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

Good morning, everybody. It's Paul Silverstein, senior analyst, networking and communications equipment at Cowen. It's my pleasure to have with us this morning Cisco Systems. We have Jonathan Davidson who is Senior VP and General Manager of Cisco's Mass-Scale Infrastructure. Jonathan has done 2 tours of duty at Cisco totaling 15 years, and Juniper sandwiched in between. His current responsibilities, he leads an organization that builds silicon optics, hardware, software and systems innovations for cloud service provider and large enterprise, hence the name mass-scale infrastructure. And he's seen an awful lot in those 15 years. Before we start, I believe there's a standard disclosure statement, Jonathan, that you want to make. Is that correct?

Sure. Thanks, Paul. So as with most of these things, I will be making forward-looking statements. For full disclosure, you can go to cisco.com and read the full statement.

So to start off with -- why don't -- for those who aren't familiar, can you provide an overview of your roles and responsibilities at Cisco? And correct me if I'm wrong, but you've got routing, automation, orchestration, optics and you sell really throughout the spectrum in terms of very large enterprises, public sector companies building out mass-scale infrastructure, hyperscale cloud and service providers. But why don't I let you provide the overview.

All right. Well, you pretty much have got it all in there. The way I'd like to think about it is we've got a set of businesses around things like mobility, which predominantly is around the transition to 5G; the mobile core, which is the subscriber experience that you get -- that's how you -- the companies that you get your multiple subscriptions from are able to know if they're going to give you a free -- a pack that's like Netflix is free this month or something. We have to be able to track all of those Netflix bits so that they know not to charge you for them, as an example. We -- you mentioned optical systems as well as optics, the cable access market. So predominantly the devices where the service is substantiated, we also call the subscriber management software. You mentioned automation orchestration, and then, of course, our routing business and then goes along with that. So from a responsibility perspective, Cisco is a functionally organized company. So I'm responsible for the product strategy, product management and engineering for each of those separate businesses. And then, of course, I partner with my peers for sales, marketing services and the other functional organizations.

All right. So why don't we start out with big picture, Jonathan. Within mass-scale infrastructure and it covers a lot of ground as you just went through, what are you most excited about with respect to product portfolio, technology, customer opportunities? Let me just point out, correct me if I'm wrong, but on the web-scale side of the business, to my memory, I may be off here, but I think this past quarter was the first time I recall Chuck proactively calling out cloud companies in terms of becoming meaningful to the equation. I think you all have been very transparent in saying historically that you weren't where you wanted to be in that customer segment. And again, this past quarter, first time to my knowledge, Chuck called it out, which suggests that you all -- that it's either become meaningful or about to become meaningful. And then on the service provider side where you had a very good quarter, relatively speaking, in terms of the order book, I think it, only being down single digits, it had been down double digits for the past 2 or 3 quarters before that. And where I'm going with this is, your new 8000 series of routers, which I don't think you've commercially shipped yet, but that's based on your new Silicon One architecture, I would think that based on the specs that you all have announced, which appear to be significantly better than all of the computing platforms out there, if it works as advertised when that product hits the market, which best I could tell would be around the turn of the year, I would think that could have a meaningful impact in turning your service provider routing business from what appears, at the risk of being unfair, what appears to have been a share giver to becoming a share taker and thereby meaningfully improving the outlook for that service provider business, which is heavily weighted to your routing. But again, let me let you respond. I apologize for that.

All right. So you started off with what are we most excited about. So certainly, we've got a phenomenal portfolio from the smallest of routers like a cell site router all the way up to the largest platforms, which would be the Cisco 8000 router that we announced back in December. We started shipping that to customers late last calendar year. And to your point, official GA or general availability actually has happened now, both for our fixed platforms as well as for the modular platforms. So we have the highest density platform in the market shipping and generally available with that 14.4 terabits per slot and 10.8 terabits in a single rack unit. So that's certainly one area that we're certainly proud of, and that the team has done a phenomenal job in building that over the past 4-plus years. It takes a lot of innovation, a lot of focus, a lot of tenacity to just stick with a project for 4, 4.5 years to get to the market. But we knew that we had to start with the fundamentals and really build from the ground up so new silicon, new systems, new line cards as well as a new approach with our market-leading software. As you're aware, we have #1 market share in the service provider routing space, and we believe that this is the right product at the right time in the market. You mentioned the hyperscalers. They are always on the cutting edge, whether it's a deep 100-gig density or the transition into 400 gig. And we certainly believe that this platform is well positioned both from a hardware as well as from a software perspective. But also from a flexibility on how our customers can consume them. You called out a well-known fact. It's not a secret that we did not do as well as we wanted to in the first phase of web build-outs. And I think you can tie that back to really just not listening well to what was happening in our web accounts and hyperscale customers. And not only building what they need, but also changing how they wanted to consume the product. And so we have learned our lesson. We spent a lot of time listening, co-creating with them. This is not a short-term endeavor. This is a long-term transition not only for us, but for our customers as they go through and build new data centers or as they refresh their data centers. So it's not something that you'll see a spike in any given quarter. This is something that's going to grow over time. And as with anything, we have to stay on the cutting edge of innovation, and we have to stay in a mode of constant listening as well.

