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

Perfect. Thanks a lot, Katie. Good morning or good afternoon to everyone depending on where you are. My name is Amit Daryanani, the networking equipment and IT hardware analyst here at Evercore and delighted to have the opportunity to host Cisco's tech talk on the optical space over the next 40, 45 minutes. We have here with us, Bill Gartner, who is the SVP and General Manager of Cisco's Optical Systems and Optics Group. We also have Raj Shanmugaraj. Hopefully, I did not kill your last name, who was the President and CEO of Acacia, prior to its acquisition by Cisco and is now the VP and General Manager of coherent products at Cisco. In addition, we have Carol Villazon from the IR team as well from Cisco. Now before we get into this entire discussion, just a couple of items in terms of lay-of-the-land, we will keep this discussion around 40, 45 minutes. Bill is going to kick this off with a few opening remarks and touch on with the slides he has a bit of a one-on-one on Cisco's optical business. I will then go through some of the questions that I have for Bill and Raj. [Operator Instructions] But before we do anything further, I will turn this over to Carol to go through the safe harbor statements, Carol?

Thanks, Amit, and thank you, everybody, for joining us today. So please note, we will be making forward-looking statements, which are subject to risks and uncertainties, which are outlined in our SEC filings, which can be found on Cisco's IR website at investor.cisco.com. And the actual results or events may differ from those forward-looking statements. No new material information will be discussed during the webcast. And with that, I will turn it over to Bill to go through his slides.

