everything here, we are accelerating growth. We are focused on the AI and on hyperscaler. We're taking that knowledge through HyperFabric to the AI enterprise on-premise data center that is managed from the cloud. We are propelling security, secure networking forward by integrating security and fusing it directly into the network fabric. We're evolving networking from having multiple franchises to a platform-based approach to make it easier to land and expand these franchises. So with that, I'd like to invite up my good friend, Jeetu Patel, who leads our security and collaboration business.
Thank you, buddy. By the way, I really apologize about that picture. That's, that's a terrible picture. Or is that how I look? So what I thought we'd do is we'd start with this notion of modernizing infrastructure, and I'll actually walk you through two category areas. The first one is collaboration. I'll walk you through that. And the second one we'll go through is security and detail. So let's start with collaboration. So once again, like Chuck mentioned, we are not TAM constrained in this market. We've got about a $90 billion market. One of the key and most promising use cases, though, that we have over here, other than, you know, what's really gaining traction with AI, which is coding, is this notion of customer support.
Customer support, every first-line call that anyone makes can actually get automated with a virtual voice agent, and that's a huge area of opportunity for us. That's actually one of the big areas of growth in TAM as well. It's a board-level initiative. The reason it's a board-level initiative is most companies really care about: how does the customer experience directly impact loyalty towards the brand when something is wrong and they call in? And how do you make sure that you continue that, that dialogue? And we happen to have a tremendous kind of footprint in that market that we can go, continue to compound. The second big area is around, you know, devices, and so I'll walk you through a couple of these a little bit more. And one of the things to keep in mind...
As we're going through this, is the strategy hasn't changed in collaboration over the course of the past couple of years. Our strategy remains the same. It's a single platform. It's got pervasive AI built into it, the fabric of, the platform itself. And there are three things that we're trying to, the three key priorities that we focus on: reimagine workspaces, specifically as campuses get refreshed, reimagine work, and make sure that you have a suite of products so that you don't have to buy individual components, and then the third one is make sure that we help reimagine customer experience itself. So that's the, that's the core experience that we're talking about.
Now, what I thought we'd do is start out by walking you through two key growth drivers that we think are gonna help us, you know, drive this business forward. The first one is interoperable devices, and then the second one is AI-powered contact center. Let me walk you through the interoperable devices first. So what specifically are we talking about in this market, and why is that such an important market? Well, as you start to see customers mandating and companies mandating their employees come back into the office, about 80% of the companies have mandated, in some way, shape, or form, either partial or full-time return to the office, that the employees have to come back to the office.
Actually, what's different this year in this survey was 72% of employees are actually excited about wanting to come back to the office, which was, which was something that was not the case two years ago, right? The goal of the year is to make the office a magnet rather than a mandate, so that when people do come back, they have a very, very productive experience. Even though employees have shown excitement and employers have mandated that their employees come back to the office, the offices themselves are not ready. Over half the office conference rooms actually aren't video-equipped. So that's a pretty important piece to keep in mind, and employees expect a fundamentally different experience when they go back to the office. They don't wanna go back and just work in the cubicle.
They wanna go into the office so that they can collaborate with each other. So what we've done is we've actually really focused over the course of the past year and a half or so, on this notion of distance zero, which is, how do we take away the distance between two parties when they're engaging with each other, even if they're virtual? 'Cause even if everyone's back in the office, the offices tend to be pretty large, and so what do you end up doing over there? And so one of the things that we've done over here is not just focus on collaboration devices, but how do you also refurbish and modernize the infrastructure with networking and sensors?
What you also need to do is make sure that the smart building technology can be infused in the power over Ethernet, so on and so forth. We've got, at this point in time, a pretty massive investment that we've made over the course of the past few years in having Cisco devices in every workspace, right? Whether it be a conference room, whether it be a home office, whether it be a huddle space, whether it be a boardroom, we've actually got all of these, these devices for different form factors for different purposes.
One of the big decisions we made about six years ago was we equipped these devices with NVIDIA chipsets, and that's actually really paying huge dividends for us right now, because what we're then able to do is take a lot of the AI capability that we had in Webex and move it to the devices. And because of that, the interoperability with the third party, so we don't just have these devices work with Webex, you know, kind of, so, you know, the software stack, but we also have it work with Microsoft and Google and Zoom. And so those products get better as you start to use these devices, because all of our AI capability that we had, for example, noise removal, can now also be applied to those other stacks.
So one big thing that we did to add to this last year was we said, "Let's make sure that we double down with Microsoft, with Microsoft Teams on our devices." And so Microsoft Teams Rooms, MTR, runs natively on our devices. And what we just have done now is, at this point, we have a complete portfolio of devices that are certified on Microsoft Teams, and in fact, this has gained a fair amount of traction in the market, you know. And our devices, like I mentioned, make Microsoft Teams better because we've got AI packed into it, and that can actually be applied to Microsoft Teams because it runs on the device, on the firmware. And the beauty about this is every Microsoft shop now becomes a potential customer for Cisco devices.
And so as you think about this, one of the big things that we've seen is there's three major demand drivers for devices growth, for video devices, for equipping modernized infrastructure within campuses and within offices. The first one, of course, is the office upgrades. Most people, as you're returning back to the office, the offices are getting refurbished. They're actually starting to get reimagined, so that that becomes a key driver. The second big driver is interoperability with other conferencing solutions. Most organizations don't actually have just one conferencing solution. They might have one conferencing solution that they might have standardized on, but their customers might have a different one, and so you need to make sure that you can go back and forth between basically three or four conferencing solutions: Microsoft, Zoom, Google, and Webex.
And so that second one is a big growth driver as well, because everyone has multiple platforms. And then the third one is hybrid work transformation, where we wanna make sure that it's not just about collaboration, but collaboration tied with all of Jonathan's portfolio on the networking side. Cisco Spaces, navigation into a property when you go into the property, so you know which conference rooms are available, which ones aren't. That can be shown visually on a map. Wi-Fi access points. All of those things tend to come together for modernizing infrastructure, which was the top priority for what, the three customer priorities that Chuck talked about. So what this has done is actually last year, our devices business was in negative growth territory.
This year, we're actually starting to see that we're not only seeing growth, we're seeing double-digit growth on devices. And I expect this to continue because we're seeing a very strong partnership with Microsoft, we're seeing offices getting refurbished, we're seeing that our portfolio is very complete and certified with Microsoft. So we have a lot of tailwinds in this market. The second big area that we talked about, the second big growth driver... So the first one was interoperable devices. The second big growth driver is AI-powered, self-learning contact centers. And contact center is a huge growth driver for us. It's basically a board-level initiative, like I mentioned. It is a single platform.
One of the big advantages that we get with this is all of the innovation that we've done on Webex, the core meetings platform. We can utilize a lot of that innovation with AI also on the contact center platform. So if you think about noise removal, if you think about speech to text, if you think about all of the language-based, you know, kind of capabilities that we've innovated over the course of the past five years, all of those things now can be applied to contact center. So what we have is the contact center portfolio is built on the same platform as Webex. It actually has three core use cases that we serve. We serve a use case, which is proactive communication.
So if a utility company wants to notify you of an anticipated power outage that's gonna happen, how do you go about doing that? That's the proactive communication side that we can do. The second one is self-service. So if someone wants to just go out and go on a website and self-serve themselves for a problem they have, how do we make sure that we can enable that for our customers? And then the third one, of course, is human interaction. And all of this can be done through a multitude of different channels while maintaining context. So if I start my conversation on a website, then I go to Webex, I'm sorry, then I go to WhatsApp, then I go to Instagram, all of those will actually consolidate together so that the contact center agency sees exactly how the status of the conversation is.
So you don't have to keep repeating where you are in the conversation. So that's basically the AI-powered self-learning contact center. And one of the things that we've been able to really keep in mind... The thing that I want you to keep in mind in this entire portfolio is this isn't just about driving collaboration revenue. As we use collaboration as a key wedge into organizations, it actually drives a fair amount of revenue for Jonathan as well, because it's all around modernizing the portfolio. So the pull-through revenue of Wi-Fi access points, of campus and branch networking, of smart building technology is actually a pretty meaningful synergy between Jonathan's team and my team.
That's something that we continue to anticipate to, you know, expect over the course of the next few years, because there's a fair amount of upgrades and refreshers that need to happen as people start actually returning back to the office. While 80% have mandated people come back, the enforcement levels tend to be pretty low. So that's where we are as it pertains to the collaboration portfolio. There's a fair amount of progress that we've made on the products itself. By the way, just as a data point, when I took over this team about four years ago, we had our NPS score was close to about 20. We are currently at about 68. So there's a fair amount of improvement that we've seen in customer satisfaction when they're using the product.
So that's the collaboration portfolio. Let's now talk about the security, you know, portfolio as well. And on the security side, hopefully, folks, you got a chance to see the keynote. Many of you might have seen the keynote that we did earlier. Once again, we're not TAM constrained in this area as well. We've got about a $118 billion TAM, and what's happening in the security market is extraordinary right now. The attacks are getting extremely sophisticated, and AI is actually being used to weaponize attacks against core critical infrastructure. So every company, every government institution, every country is starting to think about cybersecurity as a top priority.
The second thing is the architecture itself that you need to have to go out and make sure that you have a stronger security posture today is very different, as Chuck mentioned, compared to the architecture that needed to be had 5 years, 10 years ago. It's a hyper-distributed architecture. So you need to have rather than having a three-tiered architecture, you have thousands of microservices that sit across hundreds of different pieces of hardware on Kubernetes containers and Kubernetes clusters and on VMs, and it's just hard to secure that infrastructure. And third, you have to assume that the attacker has already infiltrated most organizations, and the name of the game is preventing lateral movement. And lateral movement happens on the network.
What I mean by lateral movement is if someone wanted to steal a bunch of credit card numbers, they don't go directly to the credit card database. They come, they come in through an email phishing attack, and then they might have 30, 40, 50, 60 different hops in the system before they can actually start exfiltrating data. And so that's, that's what's happening in the market. Now, what is our strategy to fundamentally differentiate in this area? It's a three-pronged strategy... is a shift from those point solutions to integrated platforms, right? And that's what's happening. So the first differentiated component of our strategy is we are one of the broadest end-to-end platform that's out there in the market. The second big area is this notion of a hyper-distributed architecture that requires hyper-distributed security, so that your security posture can be strong as this architecture shift is happening.
As Jonathan mentioned over here, what we're doing is we're taking security, melting it, and infusing it into the fabric of the network, so that every access point on the network, every top-of-rack switch, every server can actually have a security enforcement point. Every IoT device can have a security enforcement point. And thirdly, especially given the Splunk addition to the portfolio, we can now extend our leadership in the SOC. And this is the first time that we've seen a fair amount of, you know, relevance for us from a SOC perspective across the entirety of the food chain. Regardless of the maturity of the company, you can actually see that. We'll talk about each one of these in detail. Let's talk about the platform shift first. What's happening on the platform shift?
If you look at the—in 2023, the amount of innovation that we had was enormous, and in 2024, you're gonna see multiples of that. And these aren't just features that we've built. What these are, are product categories that we've built from the ground up, and they are fundamental architecture shifts that organizations are thinking about right now. And this isn't something that... We've, we've just started launching these, these capabilities. So we've been building them, we've been hard at work building them over the past couple of years, and we have just started launching these capabilities out in the market. But they are all now, most of them, are out in the market today for a couple quarters.
The second big thing that we've seen over here on the platform that's really important to understand is that the sentiment in the market is completely changing. Like—like, like Chuck mentioned, the customers are starting to see the innovation engine really kick in, and they're getting excited about it, but so are the partners, and so are the analysts. In fact, with Splunk being added to the portfolio, this is the tenth consecutive time that Splunk is a leader, and they progressed in this. And there, if you look at that red dot, that's where they are. They progressed in this in this Magic Quadrant over the course of the past year. So there's a tremendous amount of momentum that we have from a platform perspective that we've seen. Now, what exactly do we mean by a platform?
What have we delivered in the platform? We announced this, this construct of the Cisco Security Cloud, which was a global cloud-delivered service that we wanted to make sure that we delivered to the market, that could acquire and steer any and all traffic to any of the cloud providers and persist policy as you move your workload from one cloud to the other, right? And the goal over here was to solve three key problems for customers: How do you protect a user? How do you protect cloud and cloud infrastructure, both public and private? And how do you protect against a breach, so you can, in near real time, detect, respond, remediate, and recover from a breach as quickly as possible, so that you have resilience, a digital resilience from a security perspective as well?
