Welcome, and thank you for standing by. I would like to inform all participants that this conference call, as well as any Q&A, may be recorded and made available to clients of JPMorgan. We're a company that is presenting, and a recording may also be posted on their website. Views and opinions expressed by any external speakers on this call are those of the speakers and not of JPMorgan. Part of this conference call may also be reproduced in JPMorgan Research. If you have any objections, you may disconnect at this time. This call is intended for JPMorgan clients only. Fresh participants are not permitted on this call and should disconnect now. Unless otherwise permitted by internal JPMorgan policy, members of JPMorgan Investment and Corporate Banking are not permitted on this call and should disconnect now. I would now like to pass the conference over to Samik Chatterjee with JPMorgan.
You may proceed.
Yep. Thanks, Josh, and thank you, everyone, for joining. We have the pleasure of hosting the Cisco Innovation Tech Talk today. With us are Jeetu Patel, President and Chief Product Officer; DJ Sompath, who's SVP of AI Software and Platform; as well as Sami Badri, who is Head of Investor Relations and Strategic Finance at Cisco. We have a star-studded team here to go through some of the technology questions, as well as fresh off the bat from Cisco Live, a lot of the topics that have come up from investors on that front. I am Samik Chatterjee, who covers hardware and networking at JPMorgan. Before we get started, I'll hand it over to Sami to go through some of his usual disclaimer statements before we kick it off with the first question for Jeetu and DJ. Sami, over to you.
Thank you, Samik. This is Sami from Cisco Investor Relations. I'm just going to read a quick forward-looking statement disclosure. We may make forward-looking statements which are subject to the risks and uncertainties outlined in our filings, in particular on Forms 10-K and 10-Q, and actual results may differ from those forward-looking statements. Samik, I will hand it over to you for Q&A, and thank you again for hosting us and the Cisco team.
Yep. No, thank you. And thank you for the time, everyone. Thank you, Jeetu and DJ. Maybe I'll start you off here with the question on Cisco Live, which was last week. Maybe you can start by highlighting for us what the biggest takeaway for investors should be. We saw a lot of announcements around Cisco Live, but what would you highlight as the biggest takeaways in terms of your announcements at Cisco Live last week? And thank you again for the time.
Thanks, Samik. It is a pleasure to be here. Hello, everyone. We came out of a pretty exciting Cisco Live, about 22,000 of our closest customers that we hosted in San Diego. At the highest level, what we wanted to do was reframe the kind of problems that we as Cisco are solving in this new era. Because if you think about where we are right now as an industry, we are moving from this era of kind of the version one of AI, which was a bunch of chatbots that intelligently and interactively answered questions that people had, to now a version of AI, which is agents being able to conduct tasks and jobs almost fully autonomously on behalf of humans.
As we make the shift, the underlying infrastructure requirements, the underlying safety and security requirements, the underlying kind of monitoring, visibility, assurance, and observability requirements, and the data fabric requirements are fundamentally going to be different tomorrow than what they were today, or what they were yesterday, I should say. What we did was we kind of laid out what was the largest product refresh that we have had. From what I'm told, I've been here for five years. What I've been told is the largest product refresh we've had in 20 to 30 years at a single venue. We went through that, but we also went through kind of simplifying our message. Because our message in the past has been super complicated because we've had multiple different business units, and these business units all kind of ran autonomously.
It was hard to stitch a message and a narrative together that customers could effectively understand. What we have done is we started with this notion of a one Cisco message. We wanted to make sure that our differentiation was highlighted in a pretty kind of clear way. Let me walk you through the one Cisco message and then the differentiation that was in the message, and then what the payload of capabilities were. The one Cisco message was, we are here to solve three key problems for customers.
Problem number one is to enable our customers to have AI-ready data centers for this next massive build-out of data centers that is about to be embarked on globally because of reasons of data sovereignty, because of reasons of power scarcity, where the data centers are starting to get built, where the power is available, and because of reasons of inferencing workloads with agentic. You're going to see a much more sustained demand for capacity rather than a sporadic demand for capacity. If I ask a question of an agent, of an AI chatbot, and it gives me an answer, you might see a spike in inference capacity and volume. If I have agents autonomously and proactively conducting work behind the scenes, then you're going to have a spike of capacity.
The first problem that we wanted to make sure that we got everyone very clear on that we were solving is we were going to provide the critical infrastructure required for building out these AI-ready data centers across any modality of the cloud, whether you want to build it out as a hyperscaler, as a neocloud, as a service provider, or as an enterprise. Because you're starting to see that some of these data centers are getting re-accelerated in the private cloud, especially for inferencing workloads. We want to be the vendor of choice that does the networking, safety, and security capabilities for that build-out of the data center. We also talked about our partnerships that we've recently made and announced with the Kingdom of Saudi Arabia and the HUMAIN project.