Well, thanks, Carol. And Amit, thanks for hosting us today. We're happy to be here and look forward to the questions. I want to just start with a few opening comments on where we play. Amit introduced me as running the Optical Systems and Optics businesses at Cisco. And I just want to explain what that means and kind of introduce some of the insights that we've had around where these markets are going. So just to frame where we play on the left-hand side is what happens inside the data center, intra building. So you can think about it as inside the 4 walls of a data center or a central office if we're talking about a telco. And that's where we refer to optics. Those optics are transceivers that plug into routers and switches and that defining characteristics of that world is that fiber is plentiful, meaning if you add a new router or switch you pull new fiber to every port on the router or switch. And the other characteristic is that the distances are short. In this world, short is less than 10 kilometers. And that's the world that Luxtera plays in, and we'll talk more about that. And on the right-hand side, what happens when you leave the data center, now we have to move these signals across the metropolitan area or across a region or a country or even between continents. And that's where Acacia plays. And in that world, it's characterized by very different attributes, fiber scarce, meaning that if we add a new router in L.A. and a new router in New York, I can't ask to pull new fiber between those 2 routers. That's a multibillion dollar problem. So in that sense, fiber's scarce. We have to leverage the fiber that's in the ground and we do that with technologies like DWDM that allow us to put many signals on one fiber. And the other defining attribute is that the distances are long, meaning more than 80 kilometers. And in that world, we serve that market with chassis-based solutions, as you can see in the graphics. Chassis-based solutions and pizza boxes that may be more optimized for data center interconnect. And the interesting thing that's happening is that we're seeing some crossover now where that coherent technology is actually finding its way into pluggables and that creates some interesting architectural options for us. Now you may ask like why does Cisco care about optics? And I think this graphic is probably the graphic that I showed the Board 2 or 3 years ago and said, this is the reason that we need to be thinking about vertically integrating more of our optics. At 10 gig, optics represented about 10% of the spend. The port on the router or switch was about 90% of the spend. At 100 gig, that ratio shifts to about 50-50 at 400 gig more than half the spend is in the optics. Now we have many, many customers who count on Cisco to give them a fully-integrated solution, router, switch, optics, software, everything. And if we're going to be competitive and be able to continue to serve them, we need to be able to offer optics at competitive price points. And we felt that we needed to do that at a competitive margin for Cisco as well and vertically integrating that technology made sense that we're going to remain relevant for our customers. Now the other insight, which is a little more subtle, is that if you look at the Y axis here, you can think about that as capacity on a given signal, a given wavelength. And the X-axis is distance, 2 kilometers all the way up to 80 kilometers and beyond. If you think about 2 kilometers today, we serve that with direct detect technology, which means if the lights on it's a one, lights off it's a 0, and it's served with these transceivers that plug into routers and switches. And that's probably good for 100 gig, 400 gig, 800 gig. Now if you shift over and look at the 80 kilometers and above, that is served with coherent technology like Raj's team at Acacia developed. That's very sophisticated technology that's designed for long-distance applications. What we know is that physics conspires against us and physics says as the bit rate gets higher and higher, the distance gets shorter and shorter. So -- you can't send a 400 gig signal as far as your kind of 100 gig signal. And you can't send an 800 gig signal as far as you can a 400 gig signal. And what that suggests is that we might need this more sophisticated coherent technology as the bit rate increases, even for shorter-reach applications. Like we might need coherent for a 10-kilometer application at 1.6T and we might need coherent for a 2-kilometer application inside the data center when we get to 1.6T. So that suggests that coherent technology is going to be increasingly important, not only for those applications outside the data center, but maybe even inside the data center. Now Cisco has made a lot of investments here. We started our investment in optics actually back in 1999 with the acquisition of Cerent. But more recently, CoreOptics was our first generation of coherent technology. Lightwire was our first generation of silicon photonics. Leaba was acquired for next-generation silicon to be used in routers and switches. And then in 2019, we acquired Luxtera, which brought a second generation of silicon photonics technology, and in particular, automation of the manufacturing process and Acacia really brings together coherent and silicon photonics technologies. And so that's been our recent history in acquisitions. We spent almost $6 billion. And obviously, we're continuing to invest internally. And then, finally, I'd like to say that while we see a fantastic shift occurring in the optics and optical world as some of these coherent technologies find their way into pluggables that can go directly into routers and switches, we also look at what's happening in the router itself and with silicon that's expanded capacity dramatically from what was 3 terabits not too long ago. 3 terabits in a single line card is now 10 terabits, growing to 25 terabits. That drives down the cost per bit, drives up the capacity per port. But as we start looking at that as a -- at the network layer, we ask ourselves, with the combination of DWDM in pluggable and these very, very high-capacity ASICs, should we be thinking about different architectures in the network? And the answer is yes. And we think that with the combination of this massively scalable router, and a DWDM pluggable, we can vastly simplify the networks that are being offered today, and we call that routed optical networking. And that's basically saying today we have a DWDM layer, an OTN layer and a packet layer. Those 3 layers can really be collapsed into a single layer, an IP layer supporting what was private line services being delivered by an OTN or DWDM layer, and carrying all those signals on an IP layer using point-to-point connections with these DWDM pluggables. Now we have to ask, why did we end up with 3 layers? And the answer was, at one point, 20 years ago, the router was the most expensive resource in the network, and we did everything possible to bypass routers because you didn't want to go through them because those router ports are really, really expensive. But as this silicon capacity has expanded dramatically, the cost per bit on the router has come down and has come down much faster than the optic. So now the router is no longer the most expensive thing in the network, and we can start thinking about why are we bypassing the router with DWDM layer, an OTN layer. Let's go through the router, because the router is not the most expensive thing. In fact, it's becoming the cheapest thing in the network. And so let's leverage that and rethink architectures. And that's really what brings us to our routed optical networking architecture, which delivers significant economic value for customers. So with that brief introduction, hopefully, I've given you some things to think about, and we'll be happy to take some questions that you may have now.

Perfect. That was actually a really good overview on optical space and what Cisco is doing. I had a few questions for you. Before I get to that, maybe I want to start off this with OFC just wrapped up a couple of weeks ago, it was a worthful event, but there's a lot that was announced, I think, from Cisco Acacia's perspective on 400 gig plantables getting ready for deployment to coherent and access to networks. Maybe I want to spend a minute if you just highlighting what were some of the key points takeaways from the OFC announcements we had a couple of weeks ago?

Yes. So one thing I would say is the OFC was somewhat tempered, I think, given the virtual nature of it. But we were very active at OFC in some of the sessions as well as the blogs and some of the white papers that were distributed. We have a pretty strong belief that the big shift that's going to occur in the industry is driven by this coherent pluggable, the ZR pluggable. And I think a lot of the investment is going to be going into that in the industry. There's always going to be a debate is that the only answer. And we're not saying that's the only answer. There are transponder-based solutions, chassis-based solutions that will continue to be used for things like subsea and some of the long-haul applications. But we think that pluggables have a fantastic advantage in terms of simplifying the network, and the fact that they are open when none of the transponder-based solutions are open today, they're open and they're multi-vendor. We think that that's going to really change the economics of the industry. As any open standard has, if we look at Ethernet or IP having those open interfaces has really driven great economics in the industry. And this is the first time that DWDM is really offering an open interface. And so we are very bullish on the application domain for pluggables. We have competitors out there who are trying to basically put the pluggable into a very small box and say it's only going to play here. We don't believe that. We believe it's got great application outside that box. And we think that, that's where the industry investment is going to be going in terms of improving the performance over time. And so we are, I think, very bullish on that. And most of what we had to say at OFC, I think, was around the pluggable and our belief that, that's going to have a -- it's going to shape the industry going forward. Raj, is there anything you'd like to add on that?