And then all of this was built on the foundation of a firewall with AI infused into the fabric and identity infused into the fabric, so that identity would enrich every single one of the products we had, and AI would enrich all of the products that we had. So one of the big concerns that our customers have highlighted in the past is Cisco is very complicated to do business with in security, 'cause we had about, you know, 30 different products and over 1,000 different SKUs. And so what we've done is we've massively simplified the portfolio. So when you buy the Cisco Security Cloud, you can buy one of 3 suites. What are those suites? Coincidentally, those suites actually map to the 3 problems we're solving. There's a user protection suite, there's a cloud protection suite, and there's a breach protection suite, and of course, there's...
You can buy, you know, firewall in a multitude of different form factors. The suite will expand over time. So we will continue to keep adding capabilities to the suite and make sure that we have tiering of the suites. Essentials, Advantage, and Premier are the three different tiers, and we will continue to keep making sure that we add capabilities so that you can increase your ASP in the market. These suites also have a common integration point, so they're fully integrated together. These aren't just a bundling exercise. This is not just a bundling exercise. This is a full integration of the products together, and so there's a lot of cross-selling capability that you would have enabled because of the suite, right? Now, how do these suites work, and what do you actually start with the suite?
Because virtually 100% of our customers will actually have some sort of, you know, security products that are already within their environment. It's not just Cisco. It's not just a homogeneous Cisco stack. So we need to make sure that we are very clear about what are the lead products, anchor products, that can actually be used as an insertion point into an account. And those anchor products have to be meaningfully differentiated from the competition in the market. And so if you think about the user protection suite, the anchor product we have is what Jonathan was referring to, which is Cisco Secure Access. If you think about the breach protection suite, the anchor product we have is Cisco XDR, which is our extended detection and response. If you think about the cloud protection suite, we just announced Hypershield, which is gonna be available in August.
That'll be the lead product. That'll be the anchor product. And of course, you'll be able to sell firewall in multiple form factors, virtual firewall as well as hardware, and then also a completely SaaS-based version, right? And each one of these products have massive differentiation, but they also have very tight integration into Jonathan's portfolio and into Splunk. And so what we're trying to do is make sure that we are prioritizing integration across the portfolio as much as we are depth of the, you know, features and capabilities we are building for each one of these products. And that's actually starting to work really well. So an example of this is us recently announcing also what we've done on the platform, not just on the core products, but also with investments in hardware.
So we just announced a Secure Firewall 1200 series, and with this, we've basically refreshed hardware across the entire enterprise lineup for the firewall, and firewall is a pretty substantial business for us, right? One of the big challenges we've had with the firewall in the past is our quality had dropped a few years ago, and as a result, we were starting to lose share. Over the course of the past year and a half to two years, we have massively improved quality. Today, we've got those customers that use our latest version of the software of the firewall see a 40-point improvement in NPS. A 40-point improvement, right? And what we're now doing is we've actually updated the hardware infrastructure as well. So this is an up to 3x performance improvement compared to competition.
With this, what you've got now is an entire lineup of the firewall, low end of the market, mid and high, and high end, which has up to a 3x price performance advantage. In addition to that, what we're able to do here is make sure that this is a single appliance box that'll have advanced security capabilities as well as SD-WAN capabilities. That's, that's an example of what we've done in the entirety of the Cisco Security Cloud: User Protection Suite, Cloud Protection Suite, Breach Protection Suite, updates on the firewall and the quality of the software, new hardware refresh lineup across the board, and 3x price performance advantage for what we are able to do versus competition.
And what we're starting to see is very, very clear, you know, differentiation with competition, where in XDR, because organizations have kind of seen our ability to be able to correlate data across multiple different control points. So you might have a phishing attack that starts from email, then you get directed to a website, you might download some malware on your endpoint, and then that has lateral movement on the network. What we are able to do is correlate data across all of those different control points. So alerts, which might have been low-level alerts that you would have ignored otherwise, can now be correlated to graduate to a high-level alert that you know is a breach in the making, you know? And so XDR organizations using XDR and the, the...
Those organizations that use XDR the most are the ones that actually don't have a very sophisticated SOC environment. Most of them don't have a SOC, or they're just starting out building a SOC. For Secure Access, you usually see organizations, because of the simplicity and ability to integrate with the rest of the Security Cloud, they're using us quite a bit. And then our Secure Access capability also ties in with all of the technology that Jonathan has. So it's integrated into Meraki, and it's gonna continue to get integrated into different kind of, you know, points within the networking franchise. And like I mentioned, on our firewall, our latest software upgrades and our Go-to-Market programs that we have, have seen new wins and win backs in firewall at a very, very healthy rate, and we're really pleased about what's happening over there.
Hypershield, which is not yet generally available, we are completely oversubscribed on the number of customers that wanted to do early access programs on Hypershield. So we're over capacity on the early access customers. There's a waiting list, and we're starting to see some tremendous feedback from the market as we go through it. Now, in addition to this, what we've done, in not just the security portfolio, but we started with the security portfolio and have expanded out, is we're really investing heavily in AI with the Cisco AI Assistant. The Cisco AI Assistant is a universal assistant that's gonna be unified across all of the Cisco properties over time, and it's a skills-based architecture. So we now, for example, have a firewall assistant that's part of the Cisco AI Assistant, where you can set policy for a firewall using natural language.
We've also got a SOC assistant that allows you to. That notifies you of a breach that's occurring, and you can actually, as a SOC analyst, discuss strategies with the SOC assistant on what you want to do, what do you want to quarantine, what, when, what might you want to do as a remediation to the breach, so on and so forth. And so what we now have is this one single architecture for Cisco AI Assistant, so that if you have a ThousandEyes skill or a networking skill, you want to plug into it, you can plug that right in. If you want to plug in a Splunk skill, you can plug it right in. If you want to plug in a Control Hub collaboration management skill, you can plug it right in.
So you have one Cisco AI Assistant, one design language, one way that works, and the beauty about this is it correlates datasets across all of these. So your troubleshooting can now do things that you could never do before... you know, because you can look at what's happening on the network, what's happening in the security configuration, what's happening on your WebEx application, and actually tell what might have gone wrong. So, so that's essentially what we have for the first area that we talked about, which is platform. The second area then is hyper-distributed security. That's the whole architecture of moving from a three-tiered architecture to a microservices-based architecture as you move into, a very, very high-scaled distributed environment. And what you're seeing over here in hyper-distributed security is that it is an opportunity for us to reimagine the security architecture.
The reality is people are spending more and more money and more and more time in security, but the efficacy has not gone up. The efficacy has not gone up because some of the core, hard, heavy lifting is still very manual. There is an opportunity, because of AI, to go out and fundamentally rethink hard problems that could not be solved in the past that we can now solve in a completely reimagined way. So let me walk you through three of these hard problems that I think are gonna be super instructive on the kind of relevance this product has. By the way, Hypershield does not really have a direct competitor from our security competitors in the market. This is a very unique architecture, something that we've built from the ground up, as we were kind of building this thing out.
So, what are the three problems that are tactical problems? First one is segmentation to prevent lateral movement and contain the attacker, is really, really hard to do in this hyper-distributed environment. You know, and so when you had just a three-tiered architecture, it was pretty simple to go out and do segmentation because you could segment certain pieces of hardware. Now, when you have thousands of microservices across hundreds of Kubernetes clusters and containers, it's very hard to make sure that you can do segmentation right. So we got to solve this problem. The second big problem is patching, is really hard.
So one of the big things that you're starting to see in the industry right now is the time at which a vulnerability was announced to the market, to when an exploit actually happens, is single-digit days and moving down to hours and eventually will be minutes. But the amount of time that it takes to go patch a vulnerability is about 22-49 days. So you've got this window of time where there's massive exposure for the organization on what to do, right? We got to solve this problem. The third one is that upgrading dated infrastructure is extremely hard. There's 2 change control windows in a year, 1 usually during Christmas, 1 during Fourth of July. If you miss one of those change control windows, you have to wait for 6 months. Why is it so hard?
Because you have to bring down the system, right? And you have to do testing and deployment of these new versions before they get upgraded again. And so what used to be these infrastructure elements that were supposed to protect you are actually becoming elements which attackers are using to weaponize against organizations because they can now infiltrate through that vulnerability that might be there that's not been patched. So that's an area that we need to go out and solve for. Now, all of these problems have actually been solved in suboptimal ways to date, and they've been extremely manual in nature. For the first time, we can actually solve them in a very innovative way because there's some core technology building blocks that are now available that were not available before.
The three of the core technology building blocks that we've used, which we think are meaningfully gonna reimagine the way that you can go solve these problems. First one is AI, being natively built into the product for the defenses rather than being thought about as an afterthought. But the second one is this notion of kernel-level visibility. Knowing exactly what the operations are that are happening on the server, and every process that kickstarts on the endpoint and terminates on the server is really important. And there's this technology called eBPF. It's an open-source technology from a company that co-created it with Meta called Isovalent. We acquired that company called Isovalent. They have one of the largest distributions of open source called Cilium, and we use that kernel-level visibility in Hypershield extensively to go out and solve some of the problems that we talked about.
And then the third area is hardware acceleration, which is you can actually have DPUs now, data processing units, that are optimized computational systems for repetitive network functions like connection management and encryption, that you can have to really make sure that you drive a fair amount of throughput, way more than what you could do prior to hardware acceleration, the way that we had it. That's why we've actually made sure that we've had tremendous amount of success in partnering up with some of the chip manufacturers as well. All of these building blocks is what we used to build this capability called Cisco Hypershield. It's a highly distributed security system that's AI-scale, completely cloud- native.
And what we've done with this is we've actually taken security, melted it, infused it into the fabric of the network so that you have tens of thousands of enforcement points, potentially hundreds of thousands of enforcement points, that you'll have distributed throughout the network. So if you could take your security and move it to the workload, rather than having your workload come to the security. And so when this happens, what you'll essentially have is... Think about a security that used to sit on a box. Now, you've basically melted that and distributed that through hundreds of thousands of enforcement points all throughout the network, so that you have a software agent right next to every Kubernetes container... you can have a DPU on servers, and you can also have DPUs eventually on top- of- rack switches.
What this will allow us to do is not just go out and sell Hypershield, it'll allow us to hopefully have switch refresh opportunities within the switches in the data centers, right? So this is very unique, and by the way, this is something that only Cisco can do. The others can't do it because most companies either have the networking stack or they have the security stack. They don't have both. We're the only ones that have both of these combined. And so that, combined with what we've got with Isovalent, with eBPF expertise, is tremendous, is a tremendous accelerant for us as we move forward. So essentially, what we're saying over here is Hypershield fundamentally reimagines all the problems that we talked about, because if you think about segmentation, what we now have is autonomous segmentation.
You never have to write a rule again for segmentation. It can be done through AI autonomously. If patching is hard, you don't have to worry about that anymore because we have this notion of distributed exploit protection, so that you can have a compensating control put in front of a vulnerability to shield the vulnerability while the patch is getting tested and deployed. So you don't have that 45-49-day window where you're kind of exposed because you can have this compensating control within minutes of that vulnerability being announced. This is pretty revolutionary in the market, by the way, you know, in how to go out and protect data centers.
And the third thing is updating is hard, and so the mode that we've actually shifted to with this is, rather than organizations having to worry about updates twice with a change control window and bringing the system down, your updates are gonna happen very similarly to the way that your iPhone updates happen. They're fully autonomous. You actually have two parallel data paths, and, you know, you'll have a primary and a shadow. You do a diff at the end. It's in your production environment. You never have to bring down the system. Your updates can happen on an ongoing basis in a continuous mode, right?
When you think about this, you never have to write a rule again, you never have to worry about a patch being too late, as a result of which you're exposed, and you never have to worry about dated infrastructure within your environment, all done because of AI at the core foundation and a completely reimagined architecture for security. All right, the third area then is extending our leadership in the SOC. You know, if you think about the SOC, not every organization actually has a SOC, but every organization has cyber threats. We have solutions for customers for the first time, for every segment at every maturity level that a customer might have. If you have a SOC that's the most sophisticated SOC, great, we've got a solution for you there.
