Cisco Systems, Inc. (CSCO) Earnings Call Transcript & Summary

Good afternoon, everyone. I'm Samik Chatterjee. I cover hardware and networking companies at JPMorgan. For the next fireside chat, we have the pleasure of hosting Cisco. And I have with me Scott Herren, Chief Financial Officer. Scott, thanks for taking the time to be at the conference. Thank you to Sami as well from Investor Relations. Thank you.

Scott, I'll start you off with a few questions, primarily from the earnings report that you had last week. You reported last week with an update on stable demand trends and we are asking all of our companies to really comment about how they think about demand trends 12 months out. Just to offer your insights in terms of -- I know it's a difficult question and itself to forecast demand 12 months out. But how do you think about where we would be in 12 months in relation to demand as well as the magnitude of the change from where we stand today?

Yes. First of all, Samik, thanks for having us. It's always a pleasure to get a chance to spend time with you and with the group in the room. Yes, we just had our earnings call last week. And as a part of that, one of the things that we announced was total product order growth, which is a leading indicator. Obviously, of revenue growth. Total product order growth was plus 4%. That included Splunk because we had about 6 weeks of Splunk, right? That transaction closed March 18 for us. So we had about half a quarter of Splunk. If you net that out, product order growth on a more apples-to-apples basis was flat year-on-year. And that's an improving trend from where we've been. Because remember, we've had this disruption that hit the supply chain, and we had 3 consecutive quarters of 30-plus percent order growth while lead times extended and people had to buy ahead. And then as we solved those -- the kind of the supply chain, the individual constraints, we actually solved it really quickly and got a lot of products shipped out in third quarter a year ago. So our last week earnings call was our fiscal third quarter. So tough compared to a year ago where we shipped a lot of backlog out as we work through those constraints, this current quarter, we also shipped a lot of excess backlog out a year ago. So tough compare. Q1 will be the last in a tough comparison. But in the wake of that, once we got all that product shipped out into the field, what we saw is now that kind of we had this bottleneck at our level, and we got to ship out pretty quickly and now that bottleneck just move to the customer level and their ability to implement everything that we had shipped to them. By the way, they had ordered it all, but their ability to implement it as quickly as we would like them to was limited. And we said on our Q1 call as we took down our expectations for the year, we thought it would take a couple of quarters and then in Q2, it became clear to us, it was actually going to be more like the end of this quarter. So this is our fiscal fourth quarter right now. We track this pretty carefully, as you can imagine, given the impact it's had on demand. We track it pretty carefully, and it looks like the overwhelming majority of that product inventory that got shipped out as we cleared the backlog will all be implemented by the end of this fiscal quarter. So what that will mean for demand, it's obviously been a headwind. It's like, hey, I'm busy. I've got to implement what you've already shipped me before I start the next project. Starting the next project means now I've got to go ahead and place the orders for that. And our lead times have been back to normal for a couple of quarters. Backlog has been back to normal really since early in our second fiscal quarter of this year. So I think most of the perturbances from the supply chain are through the system, but now the compare points still reflect some of those changes. Feel good about -- I mean it's hard to say you feel good about flat demand. But given the trend, what we're seeing is as we were able to track product that was shipped out and we know serial number X left the loading dock on this date. When did it actually get set up and configured. And we know that, especially within Meraki because there's a cloud-based control plane for Meraki. We see those times coming back down to where they had been pre-pandemic. In pre some of the supply chain issues. So feeling good about that. Where we've seen it recover, we're seeing order growth return at the same time. And that's part of what you saw during the quarter with enterprise still being modestly down. Public sector growth, growth in service provider and cloud, growth in our security portfolio and growth in collaboration. So it was -- we're really seeing a positive trend through that.

Okay. Let's turn into campus switching and how orders also grew for data center switching and campus switching together in the quarter. Just give us more color in terms of what you're seeing on those 2 aspects, given that I think both were probably facing the same headwinds from the supply chain until the last quarter or the prior quarter, I guess, just give us more color what you're seeing both in relation to campus switching and data center and the drivers there, what is the customer really sort of now starting to reengage in?