We did the same with what we announced in Abu Dhabi with G42 and Stargate UAE, where we're partnering up with these different players in the market to provide this kind of critical infrastructure. That was the first one, which was AI-ready data centers. The second big problem, and that's where the digital workers sit, right? The second big problem that we wanted to solve and tackle was this notion of a workplace and future-proofing the workplace, which could be a campus, it could be a branch, it could be a home office, it could be your car, it could be a factory floor, it could be a store.
Regardless of your workplace that you have, where people are going to be working, we wanted to make sure that we provided the right level of connectivity infrastructure, the right level of security infrastructure, the right level of observability, and a data fabric for correlating of the data for troubleshooting effectively. That was the second big area that we talked about with a bunch of announcements over there. The third area was making sure that all of these things would run in a reliable way. Ability for an organization to be resilient when their digital infrastructure goes out was something that we really had focused on.
That is where the Splunk acquisition comes in, where the ability for us to correlate telemetry and data across networking and security and observability to be able to say, "Hey, when you have an outage, do you know what the outage is? Do not spend the first four hours determining in a war room what actually happened, but try to get that determined and detected early on in the game so that you can spend most of your time doing the response and remediation side." If it is a network outage versus a security breach versus an API overage versus an application bug, we would be able to correlate data from multiple different data sources and tell you exactly what the challenge is so that we can then spend most of the time on response and remediation.
That was the highest order storyline that we had, which was three key areas: AI-ready data centers, future-proof workplace, and providing digital resilience, with the core differentiation being we will provide a platform advantage rather than having operating like a holding company with 250 different companies. We wanted to operate like a unified platform. That unified platform would provide the benefits of a common plane for management and a precipitously lower cost of marginal ingestion of new technologies so that every single time you buy something new from Cisco, the cost is lower. The value is not just in what you just bought, but also in the things that you already have from Cisco. It creates that snowball platform effect. That was number one.
Number two was this differentiation around being the fact that we build our own silicon allows us to have a full stack, everything from silicon to the network infrastructure to the security infrastructure to models to the data platform to the application, all wrapped around with safety and security and observability. We wanted to make sure that we showed that that works cohesively well together. The third area of differentiation is we want to be AI-first in the way that we do things. It is not an afterthought, but it is something that we do from the ground up. This is not just lip service for AI. We will have security and safety products that are built to secure AI itself.
We will have products for making sure that we've got the right level of kind of capabilities for ensuring that your AI state is, in fact, world-class from a scalability standpoint on the network because AI is inherently going to be network constrained. Those were the higher order bits. There were 24 announcements in every single part of the portfolio that we announced, which were new products that we released, which I'm sure we'll talk about.
Yep. Great. Jeetu, that was great. Maybe I—
That was a long answer. I wanted to make sure that everyone had the context.
No, no. Perfect. Maybe we'll start off with the future-proofing of the workplace and the highly anticipated refresh of the campus products, the smart switches, the C9350 and the C9610. Can you outline for investors the difference in features and capabilities relative to what I think most investors will compare, which is the Cat9K portfolio? How do you sort of expect enterprises to go through the upgrade cycle relative to the Cat9K products that they probably have in their install base?
Sure. Let me actually do this. Let me walk you through that for sure. Let me actually take a step, zoom out for a second as well, and say what were the big categories of announcements we made, right, in each one of these three areas? One was around operational simplicity and management. We wanted to make sure that one of the challenges that customers have always complained about us, they've got multiple of these products and franchises. Each one of them seemed to have a different way to get managed. Complexity tended to have a negative effect on Cisco. We wanted to make sure that we actually removed that complexity in the way that we got managed, but did not just take out complexity. We leapfrogged the market in a couple of different areas. We essentially launched this capability.
One, we showed tactically in each one of these areas, Nexus, which is our data center platform. How do we make sure that we merge Nexus Dashboard and ACI in one management plane? We did that. Catalyst and Meraki, two separate kind of product lines. We showed that there's common hardware, common licensing, and a common way to manage both of those in a cloud-managed way. That was something that was released.
What we did was we said, "Now let's take it a step further and don't just simplify IT, reimagine IT." Where DJ did a demo on the keynote is we have completely built a whole new category called Agentic Ops, which was this new way to manage the entire estate through a generative UI, where it's not just something that we do once. We actually generate this dynamic UI through a prompt interface. DJ will talk about that a little bit later. That was what we did. We said, "Now let's go into the campus branch." We did a full refresh lineup of every single one of our kind of switches, Wi-Fi devices, routers that are out there. We've actually started doing a refresh of those. Catalyst, one of those categories, was a smart switch.
What is a smart switch? We announced this concept of a smart switch first at the Cisco Live in Amsterdam in February. We did that in the data center side first, where what we did was we said, "Imagine if a switch actually had dedicated compute that was isolated that could run workloads separate from the network traffic forwarding that were largely for kind of traffic inspection on security," right? We announced a smart switch, which was a top-of-rack smart switch. That essentially had a DPU on it from AMD in that smart switch. What that would allow you to do is analyze live traffic and be able to do a bunch of use cases. Hypershield, which was our product, would be able to get an enforcement point on the switch itself.