If you don't have a SOC and you're just getting started, great, we've got a solution for you there as well. As you're moving through that journey, we'll be able to do that all the way. So, that's what we've got as a combination between Splunk and Cisco and what we can bring together. Now, as we all know, security is a massive data problem. So what we need to do is make sure that we can effectively combine and correlate data across multiple control points, so that low-level alerts, when correlated together, that you would have otherwise ignored, like I said, can be graduated to a high-level alert. So what's that journey look like for customers? Well, the customers, the path that can be there for the journey is you could start with XDR.
You also have the SIEM, which is designed for enterprise SOC, kind of, you know, SecOps teams. And then you have SOAR, which is an orchestration and remediation workflow automation platform. With Splunk and Cisco combined, we've got all three of these completely integrated and working in concert with one another, right? And together, they power the future of the SOC. So we wanna make sure that we are the vendor for reimagining the future of the SOC for organizations of any shape and size, of any level of maturity, of any degree of maturity that they might have. So the nice part about this is we can meet the customers where they are in their maturity level, and we can make sure that we can protect them.
So how would this typically start, and what's the sequence and how, how it starts, and what's the upsell and cross-sell motion? You might typically start with XDR, and for many customers, that's the fastest way and the easiest way to start the journey, is through Cisco XDR. Like I mentioned, XDR correlates emails and web and network and endpoint traffic, so that you can not only detect a breach that's happening, you can also predict and prevent them over time. And then once that's done, you can actually prevent lateral movement to so that you don't have any data exfiltration. So that's step number one. You can start with XDR. Step number two, you can actually make sure that you power that you use Splunk, if you were using Splunk.
Now, by the way, many of the organizations that are very mature organizations that have a very advanced SOC, they might start with Splunk. They might not start with XDR. They might start with Splunk. And with Splunk Enterprise Security, you get a broad view of events across your entire landscape, right? And it enables SecOps teams to detect issues and go deep into the investigation. So your quality of detections get better because now the XDR also is integrated fully into the Splunk area. So you actually can hydrate Splunk with XDR incidents. Now, it's built on a common data platform, and what we've been able to do with this is, because of the common data platform, we can add modules to that at a very you know kind of easy a very easy way that customers can ingest.
So, for example, we have Splunk Attack Analyzer that can speed investigation and automatically analyze, you know, kinda complex attack chains. That'll be something that, that just got added because it's actually part of the platform. We also just announced at RSA, Asset and Risk Intelligence, which continuously discovers new devices that are in your environment, and then understand potential vulnerabilities that might happen with those new devices. So what we've got is a great platform and then modules that can be added on top of that platform, so that you continue to have better and better investigations, better and better detections, in the market. And the key differentiation over here is the integration between XDR and Splunk, because they do different things. So XDR has curated data with a limited duration, very high volume data, and Splunk has any source at any scale with extended storage. Right?
And so customers can choose where they wanna start. They can get better detections because XDR can then hydrate Splunk with all of the new incidents that they're discovering, and this creates an upsell and a cross-sell motion bidirectionally. Now, why is this, why is this a problem? Because if you think about lateral movement, and if you think about network telemetry, like Jonathan mentioned, that's a very large amount of data. And even Splunk, it's not feasible to take all of the network data from billions of packets moving through the network and actually put it all into a central ingest, right?
And so what we're able to do is we're able to take just the high-fidelity signals that actually get filtered in XDR and only move those signals into Splunk, and so your detections get better, but you don't have to move all of that networking telemetry and data into a SIEM. And that's actually the thing that's really, really, you know, powerful about this integration between XDR and Splunk. And so as a result of this, what you've seen is the outcomes for customers are very, very potent. You've got less than 15 minutes typically of that it takes for executing a phishing investigation. About 90% reduction in volume of alerts. One of the big challenges this industry has right now is alert fatigue.
There's just way too many alerts, and there's not enough signal in the noise, and we're able to go out and actually make sure that there's much more signal in the noise. And lastly, there's about a 5x faster detection and response to threats than before. So, like, you've got very, very potent outcomes that can be had. Now, the last piece over here is really interesting. Like I said, network telemetry is critical to lateral movement, because where does lateral movement happen? It happens on the network. But that data and that telemetry is too voluminous to go out and be ingested into a SIEM. So to prevent future threats, we need to make sure that we move analytics to the data rather than having the data come close to the analytics, right?
Now, customers can gain visibility in more and more events with more data that have control over it. So if you think about what Splunk has done, is they have this federated model that Gary talks about a fair amount, which is, rather than actually moving all the data from everywhere into Splunk, people are drowning in data. It's very hard to go out and move all of that data. What you do is you actually make sure that you keep the data in the lowest cost location that you want to keep the data, and you move the analytics to that data. And then you can make sure that you can conduct federated search and analytics in that data wherever it is.
So if you happen to have data in AWS and S3, keep that data in S3, and you can now do a federated search based on what Splunk can do with integrating with the data that happens to be in S3, right? So that's, that's what Splunk has already been working on, is this federated model. But then what we've been working on with Hypershield, like I mentioned, is this notion of distributed enforcement points, where you have an enforcement point or Hypershield in every single agent next to a Kubernetes container, in every switch, in every server, in every IoT device that you might have. So the combination of these is very interesting because what you can now do is you can actually move the analytics where Hypershield is running and then only have the verdict go back into Splunk.
What you've done is you've actually reduced the amount of, you know, all of this network data, which you could not put into Splunk, has been distilled down to the verdict, but that verdict can actually be much better at identifying detections and doing better investigations in Splunk because you've hydrated Splunk with very rich, potent sets of data. That's what, that's what can happen only when you combine the two of these things together. You wouldn't have been able to do it, you know, individually, side by side. When you combine all of this, the big differentiation that we have in security is three, is, is threefold: It's the data, it's the network and access to the network, and making sure that security is melted into the fabric of the network, and it's the platform.
And what I mean by the platform is it's easier to keep adding more capabilities into the platform with one common management console, common telemetry, common design language, but also deep hooks into the infrastructure as we go through it. Now, the reality is, is we've not always been in this position that we are in today. Over the last two years, there's a lot of hard work that was done to actually build out a platform with the Cisco Security Cloud. And the reason I'm so positive about our growth potential moving forward is because we've got all of the core building blocks and all of the integrations that we needed to have to fundamentally start reimagining security. And so the way you think about our business is the networking cloud, plus the security cloud, plus the observability cloud, plus the data cloud-...
plus the collaboration cloud, loosely coupled, tightly integrated, so that you can make sure that you don't have to buy all of the things together in order to get, get, get value. But boy, when you do buy them together, there's magic that starts to happen. And so that's the, that's the combined story. We've had more innovation now than we've ever had in the history of Cisco. There are some of the very, very consequential products that have been innovated. The team is amazing. We've actually completely refreshed our leadership team. And the partnership with Splunk is fantastic because a lot of our security people had worked together with the leaders on the Splunk side, and so they know each other, and so that partnership is amazing as well, which also massively has allowed us to accelerate.
So from day one, we've actually had continued levels of joint integrations that we've been doing between Splunk and Cisco, and those things have only gotten better. With that, I wanna get Gary up so he can talk a little bit more about why Cisco and Splunk are better together. Gary, come on up.
Thank you, Jeetu.
Thank you.
It is great to be here today. I'll be covering this in two capacities. The first is, I own integration for Splunk, and so I'm gonna share a little bit of detail about what we bring to Cisco, how we're approaching integration, and the value and growth that we believe we can deliver. And the second part, I will spend just a few minutes on my new role. So it was recently announced on the earnings call that I have taken on the President of Go-to-Market role within Cisco, and I'll just share a few of the insights I have in just the few number of days I've been in this role. But I'm super excited about that and the opportunity to help us drive better customer outcomes every single day. So let me just dive in.
When we look at what Splunk brings to Cisco, there's a few things that's probably important to level set. If you didn't watch Splunk at the very end before the acquisition happened, we announced our final earnings call, where we had $4.2 billion of annual recurring revenue. We grew that top line 15%. We generated $1 billion in revenue, and the momentum that we had as we closed out our fiscal year, that momentum has stayed very consistent as we moved into Cisco. So one of the critical things that we deliver is a very important growth engine for Cisco's software business. And as you heard from Jeetu, we have an incredible number of important connection points that we think we can sustain that growth or even accelerate that growth over time. Splunk comes in with a market leadership position.
You saw the Magic Quadrant, and the pace of innovation is something that we've been focused on at Splunk as an independent company, but it gets accelerated more as we come into Cisco. I joined Splunk roughly two years ago, and my number one priority was: how do we reestablish ourselves as the clear market leader, but really delivering amazing innovation that helps our customers get more out of Splunk? And that's what we've been focused on, and that will continue. And that roadmap that we put in place years ago is now continues to be in place, and we're delivering on that. One of the smart things that Cisco did in thinking about and contemplating this acquisition, because it was contemplated on growth, is all of our roadmap has stayed consistent.
And there was incremental funding given to Splunk so that we could drive the integration capabilities necessary, but not disrupt the core roadmap and path that we were on, and that's played a really important role. On the observability side, one of the very exciting things is we're bringing the broad Splunk observability portfolio, combining that with the broad capabilities within Cisco, and delivering something, frankly, that I don't think anyone in the industry can even come close to. And interestingly, one of the first decisions that got made was we moved all of the AppD development and full stack observability within Cisco into Splunk. That happened on day one. We moved all the sellers right after that into the Splunk selling organization.
So we very quickly made a set of product decisions and did the important alignment to ensure that we could go hard after this market, delivering the right level of innovation and really capitalizing on what we believe to be a tremendous opportunity ahead of us. We think the leadership that we have in security, combined with the interesting assets that we have broadly in observability and the market leadership position we've established there, only gets better within Cisco. We bring with that some amazing relationships in the Global 2000, focused on helping those organizations really drive digital resilience.
Now, this is a mission that we've been on for a very long time, and we, when we think about digital resilience, it's really about keeping systems up and running, and ensuring that they're well protected, and that they stay up and running, and they don't experience any form of outage. That fundamental view has spanned our world, and one of the interesting aspects of this is there is fundamental convergence between what's happening to drive visibility on security and what's actually happening more broadly, because all the data that you need is the same. You want that visibility across both of those environments, and Splunk has really been pioneering that convergence, and we think it's incredibly important. If you look across the broad Splunk customer base, what you'll find is that we are so focused on making sure that mission-critical applications stay up and running.
It was interesting, I was in Davos in January, and I get there, and Chuck says, "Oh, there's a meeting I want you to go to." Like, we weren't combined, and we hadn't done one sales meeting, not, not any sales meetings. So I show up, I didn't even know who the meeting was with. And I go into the conference room, Chuck's there, there's a bunch of people there, and in walks the CIO of a major bank, and I know this person. And literally, his whole only message was, "Chuck, whatever you do, don't screw up Splunk. Whatever you do!" That was the message. Because we are so mission-critical to how they operate their business, not in security, but how they have visibility across their entire application environment, and that's how they understand what's up and running across the broader bank.
And I think that if you then take the next step and think about all those amazing connections that we have within the Cisco product line and the power and visibility that we can deliver in a combined way, that, that only gets so much stronger. One of the great things about Splunk is our ability to scale, to scale in crazy hard ways. If you haven't ever followed the history, what, what made Splunk unique in the early days and continues to make us unique is we can adjust machine data, that it's just a glob of data, and make something out of it and do analytics on it. So no schema. So think schema on read. We figure out what your data is and then help you do analytics on it without having everything nicely in rows and columns, 'cause machine data, data is never that way.
We can do this at mass scale, like insane scale. I'm shocked... I spent a lot of time with customers. I am shocked at the amount of value people derive from Splunk because of the crazy complex workloads that they do and the kinds of things that Splunk enables these customers to accomplish. I was recently with a large restaurant chain, and they've been a Splunk customer for a long time. They use us in security. We are fundamental to the SOC. Great, but every single restaurant is monitored by Splunk's Observability Cloud. So the people walk in the restaurant in the morning, if anything is down, the machines that make the food, which I won't describe, the machines that make the food, they monitor those because the person walking in the door has no idea how to fix it.