Sure. Pretty encouraging a couple of data points given where we've been over the last 6 or 7 quarters to see those both return to positive growth, particularly in campus switching, I think, is the one that surprised a lot of people. I think data center network and growing. There's -- we've all seen. One of the big drivers of that, of course, are the big hyperscalers. And so I think that was probably less surprising than campus switching. And what's happening in campus switching, there's a lot of excess commercial real estate. I don't feel bad for anyone in the room that's in that space. But there is. But what is happening, and we're doing this at Cisco, too, is companies are now saying, you know what, it's time to get back to the office. Maybe not 5 days a week. Maybe I don't need the space that I needed previously, but for cultural reasons for, frankly, the training and the growth of the early in career people. When we survey our own teams and say, what's your propensity do you want to come back to the office, the people that are most interested in coming back to the office are people like me with a lot of gray hair and the people that are early in their career. Smiling, but I know it's true. And people that are early in their career because that -- it not only provides a little bit of a social structure for them, but it provides a bit of an apprenticeship aspect. I didn't learn to do what I do in school. I learned it on the job and I learned it from role models that I got to sit next to and ask questions. So more and more companies are saying, you know what, 2 days a week, 3 days a week, we want you in the office. And as they do that, it's work in the office, everyone is not going to be there on any given day. So the network intensity of the people who are there is going up. And so it does -- it seems a little bit odd that with commercial real estate having high vacancy rates that campus switching would grow, but it's really an outgrowth of what the office needs to look like in this world where we are -- more and more companies are saying, it's time for people to get back to the office.

Okay. I'll come back to that in a bit just to parse out wireline -- wired versus wireless. But since you mentioned campus switching, I think, and the recovery there, one obvious place that investors will go is how much of this is a benefit from a large M&A with some of your competitors versus actual market recovery?

Yes. I think -- so we clearly have seen some uptick in wireless. We gave a couple of stats on the previous call. We continue to see large orders in that space grow. In the WiFi and particularly in the wireless access point space. I think some of that is people looking at that acquisition and trying to understand -- I think what they're going to have to do clearly is merge the control plane for those 2. I can tell you because we've been doing this within our catalyst and Meraki lines, technically, that's super challenging to do. So it's going to take some time. And as it does, there's this which architecture is going to win out. And so I do think it's created some opportunity for us in the wireless space. It's certainly something that we've -- that we've gone after with our sales team. And going into those accounts that we know we're heavy with either of those 2 competitors who we've known for a long time. We've competed against them pretty effectively for a long time. Seeing them come together, I think has just created some more opportunity for us.

Okay. You didn't talk about wireless, I'll skip that. Now when we talk to distributors, resellers, a term that often comes up is every enterprise has a lot of technology debt that they haven't really addressed. And some of the headwinds we've seen in the last couple of sort of quarters or more have actually increased that. When you think about the recovery here, how are you thinking in terms of the drivers for customers to come out and say, okay, this is what I need to address. And which product areas in your portfolio, does that help? And maybe go beyond sort of the core networking to also help us address security and collaboration and observability within that?

Yes. I think technical debt builds up in a lot of places. You probably don't have any at JPMorgan, but other companies do that have built it up. And I think where you feel it is partly in your infrastructure, whether that's compute infrastructure, network infrastructure, just core systems infrastructure, you feel it there. You feel it in applications, lot of older applications that are not going to be retired, and they're going to have to continue to be run and that drives demand for things like observability and application performance monitoring, right? We're not going to unplug this app. It doesn't make sense to rewrite it. We're going to continue to run it and need to run it efficiently. You see it build up in data structures. And I think that's one of the benefits that we ultimately -- that will be a tailwind for us ultimately with Splunk is companies that have been around long enough to have technical debt, probably didn't sit down 15 years ago and say, here's a pristine data model that we're going to put in across the company and here's the specific owners and the governance around that. The data models just kind of grew as the company grew and so there's going to be some rationalization that needs to be done there. For us, what it means is there's an opportunity, of course, in core networking infrastructure and campus switching, as we've talked about. There clearly will continue to be opportunities in security. I don't think that's necessarily based on technical debt within security. I think it's based on the world we live in and some of the geopolitical stress that's out there that seems to be getting worse, not better at this point. That's going to drive -- continue to drive security. There is some cycles within the security business, firewalls that need to be refreshed and things like that. But I don't view that as normal refresh as opposed to technical debt. Within collaboration, there's no question that as people come back to the office, the way the office needs to be configured to work effectively in a hybrid world, which I think we're all going to be in for quite some time. There's no question that you need better devices. You've got to improve what that experience is like. If you want people to take the time to drive to the office to work, you need to make the office experience, one, that -- one of the things we talk about as having a return on commute. Okay. If I'm going to spend the time commuting, what's the return I get for that commute. And so I think there's clearly an opportunity on the device side within collaboration as well.

Let's switch gears here a bit and we are asking, obviously, as you can imagine, a lot of questions to our companies are about AI. But before we delve into the product set, what we also were curious to hear from all of our companies, including yourself, is how do you adopt AI internally? What can that drive in terms of improvement and where can it be a bit more tangible for investors to also sort of think of those improvements?