We wanted to take security, bake it into the fabric of the network, and run it on a switch. We were able to do that with the data center. We now have taken that same approach into campus branch because the goal over here is to say, "Let's take security and bake it into the fabric of the network," right? If you bake it into the fabric of the network, our new generation smart switch, which is the Catalyst 9350 and the 9610, they're essentially engineered to drive enterprise upgrades by delivering substantial performance and advanced capability required for this modern AI-powered workplace. These switches will be powered by Silicon One. They offer up to 51.2 Tb per second of throughput capacity, and latency is below 5 microseconds.
They are critical for supporting high-stakes AI applications and the explosive kind of network traffic that they generate because as we see these agents coming into the workforce, essentially what we are doing is we are augmenting agent capacity to human capacity. The agents are going to actually have use cases like computer use. Computer use as a use case is one where an agent pretends to be a human and operates like a human would do with using your computer, which means that if you have 1,000 agents added to 1,000 employees, now you have the traffic patterns of 2,000 people. You are going to need to have the infrastructure be able to go out and accommodate that increased traffic pattern that is needed. That is what we essentially did. A key kind of differentiator of these compelling upgrades was, like I said, the deeply integrated and advanced security.
If you think about how we think about this upgrade, it's going to include quantum-resistant secure networking to protect against future threats. It's going to have Hypershield readiness, which is kind of a distributed kind of enforcement point of security, a completely different way of enhancing network segmentation at machine speed. It'll have this capability that we built called Live Protect. Now, what is Live Protect? Live Protect was this capability where our hardware and specifically our switches are going to essentially, when there's a vulnerability today that gets published, it takes about 45 days for that vulnerability to get patched. Only 20% of the vulnerabilities ever get patched because most people don't have the capacity. What we've done is we've said, "Why?
Wouldn't it be great if we could actually, within the matter of the first few minutes, issue a compensating control on that switch that can go out and kind of compensate for that vulnerability and provide a compensating control? When the patch happens, the compensating control is smart enough to be lifted. We're starting this with Nexus OS, and you'll actually start to see this also within campus branch. What that Live Protect does essentially is makes every switch secure from the get-go when these vulnerabilities start getting issued and getting kind of published in the market because what's happening is it takes 45 days to patch, but it only takes three days for an exploit. We need to make sure that we compress the time for a compensating control and not have it take too long.
That was the example of how the Cat9K products will also get enhanced in the future. It's simplified management, more AI, and Agentic Ops, refresh set of devices, and security baked into the fabric of the network.
Okay. Great. A lot in there, but maybe if I can just overall take it back to how we quantified for investors, how should we think about the size of the install base that would look to upgrade into these products and any sense of what the length of an upgrade cycle into these smart switches would look like from your perspective?
Yeah. I mean, look, at some point in time, we believe that this kind of fusion of security and networking is applicable to any and all class of customers that we have. Now, we're just starting out. I think there's going to be an education process because this is not just another switch. It's a whole new category of a switch. It's actually a whole new category of an architecture that needs to be explained to customers because if you look at our networking friends, they don't have a security stack. If you look at our security friends, they don't have a networking stack. We're the only ones that have both. What we've been able to do is take that stack and combine it together where security is baked into the fabric.
That requires that different buying kind of entities within an organization, the NetOps entity and the SecOps entity, are coming together and kind of collaboratively making this decision. I think there's going to be an education process in the market. This is a brand new architecture, but this education process will be one where, as companies move to Agentic and as companies move to this newer AI-based architecture, you're going to be not just bandwidth constrained, but you're going to be kind of there's going to be a deficit of trust with the users. If the users don't actually feel safe and secure using these products, then they're just not going to.
Ladies and gentlemen, please remain on hold while the speaker reconnects.
Yeah. This is Sami. Can you hear me? This is Sami from IR.
Hey, just this is DJ here. We're just having the speaker reconnect.
Okay. Sami, I just want to check. Are you also, you're still on, right? Okay. Maybe we lost a bunch of people.
Ladies and gentlemen, please hold while we reconnect the speakers.
Yes. Okay. Sami, if we can maybe kick off where we left off, or we can go from there. I'll hand it back over to you.
Yep. Thank you. Thanks, everyone, for reconnecting, and DJ, thanks. Maybe I'll move you along here to the next question I had for you. I know you've been trying to converge the Meraki and the Catalyst portfolio for a while. Maybe sort of highlight for us, you did touch on this. What does the convergence of those two portfolios help you achieve? What does the company see in terms of benefits on that front? Obviously, the smart switch is looking to sort of further that convergence, but what kind of benefits does Cisco see as a result of that?