The POS system, if the cash register's down, obviously they're down. They use Splunk broadly across their environment to understand if in Wichita, the person who, who runs the operation there is gonna have issues before they even get into the restaurant itself. So that's the kind of crazy capabilities that we offer and deliver that makes Splunk so mission-critical to everything we do. Now, one of the big themes we're seeing in the marketplace is, and Jeetu described this, there's just this era where people have bought different tools for different things. We've been fundamentally helping customers work through, how do you rationalize the numerous tools down to something that you can really drive the right strategic outcomes? So I'll give you an example. So there's a major transportation company.
The CTO at that organization told me they have 126 "monitoring tools." 126. Like, when you think about the fragmentation you have, the duplication that you have, the issues that you're gonna have with that kind of environment, we've been very focused on driving that broad tool consolidation. Then we think about flexibility from a deployment model perspective. We live in a multi-cloud, hybrid world, and you can make Splunk work across that complex environment. We see a lot of customers that start on-prem, as you'd expect. They decide to have some number of workloads in cloud. They can run Splunk both places. They can make all of that connection work seamlessly. We have been traditionally in AWS and in GCP.
We announced last year at our user conference that we had a relationship with Microsoft, and we were moving to Azure as well. We will confirm that that product is gonna be out in the second half of this year, and so we will add Azure to that list, so customers can truly decide where they wanna be. And that flexibility provides an immense amount of power in the very complicated world with data privacy issues, with data residency issues, with data gravity issues. You have a tremendous amount of choice, and that's, that flexibility is what has made Splunk unique for a very long period of time. And, you know, I've given you a few examples, but is really one of the most enriching things about my role at Splunk has been just hearing how customers have leveraged Splunk. So there's a large e-commerce company.
They manage Splunk themselves, so this is not in Splunk's cloud. They manage it themselves. They ingest 5 petabytes a day. It's all security information. Why in the heck do they have that much data? It's because their threat or attack surface is huge. And so what they're doing is they're gathering all that data that gives them insight across their broad attack surface. They need to know at all times what's happening across that broad environment, and it's 5 petabytes a day. A day. There's another customer, who's not on this slide, that I meet with regularly. They're on-premise. They ingest 24 petabytes a day. So these volumes give you an indication of the power of data analytics that you can do to be able to see across this vast sea of data and drive decisions that ultimately drive uptime or drive better security posture.
The other two customer examples here, one is a large telecommunications company. We're broadly deployed across all their critical systems to help drive uptime. And the final example is a global sports apparel company that started using Splunk for the observability of understanding all their applications, their whole digital footprint, and we recently won their SIEM. Security was the last holdout, and they basically took out a competitive solution, and they put it in Splunk, and they're seeing all the benefits of having single product across these broad environments. But one of the things that I think you'll find is that with these kinds of relationships that we have, we fit very well into the Cisco world. If you, if you look below, you'll see some of the dollar values.
These are very large transactions because we're deeply installed into the customer, and the value that we're delivering is incredibly important to the outcomes that they need. So as we think about better together, I want to just comment on integration strategy, and I think we've made some very good decisions. So one, we were really aware of the other large acquisitions that have happened in the industry and the fallout that has happened because of those. And so just a number of... It may seem basic, but it's really important. Every customer that's been a customer of Splunk is dealing with the same person. That's not true of a lot, a lot of the large-scale integrations that have been happening. We think that's really important. Customer experience matters. As we move Splunk into Cisco, the only person that changed jobs, roles, or title was me.
Literally, one person in the whole company. Everything, everybody else stayed the same. And so we did that so that we could ensure that we could continue to drive this path of innovation that ultimately would result in growth. And I think Scott and the team have been really clear that this was a growth-focused acquisition, and we think about the top line every day. Will we see cost synergies? Of course, we will. And we've been moving incredibly aggressively on the things that are obvious from an integration standpoint that give us those cost synergies. So what was the first thing we integrated? We were done by May fifteenth on integrating legal. And you may say, "Well, that's simple," but we were done in less than six weeks. It's really fast. And do we get cost synergies? Sure.
Our chief legal officer, who's amazing, he didn't have a job. It's okay. Like, we know these things happen. And so it gives us this opportunity to, again, be 100% focused on growth while we're driving integration that's obvious, that ultimately will give us some cost benefit along the way. As we look then about in the broad context of how do we, how do we drive the value, we've been very focused on finding these connections across Cisco and Splunk, where we could deliver great outcomes for customers, and they would see value from the work that we're doing together. So we've really taken advantage of the network, the cloud data, endpoint data, and bringing that broadly into Splunk.
One of the first things we did was said, all of the technical add-ons, or we call them TAs, within Splunk that support Cisco products, we looked across those and said, "Hey, we want to make sure we're investing appropriately in those." So it's easy for loyal Cisco customers to get benefit from Splunk by bringing data easily into Splunk. We looked broadly at the Cisco security portfolio, as Jeetu described it, and there's tremendous opportunity to drive connections. Announcing support for XDR alerts to come into Splunk Enterprise Security is an obvious and big important win. And these are things that ultimately we believe will have tremendous value to customers, but also continue this path of growth that we've been on and accelerate it as we move more broadly within Cisco. And everything that we're doing, basically, Cisco amplifies.
This idea of visibility across this broad multi-cloud hybrid environment only gets better with the data that we derive from Cisco. We've been on a mission around really showing customers what the SOC of the future looks like. And with the capabilities that we're seeing across the Cisco product line, we can derive better insights for customers as a result of that. And I'm very excited about what we can do in the observability product line and the outcomes that we think will be critical there. You know, it's interesting, we had an executive session last night, and I was talking to one of the customers that I've known for a long time. They're longtime Cisco customers. They're longtime Splunk customers. And he just said to me, he's like: "I'm so excited about observability, Gary. We'd had a lot of work in Datadog.
We're actually moving to Splunk. We've made that decision because of the opportunity that we see across all these capabilities." So when we look at each of these individually and we think about how we power the SOC of the future, Jeetu covered most of this, but I just want to take a minute. We really think about how do we power detection, investigation, and response. We get the benefit from the detections that Jeetu described in the XDR solution. We bring that together with our SOAR solution, again, security, orchestration, automation, and response. While it sounds kind of boring, it's really, really important to customers. The faster that we can identify and respond, the better off we're gonna be. And secondly, and you guys all understand this because you see it written up every day, there's a new set of rules that came out of the SEC.
We have four days to report if there's ever any material breach. Determining materiality is the best thing that Splunk can do, because people will use Splunk to determine what was the extent of a breach. And if you haven't made that investment in your public company operating under these SEC rules, you'll have to do that. And we think that's another very important growth catalyst as we come together with Cisco. We've made a number of announcements. You will hear more announcements, and we're much very much focused on this ground game that I've been describing. I think that our customers will measure us on innovation, and I spent a lot of time at RSA and in other venues talking about how we come together. Pace of innovation matters. We will be judged on: Are you actually delivering on your roadmap?
Are you delivering good outcomes to us? And I feel very confident that we can continue on this path of innovation and actually accelerate that with the broad power of Cisco. And on the observability side, it's the same thing. There is so much opportunity here, and people are rethinking how they've looked at this, and I think Cisco brings an entirely unique lens. Our ability to see all the way through the network, all the way through the infrastructure, is so absolutely critical, and no one else can deliver that other than Cisco. So get my slides to work. That visibility across, as Jonathan described, unmanaged and unknown networks. How do you do that? You do that through the power of ThousandEyes, integrated into our broad observability suite.
Being able to have visibility across traditional application architectures as well as microservice-based architectures, spanning on-prem and cloud, which almost every single customer has today. But then leveraging the power of AI and driving to enterprise scale is what will uniquely differentiate Cisco in this market, which we think will have tremendous growth opportunity over the next five years. Again, with observability, we have a whole set of announcements that are happening at Cisco Live and at our user conference next week, and it really focused on how do we bring these, these components together that bring the best of Cisco and the best of Splunk and put it in the hands of our customers. So when I joined Splunk, one of the things that I was focused on is how we could simplify and drive more predictability in how we went to market.
We really focused on a very simple customer maturity model that ultimately ended up in very much of a land and expand. With Splunk, what we've traditionally done is land and security, meaning a customer adopts Splunk to get better security outcomes. They obviously can use this for their SIM, but they oftentimes will just do data analytics within the security department, leveraging Splunk. We will expand in security to some of these premium applications that Jeetu described, whether it is Attack Analyzer or Asset Risk and Compliance. These are extra paid-for modules that drive incremental value to us and continue to create growth. We focus on cross-selling observability. So help that customer who has success in security, show them value of how they can extend beyond logs, get to metrics, get to traces, and a broader observability portfolio.
And then what almost always happens in great customers is Splunk is like something that just spreads because people love the flexibility to be able to do analytics. And so we then get used in all kinds of interesting use cases. So some examples, it's not uncommon in the banking industry to see Splunk very embedded in the SOC, but then in the fraud department, they begin to use Splunk. Why is that? Because it's very similar to what people want to do in security. You're bringing in a set of data sets from a variety of systems, you need to cross-correlate, and you need to leverage all that security data that you, you got in Splunk, because oftentimes threat actors are the same. So that's a very natural use case extension that drives incremental value for customers.
So we believe that this focus on customer outcomes only gets accelerated in the Cisco world. When we look at the low-hanging fruit and where we're focused today in driving more business through Cisco for Splunk, international market expansion comes to mind. And why do I say that? So if you think about Splunk's business, you can go back and look, but we were roughly two-thirds domestic, one-third international. And for a company that's doing $4.2 billion in ARR, you would typically have more international revenue at this stage. The reason we don't, frankly, is we had so much success in the U.S., we were just going after easy revenue. No more complicated than that. So where are we underserved?
Where the opportunities are more deeply in EMEA, so we have good presence across the continent, but if you go into many countries, we're still light in terms of feet on the ground, driving business. Middle East, we're just getting started. I was in the Middle East in the fall of last year. I came back so excited because there's tremendous opportunity that will obviously be accelerated with Cisco. Latin America is tiny in our world. It's absolutely tiny, and we're just getting the teams connected. There's incredible amount of momentum and enthusiasm and excitement right now. And then if you look at Asia Pac, we, we've had presence in most of the markets, but tiny presence.
So as an example, Chuck and I were sitting, we were somewhere, he's like: "How many people do you have in India, Gary?" I'm like: "Maybe three?" Like literally, like literally nobody. And just our ability to connect into the Cisco machine that exists in that market is huge. Japan, we've got some amazing customers. Good example, in our last earnings call, we talked about a really important government win in Japan, but it was one of our first government wins. Cisco has been there forever, has deep relationships across all markets. We can go so much faster with the machine that is Cisco. Cross-selling is an obvious one. I think there's opportunity to cross-sell both ways, and then the channel really represents a huge opportunity.
I spent, in my few days that I've been at Cisco, some amount of time with channel leadership, really understanding what the opportunity is, and we're very early there because our channels are different. And so our ability to cross-train the Cisco channel and really drive growth there is just a tremendous opportunity that we'll take advantage of. So I want to shift gears slightly and just talk about my new role. So I own Go-to-Market , which means marketing to sales, and the thing that I'm excited about is I'm going to spend my time focusing on ensuring that our customers get great outcomes, and that Cisco is a company that is easy to do business with, and that we simplify things so that we can really capitalize on this inflection that I believe we will see with AI.
Sitting through all of the keynotes, I think you get the sense and the feeling of the power of innovation that we're driving within the company, and it's my job in this new role to ensure that we're getting that innovation out into the hands of the customers and out in the market and ultimately driving growth. I believe that this focus on simplifying our model of Go-to-Market will ultimately unlock velocity and efficiency that we think will be important for the next 10 years. So with that, I'm gonna turn it back to Sami.
All right. All right, thank you, everyone. We are gonna take a 10-minute coffee break, and when we get back, we're gonna hear from Scott Herren, and right after that, we're gonna get into the executive Q&A panel session. So with that, we'll see you all in about 10 minutes, which is about 3:45 P.M. local time, and then we'll pick it up from there.
... CFO of Cisco, and after that, we will be going right into the Q&A session. With that, here you go - here comes Scott.