Jonathan, I want to make sure I fully understood the previous response. So you're making progress with the web-scale customers, you're listening better, you're responding to their needs. And I will say for what it's worth, that is consistent with the checks we've done and we knew about this in terms of better go-to-market, better responsiveness as well as better technology. But the piece I want to make sure I understand is the 8000 just went GA. I thought you said that the 8000 is well positioned for that web-scale market. If it is, I assume that's for a data center interconnect or routing use cases as opposed to intra-data center top-of-rack to leaf-to-spine. But perhaps I'm wrong. If that is the case, if you're already making progress, I would assume if the 8000's successful, that success would have come on top of the progress that you're already making, which is to say the current progress is not the 8000. The 8000 would be incremental. What is the case?

Yes. So we have had, I would say, a reasonable amount of success with our existing NCS 5500 plant platform family of both chassis and fixed platforms. As you're likely aware different service providers and different hyperscalers have different religions around whether they want to use modular platforms or whether they want to use fixed platforms. And so the good news is we built it all. So you don't have to pick a religion when you're toggling on different platform hardware types. And the 5500 has seen, I would say, a decent amount of success over the past several years across a number of use cases. And honestly, inside of most hyperscales, the concept of kind of the switch of the router was blurred a long, long time ago what they would consider -- what the industry would consider as switches are often routers. But in this case, the platform that we built does not go down to the top-of-rack level at this point in time. This is much higher up in the hierarchy of platforms. We are positioning it initially as a core routing platform, but from a density perspective because of the depth from a hyperscale perspective that does fit into a number of roles inside of the number -- quite a few of the hyperscalers outside of that top-of-rack role, but moving into a number of other roles. And generally, there's 5 to 7 different roles inside of most of the hyperscalers and some had more, some had a little bit less. But we will -- we think it fits into a good chunk of those roles.

And Jonathan, I trust the 8000 also addresses the service provider market, which is obviously pre web-scale was initial bailiwick for routing?

Yes. The last couple of months has, I think, highlighted the importance of connectivity to our homes and obviously the continued connectivity between businesses and cloud connectivity. And of course, the service providers are the tip of the spear in providing that level of connectivity. And so they are continuing to see growth, but this isn't new. For the last decade plus, we've seen, on average, 30% bandwidth growth on the core, 50% bandwidth growth CAGR on mobile infrastructure. And this is -- so there's really no sign of abating. We've also seen anecdotes from different customers where they have consumed 18 months of their access capacity in a period of weeks. We have seen others where they've seen similar on average -- more telling averages between 15% to 40% growth in their peering traffic in the first several weeks of the pandemic as the different quarantines or sheltering-in-place orders were enacted around the world. And what I believe service providers are trying to sort through is, is this the new baseline for bandwidth and it's going to grow from there or is it going to go down. And we've seen both around the world we've seen that be the new baseline and things continue to grow. But I think what's becoming apparent to many is that there is going to be a shelter in place that is -- in different parts of the world, is going to be treated differently, but that the work from home is going to be -- aside from certain other parts of the world, is going to last for quite some time. And where we saw initially different over-the-top video providers derating their traffic. That's only going to be acceptable for so long and they're going to have to get back to normal because customers are paying the same price they were paying pre-COVID for being able to watch their HD and 4K videos, and they're going to want to get back to that level of quality of video as well, which is just going to add additional traffic. One data point from one of my service provider customers is they saw 28% traffic growth on their infrastructure. But if the over-the-top video had not been derated, they would have seen a 43% growth in their traffic on their infrastructure. And so the capacity is there. And I think the majority of service providers around the world, their plans of building 50%, 75%, 100% capacity in the infrastructure, additional capacity, has done them very well during the pandemic. But there's also going to be a need for additional growth on top of that at that bandwidth that they've seen.