Yeah. I mean, this one was kind of a pretty, we already covered a lot of this stuff, but at Cisco Live, basically, what we announced was a single unified platform, meaning everything across the customer's wired and wireless estate can be managed from one place, right? The platform supports any cloud, on-prem, or hybrid deployment, giving the customer a choice, a unified management platform. We further differentiated that by making sure that ThousandEyes Assurance, which delivers real-time visibility and kind of actionable insights for both owned and unowned infrastructure. Historically, ThousandEyes was good at unowned infrastructure, service providers, what the hops are on the internet, where you actually, if you have a bad experience, you can go pinpoint. What you can do now is you can also look at owned infrastructure, your Wi-Fi access point, your switch, your router.
Are there issues that are happening on those devices? If it is, we can actually pinpoint exactly what the challenge is on those devices and be able to do it. That was the second thing that we did. The third thing was this notion of the Cisco Agentic Ops, which will now essentially supercharge our networking platform and provide an AI assistant along with this capability that we launched called AI Canvas at the center of our capabilities. Maybe, DJ, why don't you kind of walk through a little bit of what you demoed and what you announced over there?
Yeah. Thanks, DJ. So here's the thing. One of the key things we did is we introduced the concept of Agentic Ops, right? When we talk about Agentic Ops, we're talking about a paradigm shift on how we're thinking about IT operations will change when we think about how we work, how humans and agents work together to be able to solve problems, right? Now, this is centered around three principles and three important ideas, right? The first one is that the agent needs access to data across multiple domains. It's not just enough to be able to see one domain. It needs to have access to all of these domains to be able to perform these tasks really well. The second key idea over here is that the entire IT operations is inherently a multiplayer sport, right? It's not just a single-player sport. It's not a single-player game.
Teams have to come together to be able to solve some of these problems. Now this team of multiplayers includes users and AI agents. The third idea over here is that you really need purpose-built intelligence that is designed for efficiency and accuracy. What I mean by efficiency is that these models have to not be really large models, right? These are small purpose-built models that can run very efficiently. On top of that, they also have to be really accurate, that they are purpose-built for a specific, in this particular case, specific purpose like the networking domain. Now, to make these ideas come to life, we introduced AI Canvas, which is Cisco's new generative UI experience that is designed to dynamically create this cross-domain collaborative workspace for the IT teams.
We basically showed how you can seamlessly fetch data from Meraki, from ThousandEyes, and cross-correlate that with data about that application with a packet loss instance from Splunk. This combination allows the teams to work smarter, not harder, and it goes beyond just visibility as well. We're using agents to be able to proactively propose remediation actions, predict the impact, but we do this all the while keeping humans in the loop, right? At no point is human not on the loop. When a change gets affected by these agents, you have the opportunity to go back in there and say, "Hey, we can potentially revert this." At the heart of this AI Canvas was our special Deep Network Model. Unlike general AI models that are available, this model is designed specifically for complex networking tasks.
It uses Cisco's extensive experience, over 40 years in networking, thousands of Cisco training courses, and tens of thousands of question and answers examples to deliver these precise solutions. In fact, this model can pass the CCIE exam with ample room to spare.
Maybe one other thing to add here is this is the second model that we introduced in the market, which is kind of this very small footprint, very high efficacy, very low cost to operate model. The first one was at RSA, where we announced the Foundation-Sec model. Foundation-Sec model essentially was open source, but that's for our security kind of products. This is the Deep Network Model. Why is this important to you folks? We will be able to have all of our AI product experiences being powered by some of these models. If you think about what this does, a platform then is able to leverage this powerful intelligence and insights from Meraki and Catalyst Center. Basically, think about 32 million devices supporting over a billion kind of client endpoints.
This reach gives us kind of unparalleled learnings of telemetry we can collect and make our AI smarter and more accurate for problem-solving. We're hiring this, we're kind of helping our customers shift from this, "I'm going to react to an issue that happened," to much more effectively predict the future and be proactive and preemptively kind of deploy solutions from manual operations to agentic workflows and fundamentally change the way that the troubleshooting arena works and the complexity is driven down for how you manage infrastructure. Sorry, DJ. I didn't mean to interrupt, but if there's anything else you want to add, I just wanted them to have the context.
No, absolutely. Not at all. I think you nailed it. I think the last piece I'll add is that this is work in progress. The fact is that we have all of this repository of information brings a quote of our newly appointed board member, Kevin Weil, who says that this is the worst it ever is going to be because of exactly what you mentioned. We're going to add so much more context, and we're going to make this model only better going forward. It's a great starting point. We've had a lot of these folks try the model out, the CCIE experts, and give us amazing feedback. Back to you for the next question.
No, great. That's interesting. Maybe on that thought, let's segue a bit from AI in your products to supporting the infrastructure players that you have been with more than $1 billion of AI orders at this point. I think in talking to investors, most are not convinced that Cisco is regaining share with the cloud companies. Maybe if you can talk about how much of a role has Silicon One played in that resurgence relative to maybe the integration of the solutions with optics capabilities that you have, just help us sort of think about the role that each of those capabilities are playing in the strong AI orders that you've reported.