Thanks, Sami. All right, thank you. There's a reason I'm last, 'cause I know it keeps many of you here, and I wanna thank you for your time, especially those of you who came in person, 'cause I know it's a commitment, and I know many of you are on a red-eye tonight to get back to New York for an event that's happening there tomorrow. I really appreciate it. I hope you had a chance to not just see the keynotes, but also interact with some customers and partners while you're here. I think that's the value of being here in person. For those of you online, thank you for your time as well. We're all busy. It's not easy to carve out time to do something like this, and I don't take it for granted. All right.
Chuck talked about us being positioned for growth, so I'd like. Let me run through that. First, I am gonna be making forward-looking statements. I'm not gonna read the slide. I think Sami did a nice job of that earlier. But we're making those statements subject to Safe Harbor, and no change in guidance. I put this up again. This is something I just wanna make sure that it's clear, no change in the guidance that we gave you for the current quarter that we're in, fiscal 2024 Q4, fiscal 2024, or for the full year. So just put that out there so that there's. I think everyone's clear on that front. All right, let me recap. Before we start to look ahead, let me recap.
When we were last together was September 2021, and we put three big targets out there. We wanted to grow the top line at a 5%-7% rate. We wanted to grow the bottom line at a 5%-7% rate. If you remember, I talked about balanced, profitable growth. Then we wanted to get to a point that where we had at least 50% of our revenues coming from subscription models. So how'd we do? We're only... That was, by the way, through the end of those targets were intended to run through the end of fiscal 2025. Obviously, with the acquisition of Splunk, it made sense to get back together earlier than fiscal 2025 so that I could update you on the model longer term.
On the top line versus the 5%-7% we expected, if you look at the average revenue growth since the end of fiscal 2021 to where the midpoint of our guide for fiscal 2024, the average growth has been 4%. So didn't quite get there, and there's a couple of things when you go back and say, "Why not?" Right? What didn't perform to expectations? There's a couple of areas. I think in collaboration, in particular, the product team did a terrific job of building the best product in the marketplace. I spent a lot of time working and using some of the other products that are out there. I think we built the best product, and the competitive dynamic changed with some of the pricing bundles from one of our bigger competitors, specifically on the meetings piece.
That's been a headwind to our growth. Instead of something we thought was gonna be a big grower, that's been a bit of a headwind. We're making really nice progress in Collab, as Jeetu just walked us through, with Cloud Calling and Cloud Contact Center and with devices. But that one, that didn't perform to our expectations when we laid out those goals. And the second is it's more secular. There's a couple of customer segments that we sell into, telco and cable, that have had a tough run over the last couple of years, and obviously that's been a headwind for us as well. So as we went back and picked apart, you know, where did we come up short relative to that goal? That's. There's always a lot of puts and takes. Those are the two biggest drivers of that.
Non-GAAP earnings per share. If you look again at the average growth in non-GAAP earnings per share over that same time frame or through the end of fiscal 2024, it's just short of 7%, so toward the high end of that range, despite coming up below our expectations on the top line. And that's really a function of focusing on driving operating leverage, focusing on the things that we can control. Obviously, we couldn't control the supply constraints through that whole process, but what we could control was our own spending and our own discipline, and we did that. And so we've been delivering operating leverage with, again, over that same time frame, average growth of earnings per share at 7%. And the 50% goal of total reported revenues, what percent of them came from subscription models?
We actually hit that in our second quarter and then, of course, stayed there again in the, in the most recent quarter, our Q3. So we got there roughly 6 quarters ahead of what our expectations were. So that's against the goals we laid out before, because I'm gonna talk longer term today, and I want you to see how we did against our previous goals. A couple of other key points, subscription revenue growth. So this is total subscription revenue. You see the green bars are service subscription revenues. The blue bars are product subscription revenues. There's a significant, significantly different growth rate. Both are growing. I went back to fiscal 2016. That's the time... and on the next slide as well, that's the time that we made the move. Cisco made the move.
It's when Chuck came in as CEO, and we made the move to focus in on driving more subscription revenue. And you can see the compound annual growth rate of our subscription revenue is 24% over that time. Obviously, there is a benefit to adding Splunk in the fiscal 2024 column that would take that CAGR from 24% product CAGR down to about a point, down to about a 23% CAGR, if you netted Splunk out of it. So nice growth in total subscription revenue, up to just short of $28 billion, and very nice growth in product subscription revenue, which has been the big focus. If you then say, "Okay, what is that equal to on total software revenue?" And I think this is also an interesting slide that is not intuitively obvious if you've been around Cisco for a while.
We've built one of the biggest software franchises in the world, with software revenue just north of, at the end of this year, just north of $18 billion in software revenue. The green bars are perpetual software revenue, and that's by design, right? As we've moved to more subscription models, we've pivoted away from selling more perpetual license revenue. That's come down, and in spite of that, you see total software subscription revenue, growing at about a 19% compound annual growth rate, going back to fiscal 2016. Again, there's about a point benefit from the addition of Splunk, so it'd be about an 18% growth rate if you netted Splunk out of the numbers.
So we've built by, by putting in the focus and by driving not just an understanding and the motivations of our sales team and our channel partners in building this recurring revenue model, but also focusing on adoption, because adoption leads to renewal, and renewal, when you've got this much, subscription revenue, renewal becomes critically important to us, and we've made good strides on that as well. So just want to recap where we are at this point. A couple of key metrics, as we look ahead and as we built that out, and I think these are both important. And again, this is still looking back. Annualized recurring revenue, so ARR, you can see at the end, and now I'm just using the Q3, the quarter we just announced. At the end of Q3 2024, total of $29 billion of annualized recurring revenue.
That includes a little more than $4 billion from the addition of Splunk. But again, you see the growth rate of that. And what's important about ARR, when you think about it over time, this is the annualized value, right? So there's $29.2 billion that we get to renew next year and new sales stack on top, and you get to renew that the following year because these are annualized values and new, new sales stack on top again. So not only does it give a durability to our revenue streams, again, you know, with the assumption that we continue to drive renewal rates, that is not always durable. It's. You get greater visibility from that as well.
It's visibility to slightly more than half of our revenue at this point, but it's a, it's a very good progress, and it's something that will dampen volatility over time as we continue to build that out. I think the second major metric that I want to make sure you, you think about and build in as you're thinking about, "How do I model out Cisco for the future?" is remaining performance obligations, right? RPO. You hear me talk about this on every earnings call, and this is basically sales that we've transacted that haven't yet accreted into the revenue stream, into reported revenues. So it's two, two big drivers. Deferred revenue is a big driver of that. That's, you know, I sell a subscription for 12 months, one day at a time, that comes out of deferred revenue and goes into the revenue stream.
The remainder sits there in the balance of deferred revenue. That's a big driver of RPO. The second is what's called unbilled. Unbilled is contracted sales with a customer, but let's say I sign you to a 3-year deal, and we agree that I'm gonna bill you annually instead of billing all 3 years up front. I bill you the first year. It does what I just said. It goes into deferred revenue and bleeds back into the revenue stream a day at a time. Those other 2 years that I haven't yet billed sit in what's called unbilled revenue. It's a GAAP metric that we have to report every quarter, and you can see at this point, we've built up just short of $39 billion worth of sales that we've transacted.
that are going to come into our revenue stream over time, but haven't come into our revenue stream yet, 'cause they're sitting in deferred or they're sitting in unbilled. And that's a nice, this is probably one of the better leading indicators that we have of how the, how the company will grow over the next twelve months. Of that almost $39 billion of RPO, you can see the green bar, $20 billion of that is what's called current. Current means it's gonna turn into revenue in the next twelve months. All right, so the next twelve months, we've got $20.1 billion coming off the balance sheet in RPO into our revenue streams. Really important metric. All right, the part I think everyone came here to see. Let's talk about looking ahead. And I want to start by just recapping.
I think you've seen all of these numbers now between Jeetu, Gary, and JD of what our TAMs are, but I wanted to capture them all on one slide. So you can see this is the blue bars are current markets, the green bars are expansion markets that we just talked about. Either markets that we're either in, in a very small, just beginning to get into those markets, or we have plans to get into those markets in the time frame here, in the next three years. That's what's in the expansion markets. The blue bars are markets, the existing markets that we're in today. And when you sum those up, $575 billion worth of opportunity just in the markets we're in today or markets that we plan to enter in the next three years.
You saw this next slide from Chuck earlier this morning. When you add the adjacent opportunities, so things that we could get into, think of AI software as an example, areas that we could get into that are adjacent. We're not in today, and we don't necessarily have a concrete plan to get in in the next three years. That adds up to just short of $1 trillion worth of opportunity by the time you get to fiscal 27. These are calendar year fiscal 27 TAMs. So we're not opportunity-constrained as we look ahead. Before I go to the next slide, fiscal 25, I think I gave you some modeling points. I know I gave you some modeling points on our most recent earnings call. There's no change to that.
I'll give you a proper guide with all the detail you're accustomed to for fiscal 25 when we get to our Q4 earnings call in August. So I don't plan on updating the modeling points that I gave you. I think looking at the way the models worked out, I think most people really understood what was there. So let's talk about, instead of talking about fiscal 25, what happens beyond that, right? Because in fiscal 25, as you recall, there's a lot of moving parts. There's the... you know, when you're talking about growth rates, you have to compare to fiscal 24. Fiscal 24 had about $4 billion of excess backlog included, mostly in our first quarter of fiscal 24. When you look at fiscal 25, of course, there's the addition of Splunk.
For a little more than a half a year, that's inorganic growth, right? Then we start to compare to a quarter where we had Splunk in. So what do we see now for fiscal 2026 and for fiscal 2027? These are annualized growth rates. We see product growing in this 5%-7% range through time, and I wanted to give you the breakout because this is one of the things that we hear constantly, is what do you think is happening in the core networking business and the growth there? We've talked about a lot of the tailwinds that are there. I think relative to some of the modeling points or some of the targets that we put out in September of 2021, one of my learnings is it pays to be a little more conservative, and we're putting out these long-term targets.
So we haven't been super aggressive on what we've laid in for AI Infrastructure opportunity within that, that networking line, but there's, there is some in there. So you see networking as a subset of total product. We see that growing somewhere between the 2%-5% range in fiscal 2026 and again in fiscal 2027. Obviously, security and observability off a smaller base, but growing significantly faster over that time. Services will continue to grow in this 2%-4%. These are revenue growth rates, will continue to grow in the same range that you've seen it over the last several years, kind of 2%-4%. And there's a natural headwind that we've talked about, but I'll say it again, on services revenue growth rate.
As we have moved more and more to SaaS models, SaaS model, you don't pay one price within a SaaS model for the product and a separate one for the services on that, for the maintenance of that software. You pay one price, right? And that all gets categorized as product revenue when you move to a SaaS model. So there's a little bit of bucket shift going on between what normally would have been services and is now coming into the product line. As we've moved to more SaaS style, more recurring revenue models, it's a natural headwind to services growth, but this is the growth range that you've seen from us over the last several years. Services growing in that 2%-4% range.
So total revenue growth that we see for both fiscal 2026 and then again in fiscal 2027, somewhere in that 4%-6% range for those two years. What does that equate to in terms of earnings per share growth? And I wanted to give you, again, there's-- I know there's been a lot of questions on this. I wanted to give you a little insight into that. We see with the top line growing each year at somewhere between 4% and 6%. We see the bottom line growing each year, somewhere between 6% and 8%. So you've seen us deliver operating leverage over the last several quarters. We've done it on a, on an average growth rate from the end of fiscal 2021 through our projection for the end of fiscal 2024.
If you look at the reasons that we see that growing over time, gross margins will have upward pressure on them. Obviously, as we continue to mix more software into our gross margins, software comes with a higher gross margin. That's a benefit, and we'll continue within that, within our overall pricing and cost discipline on the on the cost of goods sold line. We'll continue to keep pressure on that, right? So expect to see gross margins increasing through time. Some of that will fall through to the operating margin. We'll continue, obviously, to have the same level of discipline, financial management discipline that we've had, which is what's allowed us to deliver operating leverage at a time when we weren't quite getting to our expectations on the top line.
