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

Hello, everybody. Thank you for joining. Tim Long here at Barclays. Happy to have Bill Gartner, who's the SVP and GM of Cisco's Optical Systems and Optics Group, with us this afternoon. Before getting into the Q&A, I'm going to throw it over to Marty for a quick disclosure.

This webcast is educational in nature with no new financial information disclosed. However, we will be making forward-looking statements. Those forward-looking statements may differ materially from the actual results and are subject to the risks and uncertainties found in our documents filed with the SEC, the 10-Q and the 10-K. Tim, I'll turn it back to you.

Thank you, Marty, for the theatrical reading there. Really appreciate it. Bill, maybe if we could start off kind of at a high level here. Optical and optics is an important business for Cisco but doesn't always get the fanfare of, say, routing or switching. So could you just give us a little of an overview of where you guys fit in the overall portfolio and a little bit about what kind of advantages Cisco brings to the optical space?

Sure. Thanks, Tim, and thanks for having me. Happy to be here today. I look forward to some of the questions. So let me just separate those because they are very different businesses and serving different markets as well. The Optical Systems business is a classic DWDM business, where the problem that it's solving is really how do you carry signals on fiber over relatively long distances. And it's basically what happens outside the data center. So once you leave the data center, you now have to send these signals over hundreds of -- to thousands of kilometers across cities, across continents, and in some cases, between continents. And in that world, there are really 2 attributes that are important. One is that the distances are long, which makes things challenging in terms of transmitting light. And the other attribute is that the fiber is constrained, it's a scarce resource. Meaning, if I have a customer like Verizon, for instance, who puts a router in New York and then puts a router in Chicago, I can't ask them to just pull a new fiber between those 2 routers. I have to take advantage of the fiber that's in the ground and put as many signals as possible on the fiber. So that's really the Optical Systems world. In that world, we use a technology called coherent, which is a fancy word for basically taking advantage of all the properties of light to send signals over long distances. And we have had -- we've been in the optical business really since 1999 with the acquisition of Pirelli and then Cerent. And -- but we have focused our activities in that area. Historically, we played primarily in what would be characterized as a metro or regional environment or application. But increasingly, with advances in technologies that move a lot of the function into ASICs as opposed to discrete solutions that would then require a purpose-built platform, historically, we've had to build a unique platform for metro, unique platform for long haul, unique platform for subsea. With advances in ASICs, we can now serve all of those markets with one platform. We call that multi-haul. And so our Optical Systems business addresses service provider customers, web customers, large enterprises and public sector customers who want to build private networks, or in the case of service providers, that is their business. And we often endeavor to sell the optical system with routing. So we make it an IP plus optical sale to customers who may want to have one throat to choke or may want to have one consistent services interface. But we also sell discretely, both routing and optical. So I have over 4,000 customers in optical that are deploying DWDM platforms. And we've historically served a pretty wide range of those customers. But we are largely concentrated in North America. We do have business outside North America, but probably 70% of the business is in North America. So that's the Optical Systems world. And on the Optics world, the Optics is a very different business. The Optics business is basically selling transceivers to -- I thought I had one here but I don't, transceivers to customers that are buying Cisco routers and switches and need to have these pluggable transceivers to carry the optical signal. And that's a very different world. You can think about that as sort of inside the data center or inside a central office where distances are short. And by short, I mean, typically, 10 kilometers or less. And the other attribute is that fiber is plentiful. So if we add a new router or switch inside a data center, you pull new fiber to every port on that router or switch. And every port gets its own fiber and every fiber is carrying only one signal. So that is a very different world than the Optical Systems world that serve with transceivers that look like this where anybody that can see the video. These pluggable transceivers go into routers or switches. And that business actually serves -- whereas the Optical Systems business is typically serving like service providers and web players, the Optics business for Cisco serves campus customers, enterprise customers, commercial, public sector, service provider and web, basically all of our customer base. So that is really a business that serves every customer segment for Cisco, whereas the systems business is primarily focused on service provider-class customers. So that's kind of the high-level view. I'm happy to go into deeper detail on any of those.

Sure. Sure. Maybe we'll start with the systems side. Can you just talk a little bit about what is that differentiates Cisco? I mean, obviously, you -- compared to most other optical vendors, you have routing and other products to cross-sell with. But what other kind of technology software vertical integration-type advantages does Cisco have on systems?