Great. Jonathan, let me just -- I just want to wrap up on the 8000. You've had this in customers' hands since the end of '19. So they've had now almost 6 months to get a feel for it. I recognize you're just -- it's just GA. So revenues will come. But can you give us any insight based on the feedback you're getting from customers, your expectations for how meaningful an impact this platform can have? It's a not insignificant portion of your business in terms of service provider web-scale routing. Any thoughts you could share with us?

Sure. Yes, just like any other major platform introduction, you start with a specific use case. And then over a period of time, I'd like to call out kind of 24 months, you start to add additional use cases. And even in the use cases, which we've already deployed, there are often individual features, handfuls of features that a service provider might need for their -- how they just happened to run their operations environment. And so we have to make sure that if they want to deploy in service provider A that we are able to go and make sure they have those handful of operational features so they can actually get it into their infrastructure. This is not new to us. We know all these things. And for the most part, we know what those features are for every service provider. But they're always evolving, and so we have a very tight interlock process with, I'd say, the top 50 who are deploying routing infrastructure from Cisco. And we want to make sure that when they go to deploy, they have a simplified experience. I will add one additional thing. Not specific to the 8000, but we launched along with the 8000, which what I just described to you of getting new software, getting new platforms introduced traditionally has been a very long process. And what we want to do is we want to be able to decrease what I call the time to value. The time -- it's not my time to value. It's their time to value like how long does it take for them to get the platform into their infrastructure and then actually get it integrated and then deployed so they can get the benefits of all this new technology. So we've developed a cloud-based service that enables them to actually upload a configuration from one of their devices. And we worked dynamically on the fly to build a test plan for them using machine learning, which they can then use to test inside of their own labs or they can test in the cloud with some level of limitations or maybe even use that to test in our own Cisco labs as well if they so desire. And this has helped certain customers adopt that -- automate that testing experience. Now this is the first time we've gone through it. So it's not like they're getting all the benefits. But we expect over the course of the next 12 to 18 months as customers start to adopt this, that newer releases of software with newer features and newer use cases will be able to be more rapidly adopted because of this innovation.

That's a perfect segue to your optics strategy in Silicon One in terms of flexibility and time to market. And I apologize, I just want to make sure, is Silicon One within your purview?

So Silicon One is definitely one of the things that I call my babies.

Perfect. So my understanding of your Silicon One strategy, which was announced late last year, not surprisingly, Microsoft and Facebook, among others, they were showing interest. Again, not surprising. But my understanding is that is an architecture that enables you, Cisco, to relatively rapidly bring to market different silicon for different use cases in products such as the 8000 series, your first products based on that Silicon One architecture, the first silicon launched, but that's very much a time to market and a very important core competency in terms of a building block. You've always had ASICs, but this is an entirely new architecture. Can you share some insight about that before I ask you about your optic strategy?

Sure. So high level, this was part of the Leaba acquisition 4.5 years ago. So Eyal Dagan who leads that engineering team as well as the hardware engineering for Todd Nightingale and myself. So for those of you who don't know Todd, Todd leads our -- basically, our enterprise and compute divisions. We call them basically intent-based networking, the data center all sit kind of in his purview. And so we have a centralized hardware and silicon team, so both Todd and I can leverage the synergies for that. Specific to Silicon One, it is a completely clean-sheet architecture. As you are likely aware, there's different blocks inside of silicon. And no blocks were reused in the making of the silicon. Literally, everything was built from the ground up, from scratch. And what that enables you to do is be extremely efficient and be very thoughtful about every single block that you are building. And what we have seen is, from a design point perspective, is the goal was to get the bandwidth scale as well as the overall power savings efficiency that you typically would get with what is called switching silicon. But we also knew that this was going to be routing silicon. So we wanted to make sure that you have the high-feature content as well as the ability to do deep buffering so that you're actually able to have it utilized as a router. And so in my opinion, we have definitely hit the mark where we have the most power-efficient routing silicon in the entire market. We have got the highest performance from a bandwidth perspective in the market. And I believe we have really knocked it out of the park from that perspective, and we're seeing interest from our traditional service provider customers, from hyperscale and web customers and then, of course, you mentioned earlier, even from public agencies and anyone who view themselves as building critical infrastructure or has a need for significant bandwidth and capacity.