We'll see a productivity gains from AI as we go about implementing AI kind of across the product line, that some of the low-hanging fruit areas, like CX, is one of the first areas that we'll start to see some of those productivity gains. That will help. Obviously, we'll see, and Gary alluded to this in his previous conversation, we will see cost synergies over time through the Splunk transaction. That wasn't the motivation for the deal. The motivation was much more about top-line acceleration, but we will see cost synergies out through time. All those things contribute to us seeing the bottom line grow, let's call it 2 points faster than the top line. If the top line is 4%-6%, we see the bottom line, in both fiscal 2026 and 2027, growing somewhere between 6%-8%.
How about beyond that? All right, so, so you've—we've talked about 25 and some of the disruption. I think you've got the modeling points on that. I've given you a good view of 26 and 27, so that's a three-year timeframe from where we are. What happens after fiscal 27? And we've talked about a lot of these trends. You heard a lot this morning, both in the main stage and from JD, around AI Infrastructure revenue growth. We'll continue... We're investing fairly heavily. Chuck gave a stat that I, it bears repeating. If you look at our R&D spend at this point and say, how much of that is either driving AI Infrastructure, R&D, driving AI being built into our products? G2 gave several examples, and so did Gary, of where we're building AI into our products today.
What we're spending on security overall, what is that percent of our R&D budget? It's roughly half of our R&D budget is being spent. Those are the highest growth areas as we look ahead, right? So we'll have a longer-term tailwind from AI Infrastructure, from accelerating Splunk sales through the integration, both the international footprint that we've talked about, which, by the way, has an opportunity not only as we accelerate Splunk through the international channels, but the integration of the two products together will end up being a tailwind more broadly to our overall security portfolio at the same time, and of course, going through our channel.
Security being infused into the network fabric, I think the phrase that G2 used is, "We're melting security and infusing it into the fabric." It's an interesting visual, but that will continue to... I think that's a little bit of a longer-term trend for us, right? Because that requires work both on the security side and on the networking side, and so that, that trend will be a tailwind to us as we continue to roll out Hypershield, and we see that being implemented more broadly. That's a tailwind that will continue well beyond fiscal 2027. Security business has done a nice job. We've pivoted a lot of resources into security. Internally, we've pivoted a lot of investment in terms of acquisitions into security, not just Splunk, but Isovalent and several other acquisitions in that space, and we're seeing that bear fruit.
We're seeing the security business turn around even before the Splunk acquisition. That will continue through time. Core networking demand will normalize, right? We said we see the inventory that's out there, the amount that we cleared very quickly as the supply chain, kind of the constraints cleared up, and we got a lot of product into our customers' hands. We see that being largely worked off by the end of this quarter, right? When we get to Q1 of fiscal 2025, remember, we've got a tough compare, 'cause that quarter compares to Q1 of 2024, and in Q1 of 2024, we shipped about $4 billion worth of excess backlog out the door. That's not gonna repeat. Beyond that, we see the demand for core networking really normalizing and normalizing out through time.
So as we lap that point, that's where you'll start to see the seasonality that you would normally expect. And I talked about the power of revenue stacking. With more than 50% of our revenue now coming from a subscription model, I've got this many subscriptions, this much ARR this year, I get to renew that next year, new sales stack on top. Renew that the following year, new sales stack on top. So that subscription model by itself will continue to drive growth as we continue to drive adoption and renewal rates out through time. So I think longer term, there's a fair amount of opportunity, even beyond fiscal 2027, for us to continue to accelerate growth and these are the areas that we're focused on.
This is why you see us pivoting as much R&D as we have into these spaces. These are the long-term growth drivers, even beyond fiscal 2027 of our business. Okay, cap allocation. We obviously generate a lot of cash, and this is one of the ways that I think we can drive shareholder value for everyone in the room. So let's talk about our cap allocation. First, talk about the priorities, and these are unchanged. These are the same cap allocation priorities you've heard from me, going all the way back to September of 2021, but on every earnings call as it's come up. So it's first and foremost to support the growth of our business, both organically and inorganically. It's second, to continue to support and grow our dividend payment, and I'll show you on the next slide what that's looked like through time.
So expect to see us continue to do that as a priority for cap allocation. The third is share buybacks, and it's one way—not just offsetting dilution, but you've seen now for 5 consecutive quarters, you've seen us have a steady share buyback rate at about $1.25 billion per quarter, so a $5 billion annual run rate. That goes beyond, obviously, just dilution offset, and I'll show you what the benefit of that is, and we'll have an opportunistic overlay on top of that. So that's—those are the top three, and then, of course, beyond that, with any excess cash, we'll look at: Does it make sense to return this, or does it make more sense to delever? And I think that's an exercise that we'll go through as we get there.
All right, so that's the cap allocation process and the priorities, which are unchanged. What does that look like? This is what's happened over time. So going back from fiscal 2019, you see the dividend, steady dividend growth annually from fiscal 2019 through the year that we're in right now, fiscal 2024. We've retired about 300 million shares net, retired 300 million shares. Obviously, we bought more than that, but we issue shares through our equity plans, too. So a nice downward trend on shares outstanding. And free cash flow, you know, we've got a commitment of 50% of free cash flow being returned, and you can see over the same timeframe, in the last 6 years, we've returned almost double that.
We've returned about 93% of free cash flow over that same timeframe, obviously, from driving both the dividend and the, and the share buyback. So I think we've had a strong track record on, cap allocation and, and the way we've thought about how we can use that to drive shareholder value. Summarize. Is everybody awake out there?
Yeah.
Metrics are normalizing, right? We've come through what's been an incredibly difficult period with the supply chain constraints, and then the lead times extending, which then led to a lot of demand being pulled backward, and then working through those constraints, getting that pushed out, and then our customers needing to implement that. That's created a lot of perturbation. It's created a lot of, you know, spikes and valleys. We see that normalizing at this point, and as I said, Q1 of 2025 is the first quarter where we don't have one of those abnormalities to compare to in the prior year. So as those normalize, both the demand and the P&L metrics will begin to normalize. We're investing. We're investing in the biggest growth areas that are there.
When you look at both the what I just talked about in terms of roughly half of our R&D budget going into cloud, AI, and cybersecurity, but also some of the investments that we're making off the balance sheet. You saw some of those investments announced this morning in the billion-dollar AI investment fund that we've got there. Continue to be focused on consistent execution and driving disciplined financial management, which we've done, and that's what's allowed us to continue to deliver operating leverage at a time where revenues have been kind of up and down like this. So you expect to continue to see us do that. And then, with the cash flow that we talked about and our cap allocation policies, focusing on driving shareholder value with the way we allocate capital out through time. So there's a lot...
We've talked about a lot. We've talked about a lot of the tailwinds that we have longer term, kinda giving you a sense of, we'd already talked about fiscal 2025. I think everyone's got that. But what happens after that, when you don't have all of these abnormalities in one year? You get a sense now of what we see in the top-line growth for 2026 and for 2027. Talked about some of the tailwinds even beyond 2027 and what that looks like, and then continuing to drive operating leverage at a rate that's a couple points faster than what we see in terms of growth on the top line. Top line growing 4%-6% in 2026 and again in 2027. Bottom line growing 6%-8% in 2026 and 2027. That's what we see out through the future.
I wanna thank you again for your time, particularly the people in the room. I know this is not an easy trip to make. This is not my favorite place on the planet to be, and, and it's not because it's 108 degrees here. Just, like, this isn't my favorite place. I really appreciate that you allocated the time to be here with us. I hope you've had a good experience. I hope you've had a chance to meet with customers, to meet with partners while you're here. For those of you watching online, I appreciate your time as well. I know we're all busy, and it, it takes time to, to do this, and I appreciate you spending some of that time with us.
What I'd like to do then, Sami, is let's go ahead and move to the Q&A and bring up the rest of the panel, bring the rest of my peers up. Thank you.
All right, thank you. And, we're gonna have all the presenters from the ELT join us on stage. There are a couple of mic runners in the field, so whenever you raise your hand, just wait for the mic runner to approach you with the microphone. We also have a couple of other members of the ELT also available to answer questions, including Mark Patterson, our Chief Strategy Officer. We have Fran Katsoudas, our Chief People, Policy, and Purpose Officer, as well as Liz Centoni, our Chief Customer Experience Officer. So, with that, I will take a seat, and-
Doesn't look-
We'll start off with,
Doesn't look like there are any questions.
Sorry?
Doesn't look like there are any questions.
No, no, no. Whenever you guys are ready, you can raise your hands. All right, let's start off with Tim Long over there.
Thank you. Appreciate it. Tim Long, Barclays. Maybe I'll throw two at you. First, Chuck, you know, you're gonna start with this, the AI. If you could talk a little bit about the InfiniBand Ethernet use cases for kinda the full system solutions for Cisco, and particularly in light of NVIDIA, you know, now becoming more of an Ethernet competitor. So can you talk a little bit about, maybe differentiation and, and now that this is just a different landscape there? And then the second one, for whoever would like it, there is, Cisco does very strong with governments around the world. A lot more talk about sovereign AI, so maybe if you could touch on that as an end market where, maybe Cisco has, you know, outsized opportunities relative, to other players, in the industry. Thank you.
Jonathan, I'm gonna hand those to you.
All right. Great. So, two things: One, if we're talking about InfiniBand versus Ethernet, I think we all believe, and I think the industry believes, for the most part, that Ethernet is going to be the winner. I think it's all about timing. And your point specifically about differentiation in silicon, we have a very long history of innovating in that space. We have the core key seminal technology needed for Ethernet. We have actually been able to design, from the ground up, Silicon One, and we brought the team in close to eight years ago now.
They did not have a product when we acquired the company because we acquired the people, but they had an idea for an architecture that would enable us to move extremely rapidly and build derivative variants faster than any other company on the planet, and we've seen that with 14 variants in four and a half years. So we feel very good about our knowledge of the space, our closeness and proximity to understanding the largest scale that's required in this space by our work and collaboration with three of the four hyperscalers that have already gone into utilizing our products in this space. And so we feel confident that we're gonna have sustainable differentiation for Silicon One for the foreseeable future. Excuse me.
The second question, as far as Sovereign Cloud, you know, we certainly see that as a tailwind in the data center space. But as I mentioned during my talk track, there's a bit of complexity that's more than just deploying a standard data center deployment. And the NIC, as well as the network, as well as the overall thing, as well as the overall— Sorry, I should have covered up before I coughed. Apologies. As well... I think I've been talking too much today. How could that be possible?
You, you got Vegas voice.
I know.
There, there's water over beside you there.
I've been drinking water like a fiend. This is amazing. All right, back to our normal scheduled answer. So this is one of the big reasons why we have built out HyperFabric. So HyperFabric can go to any scale to enable enterprises where they might want to deploy it at an edge, where it might just be a rack of GPUs, or they might want to put it in their enterprise data center and have a whole cluster of GPUs, or if they might wanna build out their own sovereign cloud. This is built to make things easy to deploy AI Infrastructure as a complete system, and we feel really great about the long-term sustainable differentiation that we have in this space.
All right. Let's move to, let's move to Amit.
Perfect. Thanks a lot. Amit Daryanani, Evercore. I guess, Chuck, maybe to start with you, you've spent a fair bit of time over the last few years talking about the shift to subscription, the shift towards a more ARR-centric business model, and the numbers show that at 53% of your sales coming from there. As you reflect back, though, what do you think that hasn't dampened the volatility in your business, especially in the last few quarters, where revenues were down double digits? I'd kind of love to understand why do you think that happened, and, you know, how do you know that doesn't happen going forward? So maybe you could reflect on that. And then, Scott, maybe you could dig in a little bit more on the 6%-8% EPS growth. I'm assuming there's some investments being made for AI.
Is there a way to kind of maybe call that out? It doesn't appear you're embedding buybacks in that model either. Is that fair, and could that be an upside lever for you?
So you wanna go first, or you want me to answer the second question?
Go for it.
So we are embedding, obviously, continue—we've already ramped up. We weren't—If you go back a year ago, we weren't spending 50% of R&D in AI, cloud, and cyber, the sum of those three. So we continued to pivot more resources into that space, and we'll continue to do that through time. You got, you know, with—Obviously, the networking business is core to us, even outside of what we're gonna do in, in AI Infrastructure there. There's a huge amount of investment going into that space, but I think it's also important for you to get a feel that we're investing where the growth is, and those are obviously the, the faster growth areas. That will continue through 2025, 2026, and 2027. The share buybacks are somewhat embedded in there.
