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

Good morning, everybody. Thanks very much for joining us to this important call with Bill Gartner, SVP and General Manager of the Optical Systems and Optics Group. We also have my dear friend, Marilyn Mora on the phone from the IR team to keep us all honest. And I want to start, Bill. First of all, thank you so much for joining us today.

Happy to be here, Tal. Thanks for having me.

Thank you. It's -- optical is such an important -- has become -- from a very cyclical and "dangerous" area for investors to an area of solid performance and solid growth for many companies, and that's why I'm very happy to host you today.

I want to start with a general question, very, very general, just to set the stage because we normally -- when we talk about Cisco, we don't talk about optical, and we have the opportunity now to zoom in on this space. So I want to ask you, what is your optical strategy? And where does optical fit in the overall strategy of Cisco?

Sure. Thanks again, Tal, for having me. And before I begin, let me just say I'm going to be making some forward-looking statements, and I'd refer you to our SEC filings on the IR website. I think that's my safe harbor statement. So optical historically for Cisco has been a portfolio play for us to fill out the end-to-end portfolio as opposed to a stand-alone Cisco wants to be recognized as being #1 in optical. That's never really been the case. It's always been we wanted to fill out the portfolio to make sure we have an end-to-end solution for customers. That's not to say we don't win business on a stand-alone basis. We do. We win business when customers are not buying the rest of the portfolio. But in the main, we're positioning ourselves with packet solutions as part of an end-to-end play. That's been the history, I think, of the business. And it's helped us in things like Verizon wins where Verizon was looking at both packet and optical and wanted to have a strong player to support both. We participate very strongly in the MSO market, in telco market. Primarily North America is where our focus has been historically. And then more recently, our focus has shifted more to what was DCI-type applications in the web space and in the large SP space for data center interconnect. And that's where a lot of our investment has gone very recently and will continue to go. But I think as we look forward, the very exciting and interesting thing for us is really what's happening with coherent technology, thanks to ASIC developments that have allowed us to now drive power down, drive the density up and drive costs down. And the bet that we're placing going forward is that pluggable coherent technology in the form of things like ZR and ZR+ will become a very significant deployment model for many applications. And that then is -- provides us with a very, very good synergy with the packet solutions for Cisco, the routing solutions, in particular, where we can couple a routing sale and a routing customer with a DWDM pluggable. And I think that provides Cisco with a significant advantage relative to a pure-play optical solution that -- because, over time, I do believe that the pluggable plus router is going to become a very significant deployment model for many of our customers. Does that help at that level?

Absolutely. And I want to ask you even one more question just to frame out the position of Cisco. You mentioned pluggables, and we'll talk about the Acacia acquisition in a second. I want to ask you about the portfolio. Can you kind of -- big building blocks, just to explain what kind of products do you have for those that do not know Cisco's optical portfolio? What segment of the market are you playing in?