Yes. So one thing I would say about the optical systems market, and I've been in this market really for a long time, is it is a leapfrog game. And largely speaking, there is no sustainable differentiated advantage that any player has, largely speaking. If you look at the dynamics over time, one vendor gets an advantage and says, "We're going to do 100 gig," for instance. And then the next vendor comes out and says, "We're going to do 200 gig." And the next vendor comes out and says, "We're going to do 400 gig." And everybody's leapfrogging everybody constantly. So it is a business that requires constant innovation. But you're not going to get a sustainable advantage in the technology alone. It's just -- it's not there to be found. Whether you have that technology in-house or not can help you in terms of the margin structure, but you're not going to have a sustainable technical advantage. What -- where Cisco, I think, differentiates is, first of all, we do continue to invest in this technology so that we have leading-edge technology because that is the only thing you can do to drive cost down is to continue to innovate. So if you've got a signal that can travel at a higher capacity or a longer distance, you're generally going to drive down the cost per bit. But it is really a portfolio play. So for customers that either want to deploy with other products like routing products very often, or who want to have a similar operations environment, and for instance, a customer that might have standardized on IOS XR for their operations teams, we build now our products with IOS XR as the operating system for the optical product. So customers that have already adopted IOS XR and have built it into the operations environment see a very short leap to introduce optical products. So I'd say it's much more about a portfolio play than my baby is prettier than somebody else's baby. We try to play that game as well, but that's an unsustainable advantage you can gain there.

Right. Okay. Maybe similarly on the optics and the pluggable and the component layer, similar type of question there.

Yes. The optics world has -- is a little bit different for us in the sense that, historically, we have leveraged the supply base and OEM to optics. Effectively, we manage the supply chain for customers. And it's a very, very challenging supply chain, but we manage that supply chain for customers. We OEM the optics. And then we effectively resell optics. There's a lot of heavy lifting that goes between the point that the optic comes into Cisco and the time it might end up on one of our customers' hands in terms of qualifying that optic, certifying that works against all standards, making sure it works on all of our platforms. But it's effectively an OEM. That worked very, very well for Cisco when we were talking about 100-meg ethernet and gig ethernet and even up to 10 gig. And it worked well because the economics were such that the optic represented about 10% of the total spend between a switch port or a router port and the optic. Optics were about 10%. The port was about 90%. So it wasn't a huge portion of the customer spend. And most customers were comfortable basically buying optics as an accessory with -- from Cisco, knowing that Cisco would stand behind it in the event there was any issue across platforms or across vendors. At 100 gig, the optic proportional price or cost became about 50%. And so now the customer is spending about 50% of their spend in the port and another 50% in the optic. And at 400 gig, that'll be much more than 50%. And so at that point, and several years ago, we anticipated this and concluded that an OEM business model was probably not going to be the sustainable model for us to succeed in capturing that business. From a technology perspective, the industry is largely standardized. So a customer buys 100-gig optic, it's going to conform to some industry standard. It might be a CWDM4, an LR4, PSM4. And it complies with IEEE standards. So there's not a lot of differentiation that's possible there. But you have to work to get the cost out because cost matters. So we embarked, really, about 10 years ago, we started a journey of acquiring more the technologies so that we could own more of that value chain. I don't think we'll ever own all of it. We still have a very healthy relationship with all of our suppliers. We're going to continue to leverage them, but we will increasingly invest in building that technology ourselves, whether it's 100 gig or 400 gig, and owning that. And that's primarily driven by the business model needs, not so much by the technology needs. Though there is a technology element that we need to be there when our routers and switches come out with a 400-gig port, we have to have the optic. When they come out with an 800-gig port, we have to have the optic.

Okay. Great. Maybe just to follow on that, if you could talk a little bit about Acacia. I know it's not closed yet, but maybe talk about how that fits into both the optical business and maybe the broader Silicon One type of movement and what you think that will do to the value set for Cisco.