I think the thing you have to remember on share buybacks is, we also issue equity every year. So you can't just do $5 billion of spend divided by, you know, $50 share price equals this many shares going out the door, 'cause there's a, there's X amount that's just, just offsetting dilution. So out of $5 billion. So it's embedded at the rate that we had talked about, which is $1.25 billion per quarter. You've seen us do that now for 5 quarters, so $5 billion per year.
And a bit on your first question, I think fundamentally the reason that we've had inconsistencies in our business has been this demand/supply chain imbalance that we've experienced the last three or four years. I think when we are in a normal mode of demand, I think you'll feel the benefit and the effect, which is sort of built into how we thought about our long-term model today. I mean, all of it's baked in. I don't know, how much do we have in short-term RPO right now?
About $20 billion.
$20 billion of next 12-month revenue sitting on the balance sheets. I mean, that's, that's just real revenue that 8, 9 years ago we didn't have sitting there. We had to chase it every quarter. So we should feel it in a normalized environment, I think, more so.
Yeah, and I think the only other thing I'd add is, we have really good visibility to 53% of the business, right? There's another 47% that continues to be, especially given the--what we've been through with all the supply chain disruptions. That's really what's driving some of the volatility, and I see that normalizing as we now-
Right
... have cleared it. We've cleared the backlog. The backlog shifted to our customers. They're getting all that installed. I see that normalizing, but that's the piece that's been tough to call. It's not the recurring revenue piece.
Let's go to that side of the room. Ryan, could you hand Tom the mic?
... Great. Thanks for taking my question, and thanks for all the information here. Tom Blakey with KeyBank. You know, as you guys have shifted more to ARR and the predictability, I was wondering if you could maybe talk about NRR. I know Gary's a software executive is well steeped in that. And then, as it relates to Splunk, you know, the growth rate, the framework of 4-6 is below the 5-7, but being catalyzed by Splunk, I was wondering if you could share, Scott, what you're assuming in terms of Splunk prime? Like, what kind of growth are you assuming with Splunk into fiscal 2026 and 2027? I think that'd be helpful. Thank you.
Yeah. I mean, what we've assumed, yeah. So Gary talked about it during his piece, that the last quarter they announced, as they were still a public company before we got the transaction closed, they grew 15% in revenue, and we expect that... Now, that, that-
ARR.
ARR.
Sorry, ARR.
We expect that, which has an element of kind of end-of-year workflow. We expect to continue to see it grow in that space. One of the key tenets of the way that we are integrating Splunk is ensuring that we don't disrupt the momentum of the business. Back to the conversation you had in Davos, which was a year after you and I had a conversation-
That's right.
-in Davos.
But I'd also say that as we go forward, I'm very confident we will be successful integrating Splunk broadly across the Cisco portfolio, and it will be very hard for us to untangle what is Splunk revenue...
That's right
... and what is Cisco revenue? If we do it right, it'll be impossible to tell.
Yeah.
I think that will really ultimately fuel growth across the business.
Yep
... that we won't be able to isolate to just Splunk.
I agree with that. Tom, remind me your first question.
You know, having that association there are-
Yeah, it varies, as you'd expect, pretty widely across the various product lines. It's in the range that you'd also expect. I'd say there's a little bit of—there's some areas where it's been a little bit lower than expectation, as we've talked about. Some of the competitive dynamics, particularly in our meetings business, that NRR is gonna be lower, but it's in the range that you'd expect, given the scale of our ARR at this point.
Ryan, could we also get the other question also about that?
Yeah. Thank you very much, Aaron Rakers at Wells Fargo. I wanted to go back to the AI, you know, Ethernet discussion a little bit. You know, it seems to be over the last couple of weeks, there's been a lot of movement around putting stuff into the data processing unit or BlueField at NVIDIA. I know the announcement this week with you guys. So I'm curious, you know, as you think about the timing, as you mentioned earlier, what are some of the things that you're looking at that kind of set the table for the timing of these deployments to start to open up? And I'll dovetail the Ultra Ethernet Consortium in that narrative. Is there something there that you see as important that unlocks this Ethernet opportunity over the next couple of years?
Jonathan?
Yeah, I can start. So starting with the UEC, the Ultra Ethernet Consortium is really important 'cause it enables us to get to standards. However, these initial deployments aren't waiting for waiting for any standards. Where this is about being able to go and solve the challenges today that these large AI clusters are putting onto the infrastructure. And some of the concepts of these smart NICs with DPUs have been around for a long time. I mean, you know, 15, 18 years ago, we were drawing things of how you could use these architectures to become more efficient with Ethernet, but there just wasn't a need, so we didn't do it.
These are not necessarily new concepts. It's just there is a need now to squeak out the maximum amount of efficiency out of Ethernet and be able to treat the whole system as a cluster, and how you can make sure that it is lossless, has low latency, low jitter. All of these things are critically important, but it's where, in the past, the NIC or the smart NIC or the DPU kind of be a ship in the night to the data center networking fabric, and that's no longer the case, right? They need to act as a cohesive unit, and that's probably the biggest thing.
So the Ultra Ethernet Consortium is gonna be great because then we can define what those standards are so that you can be more interoperable, so that you can plug and play multiple vendors, because that's the benefit of Ethernet, is so that you can go and have multiple vendors and have the scale that all of Ethernet can bring to save costs and power and things of that nature. So, but it's gonna be a while, 'cause when you get a lot of smart people in a room, we have a tendency to argue a bit about coming up with what the right answer is, and so it'll take us time to get to conclusion, get those into products, get it out. But the networks and movement to Ethernet is not gonna wait. That's gonna happen now.
It'll just be single vendor by pod.
All right. Can we go to Ben, over there? Eric, maybe you can get to him.
Hey, guys. Ben Reitzes, Melius. Great to be here. Thank you. Chuck, you know, there was a question on the call, your most recent call about, you know, some of the backlog movement last year or, yeah, earlier this year. And, you know, it was kind of quantified around $4.5 billion, and if you normalize for that, you guys said there was some logic that you would grow your business next year in networking, maybe in the mid-singles. And your long-term framework, Scott, you know, just has revenue growth of, you know, let's call it, was 4%-6%, so it's 5% for two years. So kind of... You guys kind of backed that math.
So, what I'm trying to say is, if you normalize for the backlog, you know, you guys said on the call, you're kind of growing five-ish next year, and then you're growing five-ish, five-ish. You know, obviously, optically, it's, it's really not five for FY 2025, but let's, let's live in that world. So I'm just trying to understand. You, you got Splunk, which is much better growth business. You got this AI initiative with NVIDIA. You got some pretty good presentation material here. I mean, why isn't it better? Why, why isn't it like five, seven, eight? I mean, w- w- are you just being conservative?
I just wondering for the value investors in here who are looking at a down year, you know, what, what can really get them excited, you know, about the out year if it's just 5, 5, 5, if you normalize? And I hope you, you know, we're-- I don't mean any, you know, you know, be too harsh. I just wanted to see if you're being conservative or if that scenario can play out. Thanks a lot.
Go ahead, Scott.
I'll talk about the numbers that are there, and then you can talk about some of the tailwinds that are there. I mean, it's a range. It's 4-6. You pick the midpoint, but it's a range through there. But I'd say one of the things, Ben, I learned from having done this back in September of 2021, when I was six months in the job, seven months in the job, is it doesn't pay to be aggressive on these longer term goals. It pays to be a little conservative. I think some of the things that I talked about earlier that led to us coming in below that 5%-7% range, I think those are pretty well stabilized at this point. We're seeing some green shoots in cable and telco.
We're seeing stability in the collab business, even in the meeting side, and actually growth on devices and cloud contact center. But as we look ahead, I think it pays to, at this point, kind of given what's happened over the last two or three years, it pays to be conservative as we lay out those longer term goals.
Yeah, Ben, I would just add that if you go back to when I closed my opening presentation earlier, I made a very deliberate comment that our intent is to exceed the numbers that Scott's going to share with you later. And look, I think we're also still early in sort of what's the real AI business in the out years look like? We're modeling it, but we're not putting... There's upside with that business that's not built into this right now because you just--you can't see it, and we gotta get moving on it before you feel really good about increasing and have real visibility to what are the size of those numbers in the out years.
You can make some assumptions and get there, but I'd say that piece of it is relatively conservative as well.
All right, guys. Thanks a lot.
All right. We'll move to Samik. Eric, right behind you. Samik.
Hi, Samik from JPMorgan. So I'll ask you a question on AI, which is pretty much a follow-up there, but this goes back to your prepared remarks, where you said AI, cloud, security, that's taking about 50% of R&D. I would expect that at some point when you're sort of looking at the long-term model, these three areas start to contribute 50% of the growth as well, which is what you're investing towards. So when you think about AI in that sort of long-term model, when do you get to that point where AI, security, all of that contribute 50% of the growth? How should we think about that? And, Scott, maybe just more clarification questions.
In relation to the balance sheet, do you really expect to continue to keep the same debt level, or is interest expense sort of paid down, some level of paid down debt, paid down and lower interest expense modeled in, embedded in the model as well?
Yeah. On the AI question, we'll come back and talk about the, you know, does it make more sense to delever or not? On the AI question, there's two elements. You've heard me say that we have kind of two opportunities inside AI, right? One is the AI Infrastructure, and I think that's what's behind your question, and that will be something that we can track. We talked about the use cases that we've already... One, we've talked about the progress that we've made. We had Dave Maltz here talking about the importance of us in the Azure infrastructure. The second opportunity we have, and that is driving some of that AI investment, though, is the things that you heard from these guys, where they're building AI into our collaboration tools.
They're building AI into ThousandEyes and into our networking tools. They're building AI into security, very on a very broad way. So some of that's not gonna be in the... So now it's like, okay, do I allocate 5% of the revenue or 30% of the revenue of a given product because of the investment we've made in AI? So there's a piece of that that I think is gonna be difficult to track discreetly, as we talk about it. Clearly see that driving product efficacy and driving customer productivity as a result, but actually try, you know, figuring out how much of that security sale that has a lot of embedded AI in it, that should I attribute to AI? I think that's fun with numbers at that point.
But if you think about security in general, if you take a high percentage of the Splunk growth that's gonna contribute, and you take the growth that we expect in the security portfolio, you take the cloud infrastructure growth, and you take the AI Infrastructure that's gonna come in. I just did some quick math in my head, and uh-oh. I have a math degree. He's an engineer, so I'm the one—I'm the numbers guy up here. It's... I don't think we're too far off. I don't think it's a stretch relative to that being roughly 50% of growth or even maybe higher.
We'd have to do the math, but, I don't think it's, like, meaningfully different, like a gap, when you think about the revenue growth here, the revenue growth in security, which are primarily security, and then the AI Infrastructure, and then the cloud growth that we should expect to see on the front-end side.
On your cap allocation question or the, you know, should we delever more quickly, which I think is what you're trying to get at with that question. Whether it's cash on it, so we'll support the dividend. We'll support the growth of the business, we'll support the dividend, we'll buy back shares at the rate we've talked about with a, you know, an opportunistic overlay on that, and then you've got a, you know, some pool of money left. You know, at this point, whether that's cash on the balance sheet that's earning 5%, or it goes into delevering, and it avoids a 5% interest cost, it's a 5% impact either way, right? So it's not a big swing on the balance sheet.
I think we've been really good stewards of the balance sheet, even in the wake of a $27.5 billion cash acquisition of Splunk. We're still double A minus and not on watch. So I think the balance sheet is in really good shape, but I wouldn't think of, well, delevering is suddenly gonna save money. If it's cash on the balance sheet, it's gonna earn interest at roughly the same rate at this point.
Can we go to, can we go to Meta, right next to Samik? Thank you, Eric.
Great. Thanks, Meta Marshall from Morgan Stanley. Maybe Gary, starting with you, on Go-to-Market.
Yep.
You know, simplifying the sales process for Cisco has been kind of a goal for the past few years. And so as you think about kind of simplifying it, is it some of what Jeetu was talking about with suites? Like, how do you think that Cisco can really simplify that sales process, and how does that translate to really converting what the channel, you know, which is primarily how Cisco sells, you know, getting that channel training process? And then for Jeetu, you've mentioned a lot on the Hypershield upgrade cycle. Just how should we think about kind of the products that will be upgradable? Like, what is the cadence that we should think about those products coming out? Thanks.