Yes. That's an easy question for me to answer. The -- and I'll try to keep it at pretty simple level. You can think of the portfolio as basically 3 major product families. One is our NCS2K, which is the DWDM workhorse, and that includes things like ROADMs, amplifiers, Raman amplification and as well as a full suite of transponders, everything from 10-gig all the way to 400-gig transponders. And that's sort of a classic DWDM solution. It's chassis-based. Multiple chassis can be coupled together to form a very, very large node supporting, for instance, multi-degree ROADM connectivity. And that's been the workhorse for the business for many years, and it's evolved very significantly over time with different generations of ROADM technology. I think we're on our sixth generation of ROADM technology. And obviously, as the wavelengths have evolved from 2.5 gig to 10 gig in the direct detect world to coherent in 40 gig and 100 gig and now beyond, it's evolved in the transponders as well. And the transponders do very basic things like convert a short-reach signal into a coherent signal, but they also have -- we have transponders that do very sophisticated aggregation of signals as well, both packet-based as well as TDM-based. So that NCS2K is sort of the workhorse product. That's positioned in metro applications, regional applications and in some long-haul applications. But that's had the most significant deployment is in metro regional applications, up to maybe 1,000, 1,500 kilometers. There's nothing to prevent it from going further, but that's where we've had most of the business. The second part of the portfolio is the packet optical transport solution. That's the NCS4K, and that is really a -- you can think about that as a packet plus OTN switch. And that is targeted for customers that want to have an OTN-based solution but may not want to deploy a separate network and a separate set of products to support that. So if you think about kind of layers in the network, there's a DWDM layer, which has things like -- which connects to the fiber and has things like ROADMs and amplifiers. There may be an OTN layer where signals are aggregated onto an OTN frame signal that then goes into the DWDM layer. And on top of that, you may have a packet layer. That's sort of a classic model, but it's a very operations-intensive model. You've got 3 different networks that you're managing, that you're planning, that you're operating, that you have different life cycles, different control planes. And for some customers -- -- for many customers, they don't deploy OTN at all. For some that do, merging the packet layer and the OTN layer makes sense. As an example, Verizon has done that. And so the NCS4K is the product that supports that application, putting packet and OTN together. And then the NCS1K is a compact modular portfolio, and that started out as DCI with a first-generation product. And now we actually deploy that product in what we call multi-haul applications, which means the product can be used anywhere from DCI through metro, long haul and even subsea. So it's a very versatile product. You can think of it as a transponder on steroids. And over time, that NCS1K will become the transponder as part of the 2K portfolio. So rather than having sort of purpose-built transponders that plug into a 2K chassis, a 1K can be a much more cost-effective, much higher-density, much more power-efficient solution as a transponder serving every application, from DCI stand-alone all the way to subsea. Those are the 3 major product families. And then, really, as part of the routing portfolio but with heavy influence from the optical team, we have TDM to IP, which are blades that go into the routing portfolio, that support taking a TDM -- classic TDM signal like a T3 or a T1 signal and mapping that to packet.

Got it. So next question is about Acacia acquisition, whatever update you can give us on that. And I want to ask kind of a general question. You're not buying your vendor for semiconductors for switches, but you did buy a company that provides you with the pluggables or the technology for the pluggables. Why here and not there? Why is that important to include it in your optical portfolio?

Can you -- sorry, so I'm not sure I understood the first part of that. Can you say that again?

So I said, Acacia, if you don't mind to provide us with just an update on the acquisition, whatever you can. And second, I want to understand the rationale for buying it, just to understand why it's important in optical to have the component integrated into a solutions portfolio, while, for example, in switching or routing, you're not buying Broadcom for their switch or routing semiconductor.

Okay. Okay. So in terms of status, we have completed regulatory approvals, as I think you know, in the U.S., in Austria and Germany. We are waiting on regulatory approval in China. That's proceeding. And we're in a waiting game and having significant interactions with [ Samer ], but our plans have not changed there. We're still anticipating that, that should close by the end of the fiscal year. And that is the last thing that we have to get done. Everything else is complete. And in terms of like integration, all the activities are proceeding as we would normally expect with a company that we're acquiring. Let me come to the question of why Acacia, and I just want to correct one thing about the silicon. The -- I think Cisco has really 3 pillar technologies, foundational technologies that we invest in, in support of the systems portfolio, systems being routers, switches, optical systems, cable systems, whatnot. And the 3 pillar technologies are software, no surprise, you've seen Cisco do a lot of things in software; silicon, meaning the switch and router-based silicon. So yes, we do buy from Broadcom, but we do a lot of custom silicon, and we acquired Leaba about 5 years ago very, very intentionally to get more control over that silicon for routers and switches. And so yes, we do that. We've invested inorganically in silicon as well. And now massive investment organically in that silicon to bring to market new routing and switch products, like the 8000 series that we announced in December. And Acacia, I'd say there's 3 rationales for Acacia that I would say. Number one is simply vertical integration. We were using Acacia in the optical portfolio, in the NCS2K, in the NCS4K and in the NCS1K. We were also using it in our routing portfolio in the form of pluggables. And vertical integration helps with margin improvement. So that's one rationale. That's probably not the one that would justify doing the entire deal, though, but it's certainly a sweetener for the deal. The second rationale is probably the more important one, which is that we do see a shift over time from chassis-based DWDM solutions like the NCS2K classic optical layer to a more integrated packet and optical solution where the router has a pluggable that's supporting the coherent interface. And that can become a very significant and dominant deployment model in many applications. And we felt that owning that technology was going to be important for us to participate in that market. And so that was a very significant driver. And the third one is really, as we think about the combination of what's happening in silicon, which is that the capacity of silicon has dramatically, dramatically increased to the point we're now delivering 14 terabits on a blade, going to 25 terabits, to 51 terabits over time, we're at the first point in our life where we're seeing silicon capacity is actually exceeding ahead of the traffic demand. That's never been the case before. And so now you can almost think of silicon capacity or router scale, router capacity as almost excess capacity. And with that, the cost per bit has come down significantly. That's one dynamic. And the other dynamic is this notion of putting a pluggable into a router. And with those 2 things in mind, you have to ask yourself, does that cause us to rethink how we build networks, what the shape of a network is, what the architecture of the network is. And we think that there's going to be network architectural shifts that we can effect by taking advantage of the fact that silicon has driven capacity of a router dramatically up, the scale is dramatically up and the cost per bit is down and the pluggable DWDM technology is now going to drive that cost down as well. We think that, that's going to shape network architectures. And so we think we can flatten network architectures pretty significantly with the combination of those 2 technologies.