Yes. So I described 2 businesses for Cisco: one is the Optical Systems business, which we -- what I call DWDM business; and the other is the Optics business, the transceivers for switches and routers. And one common trait between those 2 businesses today is that they both deal with fiber optics and light as the transmission mechanism. And that's kind of why I own both is because there's some common attributes there and common suppliers. But there hasn't been a lot of crossover in those domains. I talked before about the data center, inside the data center using transceivers, these pluggable optics, and outside the data center using coherent systems, which are chassis-based solutions. Where Acacia plays -- historically, where Acacia plays is in that outside-the-data-center market. So for coherent applications where you need to send an optical signal on a fiber with many other optical signals, DWDM, and sending it over long distances. And Acacia, we have designed Acacia products into our optical systems in order to serve that market. So we have effectively relied on Acacia as a supplier to us for things like DSP technology to build our optical portfolio for certain applications. And so they have been a long-term supplier to us for optical. We've actually built some of their products into our routing portfolio as well. Now the big dynamic that is going to shift architectures in the network is that some of that technology that was historically delivered by a transponder, which is a line card in an optical system, a transponder's function is to take an optic from a router and convert it to a wavelength. Some of that function that's historically been delivered on line card is now going to be moved into a pluggable, a pluggable looks just like this, that's going to be put into a router or switch. And it's going to have a DWDM coherent interface. So in doing that, you eliminate the need for a short-reach optic on the router, a short-reach optic on a transponder and a DWDM option on the transponder and replace it with a pluggable that goes directly into the router. So now that classic Optical Systems business is going to be cannibalized at some level in order to move pluggables into the router. And so that's where the Optics and Optical Systems world crosses over. And obviously, my router brethren are very excited about the fact that we're going to simplify networks. Routing networks now won't necessarily have to have a chassis-based optical system sitting next to them to do the wavelength conversion. We can do that directly in the router with no compromises on density. There's still a need for an optical system to do things like carry the signal on a fiber, but the function for the optical system gets much, much simpler as much of that value shifts into the pluggable. And so that's where we work very closely with our routing teams on 8000 series router as well as the -- things like the 9K router and 5500 series.

Okay. Great. Great. Could you talk a little bit about routed optical networks and what Cisco is doing in that part of the market?

Yes. So let me just say what we mean by routed optical networks because that's not an industry term. That's a Cisco term. As we looked at really 2 technology transitions that are occurring here, one is that on the silicon, and Tim, you mentioned Silicon One. Silicon One is, for the first time ever, giving us the capability where the silicon capacity actually exceeds the anticipated traffic demands. We do lots of network modeling with customers, and we model things like what's the traffic demand going to look like for the next 3 to 5 years. And then we know that the system that we put in is good for 3 years or 5 years or 10 years or whatever. In most cases, in the past, the silicon capacity was such that we knew in that window, there's going to have to be an upgrade to the silicon capacity. The traffic demands were going to exceed it. Silicon One, for the first time, as we started to do this modeling, we find ourselves in a situation for many customers in many applications, the silicon actually has excess capacity, meaning we map out the customer traffic demands, and there's more capacity in the router than the customer ever envisions using. And so that's one significant dynamic. The other is this idea of moving the DWDM from a transponder into a pluggable like this. So you take those 2 together and we step back and ask ourselves, "Do those 2 dynamics allow us to rethink network architectures? Can we rethink the way networks are built by taking advantage of excess capacity in the router and the DWDM pluggable that's going to deliver 400 gig on a pluggable?" We know today that the way customers build networks that they use, they use a distinct optical layer to express wavelengths around routers, basically because routers could not afford to have those signals pass through them because the ports are too expensive and because there was no excess capacity available in the router to actually send a signal through the router, so you'd have to send it around the router. So we have that going on the DWDM layer. We have, for many customers, a distinct OTN layer. Similar arguments for why you would do that. And then a packet layer. So 3 distinct networks for many customers. Each one of which has its own operations, its own planning, its own life cycle. We think with taking advantage of the capabilities that are now available in ASICs, excess capacity effectively in the router and DWDM pluggables, we can flatten that to 1 layer that uses effectively hop-to-hop transmission of the signals through routers as opposed to around them, taking advantage of that excess effectively free capacity in the router. Using DWDM pluggables between the routers, that yields much, much higher utilization on the fiber, delivers very significant economic benefits to customers. And we see in our modeling somewhere around 35% to 40% savings for customers relative to present mode of operation. And that flattened architecture which takes advantage of DWDM pluggables and high-capacity silicon is what we call routed optical networking.