I'll go first. So on simplification, so one of the challenges has been our product line has grown really, really broad, and we have a diversity of Go-to-Market efforts across that broad product line. And so as a good example, in security, as Jeetu's described, we're down to broad suites as opposed to individual point products. And so in security, in particular, we had literally different Go-to-Markets for individual products. And so bringing it all together to a simplified, suite sale, with a single set of sellers, it can just be much more simple than having individual Go-to-Markets. So it's really just rationalizing and simplifying. There's been a lot of really good work that's already happened, and we're just gonna continue to progress that work.
Again, as you mentioned, it all gets delivered through the channel, but the simpler we can make it for the channel, where they can sell a broader set of suites for a single price and then grow and expand from that, that will go a long ways to making it easier for customers to get better outcomes, and we can drive better innovation into the hands of our customers.
To answer your question on the cadence of upgrades, you know, that's gonna be a. So on the patching side, we will actually have, within minutes of when CVEs get published, we'll be able to have compensating controls so that the patching window, we'll have a shield in front of the vulnerability almost within a matter of minutes that can be deployed throughout the infrastructure. On the upgrade side, it's just continuous deployment, continuous updates. That'll happen. And that's the beauty of the SaaS model is, you just won't have to worry about that, that dimension anymore as we move forward.
I think what you just said is so important, it bears repeating. The ability to shut down or to close off a vulnerability in a matter of minutes and have it deployed across your enterprise.
Yeah, by the way, there's another piece which is important, which is right now, only 20% of the vulnerabilities that are identified actually ever get patched, because most people don't have the capacity, most companies don't have the capacity to patch them, right? And so only 20% get patched, 80% remain unaddressed. And what most organizations are working on is a prioritization of how you sequence the patching with a kind of despondent acceptance that you're not gonna be able to patch everything that's exposed. And what we are doing is completely changing the game on that front, where all the CVEs that you have will have a 9-15-minute, you know, kind of compensating control that can be applied, that can go out to thousands, hundreds of thousands of enforcement points right away.
Like, that's why we think it's so interesting, because the architecture of security is completely changing with this. It's not just a product or a feature.
All right, let's shift to that side of the room. Start off with George. Marty, you're right.
Hi there, George Notter from Jefferies.
Sorry.
Curious about your views on the market right now. There's a lot of software companies that are out there kind of talking about SMB softness right now. There's some conversations about how enterprises are distracted by, you know, their AI strategies. Are you seeing any of these effects sort of impacting the current sort of tenor of business?
Um-
I've got a follow-up, too.
So we're, we're obviously in the middle of our quarter, so, but I would say, I, I don't think there's been a meaningful change from when we ended the last quarter.
No.
Is that fair?
Well, yeah, and I think what we said on the last quarter, George, and you heard some of this, is that, you know, particularly in Splunk, which is, you know, the biggest SaaS software engine we have, probably the part of our portfolio that's at least material, material enough to where it's like, "Okay, let's focus on that and see if we're seeing any change there." We have a lot of SaaS growth outside of that, but that's the biggest single chunk. And Gary commented on that on the last call, that we really aren't seeing a change at this point there. But I think in that software realm, I think cyber lives in a different category than-
Mm-hmm
... selling lots of other products in a lot of other segments, and that's not what-
Applications, yeah.
Yeah, and if you think about it, I think the scrutiny came on the application side, that's not the business we're in. And so I don't think that the problems haven't been solved in the cyber. People still are worried, and I, I don't think that the-
... that buyer's perspective has changed.
That's both on the security as well as the compliance side for reporting-
For sure.
Both of those aspects.
Got it. And then just for Jeetu, can you tell me what a compensating control is? I guess I think about that as a software patch, but you're not software patching, so help us understand what that is. And I gotta say, like, I've done some analysis of all the different operating systems inside the company and all the different software releases. I mean, you guys put out a lot of patches, you know, of operating systems historically. So, like, are we getting away from that model, and what exactly is going on? Thanks.
Yeah, so the compensating control is not designed to replace the patching. It's designed to make sure that you get a window of assurance while the patch is being tested and deployed. What a compensating control is, for example, you identify a vulnerability, and you wanna make sure you can block internet access for a service for a time period. That's a compensating control. That can be applied, you know, to the service, whether it be a microservice or a Kubernetes container, and you just actually apply the compensating control. As soon as a patch is applied, what then happens is the compensating control is smart enough to then remove itself, and so it's fully automated in the way that it works. It's pretty cool. We can geek out more if you want on the side.
All right, can we move to Woo Jin Ho, please? Thank you, Marty.
Thank you. Woo Jin Ho from Bloomberg Intelligence. Couple of questions. In terms of the mid-teens long-term growth and security, just to drill down a little bit, how much is it dependent on shoring up the traditional firewall business as well as, you know, Hypershield growth and the new products versus and Splunk to drive that growth? And then, in terms of your inorganic growth, you know, you've done a lot of security deals, collaboration deals, as well as networking deals, but do you have to cast a wider net to really attack the AI networking deals that are ahead of you?
Mm-hmm. I'll start on the 15%-17% that we showed, which, by the way, was security and observability together, right? It was both in there, in that range. And it's like, you hit it. It's all the above that are gonna drive that. It's continued growth and, you know, some of the revenue synergies that we see from Splunk as a standalone. It's the opportunity to, as we combine it with the rest of our security portfolio, to actually have a tailwind, not just to Splunk itself, but to our overall security portfolio. We're seeing really good progress in our firewall space. You saw security returning to growth for the last three or four quarters now.
So we're seeing that turnaround, and with today's announcement of a new firewall coming out there, that's gonna be a tailwind as well. I'm not sure I wanna dissect it any more quantitatively any more cleanly for you than that, but that gives you a good sense of what all the drivers are that are driving that out through time. Observability, and maybe I'll let you comment on this, Gary-
Sure.
'Cause you're deeper than I am. Observability, I think, is a, you know, longer term, is a very significant growth opportunity for us.
Yeah, and I think as we discussed in the presentation, there's a tremendous opportunity for us to deliver capabilities that no one else in the industry can, and I think we'll see that as a growth engine, including ThousandEyes and all of the other capabilities that we deliver across observability. So I'm super enthusiastic about that, and that will be a growth engine for us.
And, your second question was, do we need to cast a wider inorganic net for AI networking?
Yes.
Sorry, can you wait?
Yeah.
'Cause this is being webcast.
Let me repeat that again. You know, when, when I look at your M&A strategy over the past decade since you've been on, Chuck, it's been collaboration, trying to get as much recurring revenue, software, so on and so forth, right? This AI networking opportunity is just entirely new. You may wanna go deeper into the semiconductor space. That's where all the money is going, and I'm just curious if you need to cast that wider net to, to really tap into, AI networking as a whole, especially as networks are becoming more and more disaggregated.
Yeah. Let me, let me make a... I'll make a strategic statement about how we think about it, and then I'll let Jonathan comment on what, how you feel about it. There's nothing about how we think about our inorganic strategy that would keep us from doing anything that you described. It's not like we've said, "Okay, we're only gonna do them in this space or this space." I would love to go do something significant that would accelerate our success rate in, in AI networking, so, absolutely wide open to it.
I think I have nothing to add.
Okay. Yep.
All right. Can we go right here in the middle? Maybe Eric.
Oh, hi. Hi. Thank you for the questions. This is Yang Pu for Carl Ackerman at BNP Paribas. I just want to follow up on your AI market opportunity in your prepared remarks. So you earlier showed a slide that your hyperscale networking opportunity could grow nearly like 30% bigger, but the AI opportunity within the hyperscalers may only grow like 18% bigger. However, your growth in the enterprise AI are much higher. It's 29%. So first question, why is that? What's driving the difference? And second, could you discuss what AI opportunity is embedded in your 2%-5% growth rate in networking that you laid out for fiscal 2026 and fiscal 2027? Thank you.
Jonathan, you wanna take the TAM growth, and then you-
Sure.
- 5.
I can do the first part. So, a lot of this came out of the research from 650 Research, 650 Research. And what we've seen is that there's a little bit of a disparity around-
... the, where they are putting the transition of InfiniBand versus Ethernet, and we are, I think, a bit more bullish on it, as, many other analysts are in that space. So that's the majority of the delta in growth on the AI web versus just the overall web growth. That's that piece. On the enterprise side, one, it's a much smaller base, and so the growth numbers are larger. But also, we're participating in a much bigger TAM. You know, the overall TAM size is much bigger as well because we sell compute to the enterprises. We do not sell compute to the hyperscalers, and so that also is a tailwind from a TAM growth perspective for the enterprise AI versus the non-enterprise AI.
When you talk about AI Infrastructure contribution to some of the numbers that are out there, we've been pretty clear that we see line of sight to $1 billion of AI Infrastructure coming in fiscal 2025, probably more toward the second half of that. I'd say when you look beyond that, we've taken a somewhat conservative view, as Chuck said 10 minutes ago, a somewhat conservative view of what that looks like, and I think that's why it's a much wider range than you'd expect. That's quite a large business for us, and to have that wider range gives you a sense of the degree of variability that we see inside there. It's an opportunity that we think is a great opportunity, and we're extremely well positioned for it.
But again, I think it at this point, it pays to be a little more conservative on some of those longer term goals.
All right, we have a question from Michael on this side. Thank you, Ryan.
Great. Good afternoon, Michael Ng from Goldman Sachs. I just have two on Splunk, really just on the international and channel opportunity. I was just wondering if you could expand a little bit on, you know, what the competitive market dynamics are like in Middle East, LATAM, Asia Pac. Are there existing SIEM competitors there that you expect to take share from, or is this greenfield? And then on the channel opportunity, I was just wondering if you'd talk a little bit about, you know, why Splunk hasn't resonated in the channel historically, and, you know, what's gonna change on the go forward? Thank you.
Yeah, no, great question. So from an international perspective, the markets are gonna look pretty similar to the way they look today across the US and EMEA. I think what you'll see is more brownfield. We're gonna be replacing things or modernizing things, is how I would describe it, and that's a really well-known sales play that we've been running for a long, long period of time, and I think we're really good at that sales play. So that's just something we will take to these new markets and drive it that way.
From a channel point of view, Splunk had historically been somewhat schizophrenic, just being blunt about that, about the channel, and so I think when I joined, we got much more consistent and thoughtful, and our percentage of business going through the channel radically increased over the course of the two years I was there. I look at the step of joining Cisco really accelerates our momentum around channel adoption, and given the market motion that Cisco has in place, we can snap right into that, and we'll take advantage of a set of resellers that we just didn't have access to, frankly. We were more boutique.
We get to Cisco, and we become much more mainstream, and we bring along those boutique partners as well, because I also think they are going to be super compelling to drive broader adoption of the Cisco suite, 'cause they're probably not selling the Cisco suite today. So we get a lot of cross-fertilization in the channel that I think will be quite compelling.
All right, I wanted to hand it over to Chuck, just to go through some closing remarks. Chuck, I'll hand it over to you.
Thank you. Thanks, Sami. So I'll keep these brief 'cause I know everybody's had a long day. First and foremost, I wanna thank you all for coming. We really appreciate you spending time with us. Hopefully, this has been informational, and hopefully, those of you who spent the day with us also got a good feel for the, for the innovation that the teams are driving. We are very focused, relentlessly focused on driving growth as a company. We are relentlessly focused on delivering great customer outcomes, which we think will help us drive growth. We're, we're committed to consistent execution, and hopefully, after today, you realize how focused we are on winning in AI, cloud, and security, and how this time, I think it's significantly different.
You heard the Microsoft gentleman talk about the way they view us today versus the way they viewed us a decade ago with the first cloud transition. I think we are very well situated to actually help our customers, particularly these enterprise customers, as well as the hyperscalers on the move to AI. And I think that Scott laid out what I hope is a compelling initial view of the framework that we expect over the next few years. And I think now that we've gotten through the complexity of the last three or four years, hopefully, we can actually take advantage of the success we've had in this recurring transition that we've been working so hard on, and actually have that contribute significantly to a much more consistent execution as we go forward.
So with that, we really appreciate you spending time with us, and we look forward to seeing you all again soon. Thank you.