Absolutely. Firstly, we reported that we had committed, we had actually guided to the street that we were going to do a $1 billion in the full fiscal year. We actually were able to exceed the billion-dollar mark last quarter and last quarter's earnings. If you think about why this is happening, I think a lot of people say, "What role does Silicon One play?" I think it plays a pretty crucial and central role because I do not think hyperscalers would be doing business with us if we did not have the Silicon One offering because they want to make sure that there is silicon diversity in the stack.
The fact that we have silicon and our silicon is kind of P4 programmable chips that are there, which means that every single time you have a new use case, you do not have to do a tapeout, and you have a level of programmability in the silicon. It is something that is particularly for hyperscalers, this is a very important kind of dimension because they want to avoid vendor lock-in. Essentially, what these customers prioritize is silicon diversity, our ability to deliver high-quality systems that are built around Silicon One as a key differentiator. It is imperative that this is competitive in the long term. We continue to keep kind of innovating over there. Optics are a pretty big and integral part of the solution that we provide to web scalers for AI training use cases. These are for AI order figures.
The product mix of AI orders in Q3, for example, was that we had 2/3 in systems, which are based on Silicon One like G200 chips. We also had the balance in optics, demonstrating a growing relevance of our comprehensive technology solution offerings that we have. Silicon and optics are very strategic to our business, and especially when it comes to the hyperscaler side and the cloud scale side of the house.
Got it. Maybe I can ask you to sort of dive a bit deeper into the competitive position that you see Cisco in with its silicon hardware design, the software stack. Maybe we can start by just going through from a competitive standpoint, how do you think relative to maybe Broadcom's one or two sort of T-Switch chips that were released recently or their sort of portfolio of Jericho and Tomahawk? Maybe your sort of strongest competitors there in terms of Arista, Whiteb ox. We can sort of maybe go through that list in a bit, but how do you sort of think about the competitive position relative to some of those? Maybe we can tackle sort of Spectrum-X from NVIDIA at the end of that.
Absolutely. Absolutely. That is a very important partnership as well. Look, if you think about our competitive positioning, it is only strengthened by this super differentiated portfolio. That includes kind of a few things. One, it includes all networking form factors, so fixed, modular. Second, all business models, so Silicon Systems and White Box. Third is support for leading kind of Ethernet fabric designs that we have. Across the board, we have got a fair amount of competitive differentiation. Now, with NVIDIA, it gets to be super exciting because we were the first non-NVIDIA silicon provider in the Spectrum-X reference architecture that they had for the enterprise. What we showed at the event was how our switches are actually now working in tandem with NVIDIA NICs. We were able to show a demo with that actually currently working. It was one more step forward.
In addition to what we've done with being in the reference architecture, we've also worked with them on the Secure AI Factory, which is where our security capabilities of AI Defense, which is under DJ, actually get to be part of their NeMo framework, where not part of the NeMo framework as in we will provide if you are using the NeMo framework to build out models that are going to be in the NIMs architecture of NVIDIA, AI Defense can now secure them. These are open-source models that are going to be standardly kind of containerized with the NIMs framework. We will be able to then provide the common substrate of security and safety for each one of those pieces as well. I think the strategic thinking with NVIDIA, and we were with them at all of these different kind of places.
In fact, we had our keynote on Tuesday. Jensen had his keynote on Wednesday in Paris, where he talked about the partnership as well. Both of us actually talked about each other. Both of us have a tremendous amount of incentive to continue to work together so that we can forward the AI agendas for the companies. In general, what you should expect is we will have close partnerships with the ecosystem, whether it be NVIDIA, whether it be AMD, whether it be others in the market like OpenAI and the model providers. We will continue to work very closely with the ecosystem because it is extremely important to make sure that that continues. We have got a special partnership with NVIDIA for sure.
Can you dive into the competitive landscape here a bit? How do you see the competitive positioning outside of the partnership here against Arista or Whiteb ox? Whiteb ox clearly has been a big topic for investors this year. Maybe anything that you have relative to where Cisco's products differentiate relative to these competitors?
Yeah. I mean, the way that if you think about our differentiation compared to some of the others that you mentioned is, look, there is an advantage. You should think about us like the full-stack provider from silicon to systems to the OS to the optics to the and then all the capabilities around security added to it. Think about some of our competition. If you think about our network competitors, they do not have a security stack. They do not have their own silicon that they manufacture. They are a reseller of someone else's silicon. That actually has kind of downside effects. It is very hard for them to be security-first in the way in which they think.