Got it. I got a follow-up question from the audience, if I can read it. Are you concerned about the political environment between the U.S. and China and the impact on your ability to complete the Acacia acquisition?

Yes. I'd say, of course, the political environment is challenging now. We're still proceeding with the regulatory authorities. That has not entered into any of the conversations. But it's clearly in the back of our minds that the political environment can always create obstacles here. So yes, I would -- if I could -- had a magic wand, I'd like the political environment to be much more -- much less friction right now. But to date, it hasn't really entered into any of the discussions.

And is there a plan B? Meaning I've seen -- I was in -- I covered Qualcomm and I saw how they canceled the acquisition because they didn't get approval from China. Is there a plan B if you don't get approval?

I would say we are betting on the fact we're going to get approval.

Got it. Another question that I got was, will you continue to sell merchant DSP and support customers of Acacia and competitors of Cisco in China and globally?

Yes. And I want to say this as emphatically as I possibly can. We -- when we announced the deal, we made a very explicit statement that we would support existing customers, including those in China, maybe especially those in China. And that is a very strong conviction we have that we're going to continue to do that. It's a very significant part of Acacia's business today, and we want to actually expand that business. And so we're going to continue to do that. And we're -- we've had -- I have had conversations with most of their customers, and we're very sincerely committed to do that, to support not only the products that we're selling them today but the road maps that have been communicated by Acacia to those customers. And if I could add one other thing, Tal, because usually the skeptics in the crowd will challenge me on something there. When we did CoreOptics about 10 years ago, we had a similar situation where CoreOptics was supplying some of our competitors, and we basically said at the time that what we would do was we would honor the existing agreements but we would not seek to expand them or continue them. So we basically said, look, if the agreement expires in a year, we'll continue for the next year. If it expires in 3 years, we'll continue for 3 years, but we're done after that. We had no way at that time of supporting a component model in Cisco, and there was just no thought about whether that was a good thing to do. Where 10 years later, we've now made a commitment to support a component model for silicon as well as for optics. And that required a lot of heavy lifting, and still heavy lifting going on in Cisco to build the systems and capabilities to actually support that model, which is a different business model than selling fully integrated solutions or software. And so we're in a very different place and being able to say with a straight face that this is a business we're committed to support.

Got it. Bill, I want to take the discussion maybe one level higher in a sense that we spoke about your products, we spoke about the NCS1K and 4K. I want to take it to the services level, meaning what drives your optical business? What kind of services, what kind of carriers initiatives drive your optical business?

Yes. So as I mentioned, we're -- we've obviously got wins with Verizon and PLDT that I think have been publicly announced. Most of the MSOs we serve in the U.S. and many other smaller telcos and then web players. And I think if you look at each segment, there's a different thing driving it. In some cases, like as an example, for Verizon and PLDT, it was simply a network refresh, that the network that they had in place was older technology, they were not getting the economic value out of that, that they needed to, and they needed to do a complete network overhaul, and it really required like a nationwide build-out. And that happens, that continues to happen, and that will continue to happen where, at some point, the technology that was cost effective 5 years ago is now a burden and they're better off jumping to the next technology train. So that's one thing. Whether the -- and that may just be sort of business as usual in terms of the capacity needs, but their -- the economics, whether it's operations or whether it's the cost, the CapEx is going to drive them to a new technology node. So that's one driver. The second is I think we are seeing a shift that will take place over time of more traffic in the metro and edge markets, probably less in the core -- less growth, I would say, in the core as things like peering relationships move a lot of the traffic closer to the edge and to the subscriber where the content players are going to take the traffic off at the edge. And so we are seeing growth in metro networks. And then lastly, I'd say, in web, in general, that continues to be a very strong source of growth. They -- their demand is seemingly insatiable. And they clearly have a couple of different dynamics that they're dealing with, with the crisis that we're in, but there's push and pull there for them. So we're continuing to see very strong demand on the web players as -- in some cases, as folks move to a public cloud and take advantage of their services in the public cloud, but also just simple things like Netflix demand.