Yes, Bill, the one other point I would add there is coherent in the same way as being more appropriate for pluggables, you're going to see more of that in the access network as well as you get into smaller form factors as you get into 5G backhaul, it lends itself to the application once you get into 100 gigs and you can get it into a QSFP-DD, the smallest form factor. Today, I think you're going to see more applications that coherent gets into. And so to the extent you manage power and you manage the cost, I think it's going to have other applications as well. So we did have some announcement on coherent in the access network as well.

I tried to listen to one of the sessions on Shannon limits and how do you modulate that and it brought back some really bad memories from college, engineering days. So I had to shut that off pretty quickly, but it was really a great event, I think. I guess, one of the things that always comes up a fair bit in discussions is the cloud provider on the hyperscale markets, right? And maybe Bill, it's worthwhile just touching on how do you think Cisco, Cisco plus Acacia, is really positioned with the cloud providers as it relates to your portfolio? And I obviously think this is the area where you're going to see a lot of top line growth. But I'd love to understand how do you see your market share standing today? And how does that develop over there?

Yes. So I think cloud providers remain one of the significant growth areas for Cisco. We have, I think, made great progress with them recently. But frankly, it is a recent trend for Cisco. We were not really big in the major web players until fairly recently. I think we've had a very significant shift in the recent years, very recent years, and I think it remains an area for growth for us. And I think I attribute that to the fact that we were stubborn for a long time about how we want to play with the web players. And by stubborn, I mean, we would go to them and say, here's our router, how many would you like? And that wasn't how the web players wanted to engage. The web players wanted to engage in a very different way and say, look, I want to consume technology in a different way. I want to consume the portions of your technology that I'm interested in, and I will add the rest. I'll play a more significant design role disaggregating functions. And Cisco had a very tough time with that for a long time. And I think it was really in the last couple of years where Cisco made its mind up that we were going to play in a different way in order to win web. And what that meant was we had a, for instance, disaggregate our solutions and offer them hardware if they only wanted hardware. Software, if they only wanted software and even components. And that was an announcement we made in December of 2019, that we were going to develop a component business model for Cisco. And I think that was sort of a tipping point in our relationship with the web players because they saw that we were truly serious about winning their business and meeting them where they wanted to be met. Now Acacia comes to Cisco with that heritage. Acacia has always played the component model with their customers. And they play at the DSP or PIC level, true components, and they play at the embedded module level, which is more of a subsystem and obviously at the pluggable level. And really, that's how web players wanted to engage. So I think that was a real tipping point in our relationship. And we've seen the results pay off now. And they are buying things like our routers. They are buying our optical systems, fully integrated solutions, but they also want to buy just pluggables in some cases, or in some cases, ASICs. They want to buy the raw ASIC and go design their own. So we're playing with them across the spectrum, but what it took was that willingness to be able to engage with them in a different way. And I think we're seeing the fruits of that pay off now.

Perfect. I assume we appreciate the candid stats. You haven't done explaining how the history with this relationship was and how it's evolved. Maybe just step back on this a little bit a bit and you have the slide on the building material optical versus everything else, the hardware. But I want to step back and maybe you could just talk about, what is the rationale? And how does being so deeply involved in the optical space really helps Cisco versus maybe just partnering with some of the optical companies? So maybe touch on the counters in building it internally, if you may, and what advantages that encompasses for Cisco as you go forward.