If you think about our security partners, or not partners, but competitors, which also in some cases we will partner with because we want to work with an open ecosystem even with our competitors. If you think about our security competitors, they do not really have the network telemetry. If you think about what happens in the network, you have to assume that the infiltrator has already penetrated your organization. The name of the game is preventing lateral movement. Where does lateral movement happen? Lateral movement happens on the network. Who has the most amount of telemetry on the network? Cisco does.
If Cisco can take that telemetry, distill it down, and make sure that we can federate that telemetry via Splunk, you will now not only be able to compress the time for detections, you massively compress the time that it takes to investigate, have better detection, and then can do response and remediation pretty quickly. When you start thinking about the entire platform of networking, security, observability, and data, that is a huge differentiator compared to any of our competitors. You add on top of that the fact that we make our own silicon, make our own optics and optical systems. All of that together is pretty valuable. DJ, anything to add?
Yeah. So you are spot on, Jeetu. I would just add that as we think about Whiteb ox, I think the thing is this, right? I think customers more and more increasingly want a fully vertically integrated solution. They want operational simplicity. We have heard this time and again from every single customer that we have talked to. To Jeetu's point about security, nobody really wants an unpatched router or an unpatched switch that is out in the wild, right? It takes a ton of, with 45 days, that is a long time for a vulnerability to actually be on a switch known and available for attackers to exploit. You really have to start thinking about how we are fusing security into that very fabric of the network. You heard us talk about Live Protect. Those are the types of capabilities.
When you combine that with the operational simplicity of what we're talking about, the likes of the AI Canvas, you're now seeing a solution, but that's not something that you're piecing together, right? That is a core fundamental differentiation from how the Whiteb ox solution approaches this. Our ability to combine these pieces together and offer you an easy button that manages all of this and takes away the complexity of doing that, it's a huge win, especially as you think about in the age of AI, in the age of agentic era, that is an expectation that the customers are going to have of anything that they use.
Sorry.
That's a useful line. [Crosstalk]
Sorry. Go ahead. Yeah. No, sure. Maybe we can move to talking about some of the announcements you had relative to the Middle East and the investments there. You had the Stargate, HUMAIN, G42. Maybe help us think about the opportunities here, both from a technology perspective. What's different here in terms of the offering relative to what you're doing for the web scalers? As well, maybe set some expectations on what to expect from a revenue perspective here.
Yeah, absolutely. Look, I think if you think about what is happening right now with AI and this agentic movement, let's start from there, which is every company in the world is going to actually rethink their workflows with agentic augmentation and agentic automation. When you start to think about that, there's going to be a massive level of infrastructure updating that's going to be required. I think in the fullness of time, you will need to have every data center re-racked. You will need to make sure that every data center needs to get an updated kind of architectural shift for how they think about their network, how they think about their network being secure. Networking and security aren't going to be two separate markets. They're going to be secure networking as a market. It's going to be kind of combined together in some way.
You'll have to have there's going to be a huge benefit over there. When you think of what we do for the partnerships, it all emanates and it all kind of originates from that desire to say the market is going to be scarce on infrastructure, and it's going to be constrained on three big things. The market is constrained on power. It's constrained on GPU and compute, and it's constrained on network. It's actually going to be hungry for trusted systems, right? When you think about those four pieces, you're going to build data centers where there's availability of power. It's going to be global because of sovereign requirements. You're going to need to have then, in addition to there being availability of power everywhere, one of the cardinal sins in this era is idle time of a GPU.
How does idle time in a GPU happen, both for training runs and inferencing runs? If you don't get the packet to the GPU fast enough, you're just burning money, right? Because you want to have these GPUs fully utilized. You have to make sure that high performance, low latency, power-efficient networks that deliver packets reliably from point A to point B as fast as possible, both on the back end and on the front end, are super important. Intra-cluster as well as intra-data center, kind of data center interconnect, is going to be super important. What we did was we have entered into partnerships with multiple different, kind of in multiple different areas. You should expect us to have more and more of these over time. A partnership with the Kingdom of Saudi Arabia.
Chuck and I had visited Saudi Arabia the week before the partnerships got announced. I literally landed back in California, and Chuck said, "Hey, the president's going to be back there again. They want you there with the delegation." I went back to Saudi Arabia. We were able to meet with His Royal Highness, MBS, and also with the president. We've essentially announced a partnership on a multitude of different areas. In the Middle East with Saudi Arabia, with the HUMAIN project. The good news is some of the leaders of the HUMAIN project are ones that have had experience with Cisco working in other organizations where we've built out the network infrastructure over there as well. That definitely helps.
In the HUMAIN project in Saudi Arabia, essentially what we are able to we are doing is helping them do the data center build-out, where we will be providing them the networking capability as well as the security and safety capability. A similar effort then is underway with G42 and Peng and Sheikh Tahnoon and that group of people in Abu Dhabi, where Abu Dhabi is one of the leaders in data center build-outs as well. What we're doing is working with them and ensuring that we can provide combinations of Cisco 8000, Silicon One, Optics, optical systems that we've been providing to hyperscalers, now also to these neoclouds. Basically, the Stargate project is an area that we have in the Stargate UAE project is where we're going to be working closely with as well.