Okay. Great. That's helpful. Maybe just if you could touch on a few of the different verticals that you sell into, maybe start with the service provider side. What kind of trends are you seeing there? I mean, obviously, there's a kind of a steady speeds and feeds upgrade cycle. And they're trying to push to be more programmable, software-driven. So what are you seeing on the optical side from the service provider community?

Yes. So on the service provider community, there's always, always a need to drive costs down. And we do that in the DWDM world by driving down the cost per bit per kilometer, so being able to send a signal over a longer distance or send more capacity at the same cost. And so that's a constant innovation cycle, going from 10 gig to 40 gig to 100 gig to 200 gig to 400 gig and beyond. And that game is going to continue. I think that most service providers today are assuming that continues, and that's just kind of built in for them. And where they are focused is how can we simplify their network and reduce operations costs. The OpEx costs are anywhere from 4x to 5x the CapEx costs for most customers. And so if I look at how do we get control of the total costs here relative to revenue, which is not growing at the same rate as capacity, they've got to focus on OpEx costs. And we can't go to a customer and say, "Look, I'm going to go help you save OpEx, but ignore the CapEx." We always have to start with the CapEx because that's kind of a table stakes for the discussion. But increasingly, the conversation we're having with customers is, what kinds of things can we do to simplify the network? And things like routed optical networking, a big part of that conversation. What kinds of things can we do to automate functions that are error prone or require human interaction today? And what kinds of things can we do to simplify the installation and provisioning process? And so that's where a lot of our effort is and that's where a lot of the attention on the part of our customers is.

Okay. Great. And then maybe just hitting on kind of the more of the web scale, large cloud players. Obviously, cost is very important to them, and they also like vertical integration. So what are you kind of seeing from that community?

Yes. So for the web scale players, the dynamic is a bit different in that the web scale players had somewhat of a luxury of starting with a clean sheet of paper for many of them, whereas many of the telcos have 100 years of legacy, in some cases. And many, many networks that have come together over time, if they've been part of roll-ups, that really complicate their world. They have massive, massive legacy issues to deal with. Most of the web players are dealing with none of the legacy issues, don't have things like private line services that might be a customer requiring a T1 service to continue to be delivered that was delivered 20 or 30 years ago. So many of them have more degrees of freedom in terms of how they've built out their network. So that's one thing that's very different. And the other thing is, for the most part, the web players, because they came from a world where they were building out data centers largely with servers and storage, they started out in a world that was highly automated and did not assume that they were going to build a large labor force for operations. So they are very, very lean in terms of operations. And the way they manage is by automating everything. From the moment something is taken out of the box to the time it's finally retired, everything is fully automated. And so when we approach the web players, it's much more about how we fit into their operations environment and making sure that we conform to their operations environment. That may mean that we have to do things in the product to make sure that once you plug it in, everything is automated. Like the software that's supposed to be running in that is automatically downloaded, and it automatically detects that it's got the right software, it doesn't have the right software, and figure out what to do about that. And it automatically provisions itself with the right attributes. And then things like supporting NETCONF and YANG or OpenConfig models or whatever management model that customer might have, we have to conform to that. And so that's -- the mindset there is much more on operations and automation, and obviously, cost. Cost is kind of, again, table stakes for the web players. You're not going to get in there and say, "I can conform to your operations environment, but I'm coming at a high premium."

Okay. Great. Great. Maybe just squeeze one quick one in. Could you just talk a little bit about Cisco's participation in the subsea market? We've seen that be pretty strong with a lot of these big players building data centers all over the world. So what's Cisco's focus there?

Yes. So we -- subsea is a new segment for us. It's not a particularly large segment, but it is one that we are participating in, and we're participating in, I would say, fairly opportunistically. We've won a number of deals. We are deployed in a number of subsea applications right now. And the reason that we can participate in that market is because, I mentioned earlier, we have this capability now to support multi-haul, which is you don't need a purpose-built platform for subsea. You can use a platform that's serving metro or long haul and have that work in subsea applications as well. And so we're approaching that market fairly opportunistically, but very excited about the deployments we've had and some of the opportunities that we are now pursuing.

Okay. Great, Bill. I think we're running up against the end of our time here. Really appreciate the time. Thank you, Marty, for joining as well. Thank you, everybody else. Have a great rest of your day and stay safe. And we'll talk to you soon.

Thanks, Tim.

Thanks, Tim. Bye-bye.