So Cisco has had a pretty long history of being in optical, starting really, I think, as I mentioned, with the acquisition of Cerent in 1999, well before the time line was with Cisco. But 1999, but its view about the role of optical has shifted over time. I think if you went back to 1999, Cisco would have said we're going to become the optical leader. And what it has morphed into now is really Cisco keeps optical in the portfolio primarily for customers who want to buy IP and optical together and want to have one throat to choke. And we succeed with that. We succeeded with that, for instance, at Verizon, a very significant win in Verizon's metro business. We succeeded with that at PLDT in the Philippines, a nationwide build-out. Their long distance network, where they wanted one vendor that would deliver both IP and optical. Now we sell to 4,000 customers. So not every customer buys integrated IP and optical. But largely speaking, customers want to deal with Cisco and say, look, I just want one company to deal with on this. Now we do go out opportunistically and win optical business when we're not selling IP. In some cases, we've won web business, for instance. We're not selling a router to that web customer, but we're selling an optical solution. So I would say opportunistically, we go sell optical on a stand-alone basis, but really strategically, it's there for the combined IP and optical play.

Got it. And then maybe talk a little bit on the routed optical network solution that you have just recently launched, right? And I think it's essentially integrating Cisco's routers with Acacia's pluggable optics. Can we just talk about, what drove you to develop the solution? What kind of problems or bottlenecks does it really address that you weren't able to do before? And what are the efficiencies you're going to get out of integrating these products?

Yes. So I mentioned in the intro, and I think I used a graphic there that shows 3 layers of the traditional customer network. And if you go talk to almost any carrier, today, you will see at least 2 layers in the network. One is a DWDM layer that has very sophisticated components like ROADMs that are designed to switch wavelengths. And those -- that ROADM was developed, and I was actually part of a company that developed the first generation ROADM, but those ROADMs were developed to bypass routers because the router was the most expensive resource in the network. And so we built this very complicated, very sophisticated and very expensive DWDM layer, trying to bypass an expensive resource in the network, the router. And so if we went from L.A. to New York, for instance, you might have 10 routers between L.A. and New York, but you would send a wavelength all the way from L.A. to New York bypassing all those routers in between. And the idea was it was expensive to go through those routers because the route reports were very, very expensive. But what that drove was it drove complexity in the optical layer because now you had to have a wavelength that could actually travel from L.A. to New York. And you had to have something that could switch it on and off at L.A. in New York, so you needed some optical switching capability. And so it kind of drove up the cost in the optical layer in order to avoid cost in the IP layer. And we've seen this happen over 20 years. We've built much, much more sophistic capabilities in the optical layer. And by the way, the OTN layer as well, which I'm not even mentioning but it's kind of a third layer that was designed to do similar things bypassing the router layer. Now as the cost of that IP layer has come down, it causes us to step back and say, hey, does it make sense to rethink how we built networks. And by the way, in bypassing the routers, what we ended up with was a lot of very lightly used wavelengths. So they're lightly loaded wavelengths that go -- that are express wavelengths effectively. And so if we step back and say, well, the router port is now cheaper, I can go point-to-point through those routers, hop-to-hop through those routers using the pluggable DWDM. I can really, really pack those wavelengths now at 400 gig as opposed to having a lot of lightly loaded wavelengths. What I get now is far fewer wavelengths, but they're much more densely packed. I'm going through routers and dropping traffic of using IP and ultimately, it results in about a 40%, 45% savings in TCO. That's both CapEx and OpEx. And actually, there's a white paper that's been written by ACG that does a pretty good job of talking about that analysis. But it's that economic value that we think is really going to drive customers towards this architecture. I should say, architectural shifts like this don't take place overnight. We're not going to see all the customers on earth transition to routed optical networking overnight. What we will see is customers as they deploy a new network, we're going to be talking with those customers about don't build a network that we've been building for the last 20 years, build a network that we're going to be building for the next 20 years, and that's routed optical networking. And I expect this is a 5-plus year transition for our customers.

Got it. I guess, maybe if I just continue down the discussion around the routed optical networks and actually you folks have been a big proponent of this. Maybe just talk about are we going to continue to invest in merchant modem products that are not intended for integration into these platforms?

Yes. Actually, let me just make one other point, Amit, that I think is an important point because you introduced the concept of routed optical network with Cisco routers and Cisco pluggables. I think one of the advantages is that the DWDM pluggable is now open and interoperable with other vendors. Routers have always been open and interoperable. So this is not a Cisco-only solution. This is an architecture that could be adopted by a customer for use with third-party routers and third-party optics. We certainly hope they use all Cisco. We're certainly going to make that a compelling argument for them. But I think one significant advantage is this is truly open. No optical solution that they deploy, so including a Cisco optical solution offers the level of interoperability that these pluggables will. So I'm sorry, your follow-on question.