Got it. I think I know the answer for the next question, but I want to ask it regardless. What's the strategic importance of optical for Cisco? Meaning it's an isolated market. Most of the contracts or tenders are being done in isolation, it's just for optical. Where is the strategic fit? Why is it important for switching and routing and services and software company to also own optical?

Yes. So let me -- I'm going to expand that question a little bit, Tal. I think if you confine optical to say it's just the way we've always thought about optical, then it is more of a portfolio play where we want to have it to fill out the portfolio for customers that want an end-to-end solution and want one throat to choke. And that's historically been our play. But as we go forward, I think the broader question is around optics and optical and what role that plays in the network for customers. And as the bit rates have increased from 10 gig to 100 gig to 400 gig, what we know is that the relative cost of the solution has shifted more to optics. So at 10 gig -- and I'm talking even on the short-reach side now, not so much the coherent side, but at 10 gig, optic represented about 10% of the cost between a switch port or a router port and optic. 90% of the cost was in the port. At 100 gig, it's about 50-50 split. And at 400 gig, more than -- certainly more than half the cost is in the optic. And so that's one trend that says we better be relevant in optics for our customers if we're going to be relevant for them. We better be able to have a competitive solution in optics. And that's driven our investment in optics. It's driven things like the acquisition of Luxtera and before that, Lightwire. But the other trend, the other dynamic that we have our eye on is that there's this sort of unfortunate fact of physics that says, as you increase the bit rate, the distance that you get to send the signal gets shorter and shorter. You don't -- there's no free lunch. So going from 100 gig to 400 gig, you have challenges in sending that signal over the same distance. And we have a belief that at some point, the coherent technologies that are used exclusively in the optical systems world today, the DWDM world sort of outside the data center, that some of those technologies may have to find their way inside the data center or inside a central office even for short-reach applications. So the boundary between what was the short-reach world, conventional short-reach optics, and the coherent world, which was typically for metro and long-haul type applications, that boundary, we think, may blur. Now it's not going to happen at 400 gig, but it might start to happen at 800 gig. And for that reason, we felt -- we feel owning that coherent technology is important, and that has classically been part of the optical systems world, but increasingly it could be a very significant factor for even short-reach optics.

Got it. I want to ask you about the convergence of optics and IP. You touched on it in your answer. What are you doing to drive this convergence as a leader of IP?

Yes. And I think that is related to a previous question is like why is optical relevant for Cisco. We've had over the -- since as long as I've been with Cisco, which is about 10, 11 years now, we've had a belief that bringing optics and -- or optical and IP together made sense for customers. Now I think there have been some false starts with that, things like IP over WDM never really took off in part because there was always trade-offs that had to be made in one world or another in order to bring those 2 together. You got benefits, but there were trade-offs. But we still had a belief that there were long-term benefits if you could overcome some of those trade-offs. And with pluggable optics, the DWDM pluggable optics that are emerging now, the ZR and ZR+, they do eliminate all of the obstacles that we had in the past with what was known as IP over WDM, which were things like density trade-offs. In the past, if you wanted to put a DWDM pluggable into a router blade, you basically -- you took the density of that blade and cut it in half or worse. And that was the trade-off. And that wasn't necessarily a good trade-off for a lot of customers. It was, for some, the right trade-off, but for many, they said, I'd rather have the density. Now that is no longer the case. There's no density penalty with a 400-gig ZR or ZR+. So I think we're now approaching the point where bringing that optic together with the router makes perfect sense with no trade-offs that existed in the past, and it would make very significant economic sense as well. This is going to be economically a much better answer than deploying a discrete optical layer. And so at that point, that is, no question, cannibalizing the classic optical systems market. And because we play in both markets, we can make that choice pretty objectively and say, look, we're willing to trade off a loss in one segment for a gain in another segment. And we can do that pretty cleanly. So we do see bringing together IP and optics through pluggables is a trend that we're excited about. It drove our acquisition of Acacia, and we're going to be driving that into the market.