So what other companies should everybody invest in around AI.

Or how sustainable is this?

No, I think it's a great question. And there's a lot of use cases, obviously, just within the finance organization that we're looking at. Everything from account reconciliations, which are pretty basic to the way you prepare and kind of self-service data, right? If you think of what happens inside the finance organization of big companies, it's a lot of, hey, can you get me x. All right. Yes, I got to go full data from here and here and here and here and put it all together and then try to massage and is something that makes sense, that's going to lend itself very clearly to AI beyond RPA, which has already helped in that space. . But then if you zoom out from finance, there's huge opportunities for us to be more efficient in the 1-800 Cisco in our call center space, both on the inbound calls that we take. One of the big advantages, I think, of these large language models that have come out, I know you've played with it, I have 2 is there's a dialogue capability to it. It's not submit kind of a query and get an answer, there's a dialogue, and it's back and forth. And I think that lends itself, as you know, very well to customer support, customer service. We've just moved Liz and Tony, who was our Chief Strategy Officer, just moved into the role running our customer experience team, which includes customer success and our customer support teams. No one is better positioned to drive AI through that space than Liz is kind of given her background, she's engineering background. So I think it's a -- there's a real opportunity for us there. And of course, there's the opportunity to make sales and marketing much more efficient. I think everyone sees that as well.

Okay. Okay. Let's move to then the product side, and this is a big change for Cisco, particularly for the investment community, where Cisco's positioning for the upcoming cycle relative to some of the prior generations is materially different, materially improved. Talk to us about the impact you think AI will have on Cisco's product demand, how do you think about growth beyond the $1 billion of AI orders that you've talked about for fiscal '25.

Yes. The $1 billion is kind of the first step, right? And just to be clear about what that number is because you've asked me about it a couple of times, everyone has asked me about that. That is within AI, it's back end, not front end. We already have a significant business in the front-end networks. It's back end, it's Ethernet and optics in the back end, driven by largely the large hyperscalers. There's a couple of Tier 2 in there, almost no enterprise in there today, and we should come back and talk about the opportunity in enterprise as well. So that's what drives it. The way that works is they don't just buy off the price list, right? All of these hyperscalers have slightly different architectures. And what they do is they come and say, this is what I need. So they take their own network topology or their own network architecture and say, here's 3 or 4 different places that I need networking gear in here. I need it at the Spine. I need it at the top of Rack as an example. This is exactly what it needs to look like in terms of technical specifications, we go off and build that. And sometimes you have to build the chipset that goes in it, but we build that, ship it to them. They go through a full certification process. Once that happens, you hear us talk about design wins. Once that happens, they say, you are designed in on this use case. It doesn't mean we're the only ones design on in this use case, but you're designed in on that use case. And then as they look to scale up whatever they -- however data centers they're looking to scale up, they look at those particular use cases for networking gear as an example and say, who's qualified at each of these. And then what they want to do is have more than one viable supplier at each place, right? No one wants to have -- I don't want to have it at Cisco, but particularly the hyperscalers because of the magnitude of what they buy, they don't want to have single vendor dependency at any level across the board. And so if you're designed in at that level, then you get an allocation of the demand that they're looking for at that space. What drives the $1 billion then is clear line of sight to use cases where we've got design wins and with conversations with the hyperscalers knowing what that's going to look like for fiscal '25. So our fiscal '24 ends at the end of July, fiscal '25 starts at the beginning of August, we see that $1 billion in orders coming in during fiscal '25, more towards the second half than the first half, right, as they're working their way through. And of course, there's a lot of things that go into that. It's not just the networking gear, they need GPUs. They need electricity. They need facilities, et cetera. So the delivery is what will drive the revenue side of that, but those orders we see coming in, in fiscal '25 largely. I think the enterprise opportunity then for AI is on top of that. So that's the -- that's our AI opportunity to sell into the AI infrastructure build-out, which is significant opportunity. It's well beyond the $1 billion that we've talked about. By the way, our pipeline is significantly bigger than that. That's not total pipeline. We talked about pipeline being about 3x that number, $1 billion is where we have clear line of sight to those orders. The enterprise opportunity is much more nascent at this point. And I guess we'll come back and touch on that.

Yes. Maybe before we get to the enterprise, so you talked about the pipeline being 3x, how much of that pipeline is hyperscaler versus the Tier 2 cloud.

It's the majority, the large majority. I don't want to give you a percentage on that, but it is the large majority of that total pipeline. Not just...

And you've talked about wins with 3 out of 4 hyperscalers, but the way you're building up to that visibility into the $1 billion of orders also suggest you have a fair allocation or a fair view into what allocation or share you're being sort of promise. So in terms of visibility into that $1 billion of orders, how much more sort of new customer wins do you need versus you already have a line of sight and this is what's ramping, we have the share we're going to get to it easily.