Yes. Really the follow-on question here, you sort of touched on this a bit, but are we going to continue to invest in merchant modem products that are not intended for integration into these router platforms?

Yes. Yes, thanks. Yes. So -- absolutely because the pluggables go into the routers and the pluggables will serve a specific set of applications. Certainly, DCI, certainly metro, even some of the long-haul applications. But we'll have customers that have other applications like subsea, long haul. And those customers are going to demand that we deliver more traditional chassis-based solutions, transponders and DCI platforms like Cisco's NCS1K. So I think there's going to be applications that drive us to say, it's in a more traditional platform. Those technologies find themselves encapsulated in a different form factor. And I think the other factor that's important is customers, in some cases, are going to say, I want to live with the operations model that I've always had. And I'm not ready to change my operations model. And so I want to think about IP and optical separately. And therefore, I want to have a distinct optical system and a distinct IP system. And some of those customers are going to continue with that architecture for some time. And so we'll continue to serve them with our existing systems business that are including advanced innovations that are still coming out of Acacia for those applications.

Got it. Perfect. We have a few questions lined up here. But maybe before I get into that, maybe I want to start a little bit of discussion on the 400 gig ZR, the product, I think, really looks set to go from testing to commercial deployments. I think later this year, you can correct me if I'm off on that. But maybe just touch on, how do you think the development of these products have gone? What do you think is the potential for 400ZR as you scale this over time?

Well, why don't I toss this one over to Raj and let him offer some views on that.

Yes. Thanks, Bill. So the 400ZR, we've talked about this before. It's well past the design. And I mean the qualification stages. So it is what we would call in the high-volume ramp mode right now, Amit. So we have engagement with all 4 of the largest hyperscalers including many users of the colors and other platforms. So from an Acacia perspective, Cisco perspective, we are very well positioned in all these hyperscalers. And in terms of -- it has gone through, I would say, in the thousands, a few thousands that we've already shipped. So it's well underway in terms of needing to -- what the deployment -- and it's being deployed already. So it is early days. It's not gone that far, but it is already being deployed. And what we are seeing is the demand has been very robust because of the 400ZR allows you to transport the 400-gig Ethernet links on a single wavelength. And to Bill's point about pluggables and the low power as well as what we see is some of the customers are building their own. They're buying their own components putting together, but they are buying the modules. So we feel -- as Acacia stand-alone, we've scaled multiple pluggable products, the CFP and the CFP2, which are obviously much bigger. So at this stage, I would say it is well into the deployment phase. We see a multiyear, these are just starting out, and we anticipate the volumes to be very healthy over the next few years.

And then Raj, maybe I'll go into some of these questions here. And maybe just to wrap up the ZR in just a bit. Where are you seeing the significant uptake? Is it service providers, cloud providers? And any sense of kind of where that's happening?

It is all of the above. The -- as Bill talked about, we are seeing tremendous pull initially from the hyperscale to web-scale providers. That's where we are seeing the early strong demand. As we said, it's the same platform. The 400-gig platform is going into a couple of different form factors into the CFP2, which is a little bigger than that, and that has also things like the [ open road ] interoperable standards. So that is being used in the traditional service provider business. And also, if you look at the web scale providers are, Bill talked about submarine, but there is also pace stuff that is evolving that is using pluggables as well. And those are using 400-gig pluggable. So you are -- it is broad-based demand, not just a single -- and it's in multiple sectors. And we see, as you get into the access and some of the large enterprises, we see that evolving there because the power is manageable. It is standardized. There's multiple interoperable standards there. So we see this more a broad-based, but initially, it is the web-scale moving into the traditional service providers.

Perfect. That's really helpful. And then Bill, maybe a few questions for you. We hear it from the audience -- maybe just elaborate on who are your primary competitors? And what are their relative market shares? Maybe it's a bit of an optical 101 TAM, how big is Cisco? Who do you compete with? Market share stuff.