Right. So correct me if I'm wrong, but I think since Cisco started paying a dividend in 2012, of the dozens and dozens of acquisitions that Cisco has consummated, I believe only 2 were hardware-based and those have been done in the last year, one of which is still pending. And both of those, of course, for optics deals. First, Luxtera for silicon photonics focused on intra-data center switching applications; and now with Acacia, we're coherent DSP, more telecom or distance oriented. Again, that hasn't yet closed. Can you talk to us about your optics strategy? And before you respond, I'd note there was a slide that you all put up recently, and I apologize, I think this was at an industry analyst event. But what you did show, not surprisingly, that optics is an increasing portion, increasing percentage of customer spend on routing and switching platforms, which I assume is part of the reason behind the strategy. What is -- once upon a time, you all bought Cerent and you all bought Pirelli to get into the optics business? And then it seemed like for many years, you downplayed optics before coming back to it in the last 2 or 3 years. What's the thought process behind the renewed focus?

Yes. So a couple of things. One, we also have acquired Leaba 4.5 years ago. You could consider that a hardware company although they were preproduct at the time. And to your point, from just purely a service product perspective, we've made 2 additional acquisitions, mostly in the software space to help with what I call the hybrid connectivity as well as more around the automation orchestration area. Specific to optics, you bring up a really good point and this is one of my favorite slides, which basically shows the amount of cost to move a bit from point A to point B and how do you go back 10-plus years ago, 10% of that cost is in the optics and today it's about 50/50. And if you move forward, we believe it will be likely over 70% especially when you're talking about digital coherent optics of which Acacia specializes in. Now this is a rule of thumb. These are not -- don't go and build a business model based upon that -- the high-level statistics because, obviously, there's short-reach optics, kind of medium and there's long-reach optics. Obviously, the longer the reach, the more expensive the optic itself gets. So when you think of this, 2 things. One, we actually acquired an optics company a while back. I don't know exactly. It's probably pre-2012 with CoreOptics so we've been building...

And Lightwire.

Lightwire. Yes. So this is -- so we've been building optics internally for quite some time. And what we've seen is that there's obviously a lot of different types of optics. I mentioned short-reach, kind of medium and then long-reach. There's also QSFP, there's different form factors, CPAC. So you have to build a pretty broad range and there's pretty seminal technology that you really have to forward invest just like you do with silicon. And our belief is that we need to make sure that we own the core intellectual property around some of these key areas inside of the optics environment, not only because of what you mentioned around simply the amount of the TAM that's moving towards optics. But also, we see technology transitions where the optic technology and silicon technology is going to become closer and closer over the coming generations of time. And I don't have a Magic 8 Ball. I can't tell you exactly when that's going to be. But we know that it's going to happen. It's just a matter of time, and we need to make sure that we've got all the different pieces of the puzzle. One example of optics getting closer, and there was a press release and just in the last couple of weeks, talking about how our Luxtera team actually was partnering with one of the ODM players to do an onboard optics play to make it easier to cool the device, to make it -- if you know you're going to use all of the ports from day 1, there's no reason to use pluggables, right? You might as well just go ahead and put the optics on board and see if you can bring cost efficiencies out of that, operational efficiencies out of that. And there's -- and I think that we're still early with onboard optics, but it was a good learning experience both for us and for the other vendors who were involved in that program.

Well, so to your point, I mean, I saw the Luxtera announcement, our first take was this is more about silicon than it is about optics, i.e., silicon optics eventually coming together on the same die, on the same board addressing the distance issue. But -- so the slide that you referenced, increasingly, optics is a bigger portion of the equation of TAM. But I know time is short, and let me move on and return to the big picture question. We discussed the 8000, we discussed optics, we've discussed the Silicon One strategy. What else are you most excited about that can have an impact on Cisco's business for the better, if not now, then in the coming years?

Yes. So I would call out 2 specific things. One is around the transition to 5G. We certainly see that this transition is happening around the world. You need to have an alignment of a few key elements. First of all, obviously, you need to have spectrum. It needs to be available. It needs to be reasonably priced, and I think that has been a challenge in a number of countries around the world. The second thing is you need to have handsets ready to go. And while I think Samsung has done a good job of that, there's definitely a dearth of 5G-enabled handsets out in the world. And then -- good, you're actually going to make some pretty big technology decisions around your mobile core provider, your RAM provider to see that there definitely are more service providers looking at virtual or O-RAN-enabled technologies than even 12 months ago. So we think the disaggregation and opening up of the interfaces -- more standardized interfaces is good for the technology environment, good for silicon customers and will create more choice, which usually accelerates markets. And then, of course, our -- the service providers need to go and build out new transport infrastructure from IP as well as from an optical perspective. And so we have been working closely with a number of them. You might have seen in the press release a couple of months ago now with NTT that they had selected Cisco as their primary backhaul supplier for their new 5G network. And as you know, they've been on the cutting-edge of 5G in Japan for quite some time in preparation for the Tokyo 2020, now Tokyo 2021 Olympics. And certainly, we're tracking what's happening with 5G around the world. And then the last thing I would say that we're excited about with relation to the 8000 is just the transition of architectures, what ZR is going to bring into the fold and how that's going to enable a simplification of the IP and optical networks, including the transponders without really a tax directly into most of our service providers. And that will remove network elements, that saves power, that creates a more automatable network and so we've seen early interest in that technology transition as well. Obviously, there's some Z optics start shipping at scale and so that's likely to happen towards the end of this calendar year.