I would say that the way that we want to do this is over time, as this model scales to different regions of the world, we want to make sure that we actually are participating in all the regions because I see this happening in every part of the world. You will see it happening in Africa. You'll see it happening in India. You'll see it happening in Europe. That'll continue. As those things happen, we want to continue to stay very engaged with all of these companies. The good news was a lot of these people rearranged their schedules and came and witnessed Cisco Live. It was actually great for them to see the breadth of the portfolio as we move forward.
There is a tremendous amount of learnings to be had for these neoclouds from a hyperscaler experience and also from our enterprise experience and how we can partner together and make sure that we can take this to the enterprises, enterprises start thinking about moving data centers and reaccelerating data centers also for the private cloud in addition to the public cloud for some of the inference requirements.
Sorry. Maybe for the last 10 minutes or so that we have, if I pivot here to security a bit, and maybe I'll combine, Jeetu, for you a couple of questions together just to make sure we get through some of them. A lot of positive feedback from distributors we've talked about Cisco recently on the security side in particular. I'm just wondering if you can sort of help us just walk us through the strategy here that's worked because what we hear from a lot of distributors are Cisco is now defining the market in security and use cases rather than what they used to perceive as a fast follower in the past.
Maybe just help us with what the key change in strategy here was and how do you expect sort of customers to then come in and adopt, sort of see the adoption curve going for some of the new security offerings that you've rolled out?
Yeah. I think the reality is when you undertake a replatforming effort like we did in the past few years, the first few years are met with skepticism until you actually start rolling out the products. The good news is the kind of people that we've attracted, in fact, we just hired one of the key leaders from Google who is now running our security business that used to be the Chief Operating Officer at Mandiant. Peter Bailey, who ran a large part of the business at Google as well. He's on with us as well. If you just think about firstly the team that we have, we've got people from VMware. We've got people from Google. We've got people from Microsoft. We've got people from Palo Alto Networks. We've got people from Zscaler. We've got startup CEOs like DJ.
All of these people have come on board at Cisco in the past couple of years to fundamentally rearchitect this because they see this opportunity of, "Hey, if we can fuse security into the fabric of the network, you can build something really magical because the telemetry that we have can really allow us to find the next scale of breaches and attacks that we are currently not equipped to do in the world." With that, let me just walk you through the security strategy at the highest level. We have fundamentally rearchitected this notion of firewalls, where rather than thinking about perimeter-based firewall, we now think about the concept of a hybrid mesh firewall.
Any kind of form factor that you have for a firewall, by the way, we just finished at Cisco Live a fully refreshed lineup of the firewall from the ultra high-end all the way down to a branch firewall. At this point in time, all of our firewall refresh has been finished. We used to have some quality issues in the past. Those are behind us. That's actually exciting to see.
What we've done in this hybrid mesh firewall is whether it is a firewall appliance, it's a virtual appliance, it's an eBPF agent through iSurveillance, it is a smart switch with a DPU on it, or now a CPU with the campus that's going to be Hypershield ready, whether it's Hypershield or it's anything else that we have in our estate, all hydrated with identity intelligence, is all managed through a single management plane. It's called Security Cloud Control. The first thing that we did is we said, "Rather than having a perimeter-based solution for firewalling, we wanted to make sure that we have a completely hyper distributed mode of firewalling where you have a distributed enforcement for firewall. You can enforce a firewall in a top-of-rack switch. You can enforce it on a server.
You can enforce it as an agent that's an eBPF agent that sits in the user space but observes every piece of traffic that terminates on a server and the host. That end-to-end kind of world of hybrid mesh firewalling was the first thing that we announced. The second piece was this notion of universal ZTNA, Zero Trust Access, which is least privileged access. We built our SSE product from the ground up that competes with the likes of Zscaler and Palo Alto. We are seeing tremendous success over there. What we have done over here is in that universal ZTNA, it's not just about a user securely connecting to an application. It is also about IoT devices like printers securely connecting to each other or to a user.
It is about agents connecting to agents and agents collaborating with humans and agents having computer use with applications. All of that is managed from security from the same exact platform, right? That completely changes everything. That is the second big area on security that we are focusing on. The first one is hybrid mesh firewall. The second is this notion of a universal ZTNA, which is we think that agents and printers are people too. We want to make sure that those get connected. The third big thing is fundamentally reimagining the SOC with our Splunk estate that we have as well and making sure that the SOC becomes agentic in nature as well. We had XDR and then RSA. We made an announcement of XDR being agentic. We are also making sure that that gets tied into the Splunk SIM.
That whole thing, all those three pieces, hybrid mesh firewall, universal ZTNA, and reimagining the future of the SOC to be fully agentic are the three things on the security strategy. All of our products that we have all get managed in a common management plane, all AI-first. We are solving two problems: using AI to fend off cybersecurity attacks and securing AI itself. Those are the two big problems that we are solving.