Yes. I probably don't have the market share numbers in my head, but I'll give you an indication of who the major players are that we compete with. So Huawei, ZTE, for sure, in -- especially in Asia, but really almost everywhere, except North America. We see Ciena as a significant optical player. They're probably one of the top 2 or 3, Huawei being certainly in the top -- is probably the top vendor. Nokia plays pretty strongly in optical. But it's a very fragmented market. So Cisco, I think, probably has less than 10% share globally. And we do not seek to be #1 player in optical. As I said before, we really target the optical opportunities primarily for those customers who want IP plus optical and then much more opportunistically. But that's a very fragmented market. So then beyond the big players that are Ciena, Huawei, Cisco, Nokia, you would generally see guys like Infinera, ADVA, Fujitsu and then regional players like Padtec in Brazil, ECI now Ribbon. So those are some of the major players, highly fragmented market, probably 10 players that make up most of the share, and there's probably another 10 or 12 players regionally that are very, very focused on one small region like Korea or Brazil or Israel.

Perfect. That's really helpful. I guess, there's another one here on -- it's fully around co-packaged optics or I think Cisco sometimes call it in-packaged optics, but Bill, actually, this would be a good one. Maybe you step on top of what that really means. And when you see products being introduced with this co-packaged in-packaged optics, what does that really mean for Cisco? Because I think it should be a big differentiator for you over time.

Yes. Yes. And one thing I want to state at the outset, we are -- we're passionate about making sure that we have a solution here, but we're not religious about like co-packaged is the way to go. And the reason is that today, the basic model is you have a router or a switch with a bunch of ports on it, like, say, 36 ports and you plug a transceiver into those ports in order to add capacity. That's the basic model. Now as the bit rate of those ports goes up from 10 gig to 40 gig to 100 gig to 400 gig to 800 gig, the signals that have to get to that base plate have to go faster and faster. So that's one issue. And the other issue is that the signals coming off the ASIC. In ASIC, if you look at the raw ASIC, it's got a bunch of pins or little balls on the bottom of it that attached to the printed wiring board. We're not going to get many more -- we're not going to double the number of pins in order to go from 400 gig to 800 gig. We basically have to make the IO go faster and faster. And so now that I/O coming out of the ASIC have to travel across the line card, the switch or router to the face play, where you have a pluggable. And the signal integrity issues get very significant as you increase the bit rate. But more importantly, the power grows very significantly as you increase the bit rate -- double the bit rate. So every time you double a bit rate, you've got to deal with more power in the same footprint. And that becomes a cooling problem. So at some point, what we expect is that the power issues, because every time you have to send a signal across a line card, you have to basically do things to condition it as it passes through the line card, things like retimers and those things consume power. And so the power issue is going to drive us to say, we need to put the optic right next to the ASIC and not have the signals travel across the switch to a pluggable, which has got more power now. We need to put the optic right next to the ASIC to try to reduce the overall power footprint. Now when will that happen? Well, today, we've got largely speaking, 50 gig, 50-gig lines coming out of the ASIC, and we're going to move to 100 gig. We'll probably be okay at 100 gig using the conventional approach. And the next jump is going to be at 200 gig. Now the industry is scratching its head on 200 gig, even 100 gigs is very challenging. And -- but at 200 gig, the industry is really scratching its head saying, I don't know if we can use the conventional approach here, if we're really going to be hitting a dead end. Now I would say we will root for the industry to solve the problem, like go solve it guys because coming up with a new architecture around co-packaging or putting optics and silicon together is a big, big lift for the industry. It requires that we develop an ecosystem that can do that packaging and there's issues around what gets standardized and is the silicon standardized? Is the optic standardized? Is the whole package standardized? There's a lot of issues that have to be grappled with. So we would root for the industry to go solve the problem at 200 gig and just postpone that pain of introducing a new architecture called co-packaged. But we're not sure that the industry is going to be smart enough to do that. And so we are investing right now in co-packaging. And I expect that later this year, early next year, we'll start to see some early demos of that capability, maybe at 100 gig, even a 50 gig just to demonstrate the capability because it will ultimately have big operations impact for customers, too. Today a customer buys a router or a switch with 36 ports. They're all empty Day 1, and they plug in a little transceiver when they want to add capacity. What co-packaging suggests is that all those ports are going to be filled Day 1. And so now it raises questions like, well, what if one of them fails? In the old world, you went and replace the transceiver. In the new world of co-package, do you throw the whole switch away? Or do you have to have software that runs around that broken port? So there's a lot of operations issues that have to be dealt with. And I think what we're going to see is probably some customers that deploy on some small-scale, early-stage versions of co-packaging to gain some experience with operations and understand what does it mean from a maintenance perspective, from a fault locate perspective, from how I think about consuming this technology that's different from the conventional world. So I think we're investing. We're investing with our silicon, our optics. We're also open to saying it's third-party silicon in our optics or third-party optics in our silicon. All those issues have to be dealt with. And we're working very closely with customers who are very anxious in driving this. Microsoft and Facebook have created a forum where they are driving some -- driving vendors towards solutions whether it's the right long-term answer. The industry doesn't know that, but we've got actually a couple of efforts going on internally to make sure that we're covering our bets here.