Jonathan, in the last minute or 2, I think many investors may not know or may have forgotten, but way back when, I've lost track, it probably was a decade ago, maybe longer, you had what for many years was an extremely successful acquisition. If I remember it was Starent, which brought you into core mobility. And in LTE, they did extremely well for a while. We don't -- the name, it's now part of a large organization and I don't think we've heard a lot about this, and it doesn't seem that it had -- I don't know if it lagged over the past several years, but you just referenced 5G both, the transport piece, IP routing and optical, as well as the core mobility piece. Any update you could provide with respect to your competitive positioning? Are we going to see? Is there the prospect at, again, at the risk of being fair, that you'll do meaningfully better in 5G core than it appears to have been the case perhaps, again, at the risk of being unfair, over the last couple of years.

Great question. So Starent is now our what we call the Cisco Ultra Cloud Core. And when we acquired Starent, the industry was all about appliances. And since that time, it's moved pretty dramatically to be a software business. So initially, it was a virtual software and now we're well enough got to cloud-native software. This is what I described at the beginning of the call around that transition to 4G, 5G, you need subscriber management software. That subscriber management software is the mobile core, whether it's a 2G, 3G, 4G, 5G. There also was a blog and -- I'm not sure if it's a press release or just a blog from T-Mobile and their Neville Ray there talking about our partnership around helping them build not just a 5G NSA network, which they actually started offering publicly towards the end of the last calendar year, but also they announced that they were planning to be the first to deploy a 5G SA infrastructure. And that's built upon Cisco's mobile core. And if you go and look at who's deployed what around the world, this, as far as I know, no one has disputed this yet, I track it pretty closely, it's going to be the first 5G SA core deployment in the world. And so we're excited to be able to partner with T-Mobile on doing that. And clearly, with this current acquisition closing, there's going to be more subscribers on that network over time.

One last question, and I apologize, I know we're over time. I'll try to keep it short. The U.S. government has made -- has been very vocal in connection with geopolitical vis-à-vis Huawei and China in terms of championing how important 5G is. Obvious thought is there aren't a lot of players in the U.S. when it comes to 5G. At least at the systems level, you're one of the very brief few. Any thoughts you could share with respect to the U.S. ION initiative or anything in that vein in terms of 5G and U.S. push? Anything you can share with us?

You were breaking up a tiny bit there, so I'm not sure I got all the right. I think I got the gist of it. So the basics around 5G and where we're at, there are a few smaller radio vendors inside of the U.S. that it will be building -- or are building O-RAN, DRAM software or they actually are building O-RAN, DRAM software plus the radio themselves. Obviously, we've got some of the key seminal technology with large companies like Qualcomm that do actually build the key components for 4G and 5G infrastructure. We certainly believe that we built some of the key elements around transport and then, of course, from a mobility perspective, what we're building around the mobile core, and I mentioned what was happening with T-Mobile. We certainly think that -- and the government has called us out that there is a need to help the more rural telcos migrate from their existing deployments to, I would just say, goes to more in the Western hemisphere, and that's the initiative that the government started and we're certainly here to help them as much as possible with that effort. And so we're going to stay laser-focused on that and see how the industry continues to evolve. And I've said this before and I'll say it again, we stay focused on listening to our customers. We stay focused on innovating with kind of long-term differentiation in mind, and that's our key focus areas for this space. And what happens kind of on the geopolitical side is outside of our control. And so we -- personally, we don't spend a whole lot of time, doesn't impact our innovation agenda from that perspective.

Understood, not surprising. I'd love to have another hour or two, but I'm afraid we're going to have to leave it at that. I appreciate the time and insight this morning. I hope you and everyone on the line today, I hope you and your loved ones and friends, family, et cetera, I hope everyone's safe and sound. Thanks, Jonathan.

Same to you. Thank you, Paul.