Yes. Well, you're getting to it easily. It's $1 billion at the end of the day. But it's clearly -- it's more of the former. It's more we understand the demand. We understand what our allocation is going to look like. We understand what they need to buy, when they need to buy it. We understand our ability to deliver at those levels. So it's that. If new customers, just that's where you talk about the $3 billion pipeline as opposed to the $1 billion that we have line of sight to.

And before we go any further, what's changed in that whole sort of landscape because it's not a surprise to anyone that Arista did have major share gains in this area in the prior generation. What if you had to identify, is this saying, okay, we -- now cloud customers are more aware that they need to have 2 sources versus 1? Or is this just the product portfolio that is making sure that you have your fair share? Just walk us through that.

I'd say it's all the above. So if you go back to the original build-out of the big compute and storage infrastructure, so a decade, 15 years ago, as that was being built out. And this was under our previous CEO, but Cisco kind of had the mindset of no one knows more about networking than we do, so buy what's on the truck kind of mindset. And the hyperscalers kind of said that's not the way this is going to work, right? We've got our own designs. We've got our own architecture. We're going to build our own control layer on top of it. This is specifically what we need from you. . Scroll forward, so we missed that initial wave of build-out. Scroll forward, Chuck becomes CEO. We acquired a company called LIVA. We started the Silicon One strategy of building our own chipsets inside there and really went back to the hyperscalers and said, look, we'll meet you where you are. So if what you want to buy from us is the chipset and you want to build all the -- the rest of the switch infrastructure around it, we'll sell you just the chipset. If you want to buy a white box, we'll sell you the white box. You want to buy full systems. So you tell us how you want to buy, and that's what led to the pretty rapid growth that we had a little more than a year ago within our service provider business, right? If you remember, we kept giving you those statistics on the calls. So we actually ended up -- we were late to the game but ended up making pretty good strides in that space. Now the AI infrastructure build-out is beginning. We already have the relationships. We already have a very effective chipset that's designed for low power consumption. We have that -- we'll meet you where you are instead of business model. So we are far better positioned to catch the initial wave and then, obviously, the subsequent waves of the build-out of the AI infrastructure than we were when the initial compute infrastructure got built out.

Okay. Let's move over to talking about the enterprise AI opportunity. And as much as I want to sort of give you in that open-ended question of tell me how do you think about enterprise AI in itself. Also, what's interesting to me is NVIDIA, you announced the partnership with NVIDIA. To me it read like NVIDIA wants your expertise in accessing the enterprise market. So help me think if I'm interpreting this correctly as basically, your expertise in managing enterprise networks is what's being valued and the partnership is -- that's what you bring to that partnership as well.

Yes, that's right. It's their GPUs at the core, but it's built in a chassis and by the way, we're working through the architecture. So watch the space. We have our Cisco Live. Our big end customer event coming up the first week of June in Las Vegas, so kind of watch this space. But the architecture that we're working on, of course, has their GPUs at the middle, but they're built on to servers that are -- sit in our chassis with the Intersight, which is our -- inside the RAG control system. And our top-of-RAG ethernet switches on top of it. So kind of preconfigured. If you're an enterprise and you're looking for inferencing or low-level training, this is what you need AI in a box for you. NVIDIA came to us for a couple of reasons on that. One is what you just said, our expertise in managing enterprise networks and the significant footprint we have on the front end of all those, but also because of the reach that we have, the enterprise reach that we have from a sales standpoint made it attractive to them. So that's -- I think that's today nascent but a pretty substantial opportunity over time as the sensitivity of the data that companies use is going to drive them to. I can do some training up in the cloud. But if I ship my data up and train my model, others can leverage that same data and train their model on my data. I don't want that. It's not that it exposes your data as much as it allows other people to train on your data, data becomes a source of competitive advantage over time. So I think you'll see -- more and more, you'll see, particularly if there's sensitive data, you'll see those being run inside the enterprise, potentially even at the edge within the enterprise.

How do you think about the AI demand on enterprise from the enterprise segment carrying over to your campus products as well outside the data center.

Yes. Well, I think there's -- I think it's pretty clear that in a world where we're using a lot of AI to drive your business in whatever way, whether that's to drive your sales and marketing team or to drive your customer support desk or your finance team, however, you're using that internally, there's a lot of data that has to move. There's a lot of network intensity involved in leveraging AI to drive your business. So I don't think there's any question that, that network intensity accretes to a stronger growth segment for us in campus and in enterprise.