Got it. Got it. Seems like simplification across networking and security is a big theme overall. Let me just go through the last question here that I think I'll be able to fit in. A lot of investor questions have been on Splunk and the progress on the integration there, but more so what's remaining to be executed upon and how does it really feed into value-add for some of the platforms like XDR or Cisco AI Defense? Maybe just go through what's remaining for us to see on Splunk and how does it add value to some of the existing platforms. Thank you.
Yeah. That's a really good question on Splunk. Firstly, when we bought Splunk, we wanted to do this in three phases. Phase number one, do not affect the Splunk organic roadmap. Keep that going because it is a healthy business. It is growing well. Let's make sure that we keep that chugging along. Phase number two, which was we need to make sure that Splunk ingested data from a lot of different systems. The mechanism that they use to ingest that data is through this thing that they call a technical add-on or a TA. That technical add-on, Cisco had a bunch of technical add-ons. The market players have a bunch of technical add-ons. We want to continue to make sure that Splunk is open, right? It is working with everyone to ingest data because that is one of the key values of Splunk.
We wanted to make sure that the technical add-ons that we had from the Cisco estate were best in class. We upgraded all of those technical add-ons, whether it be from Talos and XDR and our firewalls, those are all upgraded. The third thing we're doing is saying, "Okay, it's not just about ingest. It's about federation. It's about making sure that the data can remain in place. We can federate." Most importantly, it is about ensuring that we take, instead of taking high-volume, low-fidelity data and trying to jam it into Splunk, which can be cost-prohibitive and scalability challenges when you think about the telemetry store as large as the network, right?
What we wanted to do was distill that data and only have the high-fidelity data that actually gets ingested by Splunk and actually gets federated analytics and federated search being enabled by Splunk. When you do that, you can have a very, very different kind of insight that can be had in place because now you know what's happening on the network. What we also did at Cisco Live, because we feel like one plus one equals three, is we want to make sure that when customers are Splunk customers, they are incentivized in using Cisco technology and the Cisco customers are incentivized in using Splunk. What we've done is we announced at Cisco Live that any telemetry that gets ingested from a Cisco firewall will be at no charge up to 5 GB a day into Splunk.
What that does is gives the Splunk customers a tremendous incentive to actually use the Cisco firewall. The Cisco firewall customers have a tremendous incentive to use the Splunk CIM, right? We have done that. In addition to that, the combination of AI Defense, which is under DJ for securing AI, the telemetry from there goes into Splunk. The Splunk sellers are incentivized in going out and selling AI Defense. We will actually see kind of common integration points between our Cisco technologies and Splunk technologies like XDR, which is a SOC for the kind of a capability for detecting and doing response on attacks that are happening and breaches that are occurring.
For those companies that might not be as sophisticated to have a SOC, we need to make sure that that XDR telemetry is also flowing into Splunk, who is for the most sophisticated of the SOCs that are there in the world, the financial services companies and the large E-commerce companies and what have you. We wanted to make sure that we provide this end-to-end perspective for customers so that if you happen to have a model where you have customers or suppliers that are not as sophisticated and you have customers and suppliers that are extremely sophisticated that have a SOC, you can now turn to one company for both those needs. You do not need to go to two different companies for those needs.
We have the most sophisticated and the most easy-to-use way to go out and get a network IT manager operational in being able to be a security SOC analyst because in most companies that are small or aren't as mature, the NetOps person and the SecOps person and the SOC analyst might not be three separate personas. They might be the same person doing it all. We want to make sure that the technology is simple enough on that front as well. DJ, any other add-ons?
I would just add that the fact was one of the key things we announced at Cisco Live was also the capabilities from an observability point of view to be able to do AI observability from within Splunk. When you combine that with AI Defense, it gives you a complete comprehensive view of it because token economics is a big topic of conversation. When you start thinking about financial operations, a lot of people are consuming these tokens back and forth. Being able to observe and be visible is really important. No, I just wanted to add that, Jeet. Back to you.
Yep. Anyway, I think we're out of time. So Samik, hopefully this gives your audience a little bit of an overview of what we've been up to.
Yep. No, this is great. Thank you both for the time. I'll just hand it to Sami in case he wants to close out with any remarks. Thank you to the audience and thank you to you two both for joining as well. Sami, over to you.
Thank you, Samik, and Jeetu, DJ. Thank you as well. As well, all other investors joining us. If you guys have any questions, please feel free to reach out to me on the Cisco Investor Relations page. Thank you very much. Everyone, have a good week and a good day.
Thanks, folks.
Ladies and gentlemen.
Thank you.
Bye. Ladies and gentlemen, thank you for attending the Cisco Innovation Tech Talk with JPMorgan. This now concludes the conference. Please enjoy the rest of your day. You may now disconnect.