Got it. Cloud, cloud customers. Can you discuss your position with cloud? You recently spoke about solid traction of this group of customers. So can you give us more details on what are you doing with them and the history maybe a little bit and then the future?

Yes. I can't talk about specific customers, but I'm going to talk generally about the hyperscalers. So we initially -- so I mean, we've had a history with them when we deployed our NCS2K product initially. But that was really a poor fit for their application in part because, number one, the application itself was at the time a very short-reach application like 80 kilometer and 2K was sort of overkill for that. And secondly, the 2K was optimized for a central office environment and had all things like cooling that were optimized for a central office environment. That's very different than a data center environment. And so from a density perspective, from a power perspective, it was not really a good fit. But over time, that -- the portfolio for 1K is what we developed to really target that application. The 1K is -- you can think of it as a pizza box. It's compact modular form factor that is really optimized for a data center environment. So things like, it has front-to-back airflow, which is what you'd have in a data center server environment. And once you have front-to-back airflow versus something like a 2K, which has air that comes in from the front and then gets bended and like directed down or up the chassis and then gets bended again to send out the back, that's a very, very inefficient cooling mechanism. And in fact, you waste like 30% of the power just bending air. But that's the way you have to build things for a central office environment, whereas in a data center, you can go straight front to back. And so when you take advantage of things like different environmental standards in a data center, different cooling standards, you can build a much more purpose-built product that's much higher density, much more cost effective for that application. And so the first generation of 1K we sold into DCI applications, basically, this short-reach data center to data center interconnect, and we did that with many of the hyperscaler players. In the next generation, which, by the way, is using Acacia technology, the NCS 1004, we -- again, thanks to ASIC advances, we now could incorporate many more transmission modes, so not only short-reach applications like an 80 kilometer, but we can support metro applications or long-haul or subsea applications with the same ASIC. It's really now under software control. And under software control, we have 6,000 different modes we can actually enable that will optimize the solution for a given application, whether it's very short reach or very long reach. And so under software control, we can actually do -- provide the customer with significant flexibility in that platform. So the NCS 1004 is actually deployed in subsea applications, like crossing the Atlantic or crossing the Pacific. It's deployed in long-haul applications for web players, and it's deployed in metro applications for web players. So it's -- and DCI. So it's actually every application from DCI short reach all the way to subsea is deployed in applications today.

Got it. I only have 1 minute left, and I want to ask you an important question. In the last few years -- 2 years, you, Alcatel and Ciena reported good numbers. So optical is going through some kind of growth phase. Is it cyclical again? Or do you see something changing that will make the cycle longer? Everything is cyclical, but the question, is it 1-year cycle or 5-year cycle? So what's your view on the sustainability of growth in optical? And I talk about Cisco, but I also talk about the market as well.

Yes. I think there's -- I don't -- for any one customer, there's a cyclical behavior because the customer might, for instance, every 5 years upgrade a network. So we do see that as a cyclical dynamic. That's more customer specific. It's not a macro effect because every customer is on a different life cycle. I think there are trends like a shift to cloud that are driving a lot of the demand for web players right now. As long as that continues, I would expect demand to continue. I think there are trends like 5G, I think, will have an impact in terms of the -- and putting stress on existing networks that will require capacity upgrades, and we've seen some of that already. And I think this idea of terminating more of the traffic at the edge will put more demand requirements on metro networks. So I guess I would expect core networks to see lower growth over time. But I think there are trends out there like 5G and shift to cloud that are continuing to drive capacity deployment for optical networks. So as long as those continue, I think we'll continue to see growth.

Great. So Bill, great discussion. Unfortunately, we ran out of time as usual.

Yes. Thank you, Tal.

Thanks so much for the time and the effort to explain us the complexity of optical. And I did get more questions from the audience, so if you submitted me question, I'll try to answer them to the best of my knowledge. And if I don't know the answer, I'll approach Investor Relations of Cisco for more answers. Thank you so much.

I'm happy to help, Tal, and thanks for having me.

Thank you.

I hope everybody stays safe.

Yes. Thanks.