Perfect. Maybe I'll just shift a little bit in terms of Acacia and the integration, how that's going. And I think there's a lot of discussion we tend to have with investors on the integration of Acacia and how Cisco has managed that internally. I think you have been very clear and continue to support -- you've been very clear in saying you want to continue to support third party that wants to use a case here. But maybe just touch on how is that integration going? And what are the benefits of combining it with Cisco if you intend to continue supporting third-party users?

Yes. Let me take a shot at that, and then I'll ask Raj to offer his perspectives. So yes, we made a decision -- well, actually, let me back up a little bit and give a little bit of history that people may not really be cognizant of. When we acquired CoreOptics, which was like 10 years ago, CoreOptics was a first-generation coherent supplier, and they were supplying to us. They were also supplying to some of my competitors. And at the time, we had no way to support a component business model. It was just so foreign to us that we told the competitors, look, we will honor your contracts however long they exist. We are going to continue to support you. We'll continue to sell to you, but we really prefer that you go find another supplier. And that's what we told them. And it was an interesting conversation. But -- and by the way, we honor those agreements for like 5 years. So we continue to support those competitors for 5 years, but it was an unnatural process for Cisco, and we really had to do everything manually. We didn't know how to take an order for a part. We didn't know how to do an RMA apart, everything had to be done in annually. When we looked at Acacia, we really grapple with the fact that we said, look, we cannot simply cut off those customers. And for very different reasons that I mentioned earlier, we were already contemplating our component model and saying, look, if we're going to win web, we have to have a component model. So we put those really 2 pieces together and said, look, we're going to embrace the component model and a component model for silicon or optics means that you serve the industry, and that may mean you're serving your competitors. You can't just sell to friends because those friends ultimately become competitors. And I don't think anybody should be too anxious about this because this industry sort of breeds strange bed fellows. We have customers who are our suppliers. We have customers who are our competitors. And that's just the world where we're in. And so when we made the decision to acquire Acacia, we did it with our eyes open that it was going to be a full component model. And we'd be selling those compounds to our competitors, and we'd have to organize ourselves in such a way that we could do that with integrity and not basically create huge skepticism on the part of our competitors Day 1. That's very natural. I mean -- and I've met with all of our key customers, as Raj has -- meets with them regularly to reassure them that they will get fair treatment when it comes to things like allocations, access to technology, access to the team. And so we've been very upfront with them about that. And we're very determined to make sure that a year from now, they look back and say, Cisco really is a great supplier. And my objective is basically that we offer them the best technology at a competitive price point, at the right point in time and force them to make a decision that says, I'm going to adopt an inferior technology rather than have a product that happen to have a Cisco label on it. That's the decision I want to force them to make they're going to have to take an inferior technology or they take a product that happens to have a Cisco label on it. And that's the choice I'd like them to make, but we want to be a good supplier to them. We've organized ourselves in such a way that we can firewall their information. So as an example, I don't see unit pricing or unit volumes for products that are going to our competitors. Raj sees that, but I don't see that. I see an aggregate, I see aggregate how we're doing for the competitors. We call them NEMs, but I don't see anything at the detail level. And so Raj is basically the firewall there. And we've organized the team in such a way that we've got a completely independent sales team. So the Acacia sales team remains in place serving those customers, and they are not part of our Cisco sales team. They're actually completely independent. They only come together at Chuck's level. So we've done some things organizationally. We've done some things structurally like we've got a completely independent ERP system to manage those customers. So there's no information that's basically being shared across ERP systems. And we've done things from just a pure integrity perspective. We've built into things like our code of conduct, like guidelines, like, hey, the Cisco sales team is trying to sell something to one of our customers that happens to be competing with one of our optical competitors also selling to one of those customers who might be using Acacia, you guys can't share information. That's ultimately an integrity issue, and that's where we build things like code of conduct and say, look, there's so much we can do structurally and from a firewall perspective and then ultimately, you have to count on people's integrity. But we do ask people to sign a code of conduct statement saying that they understand the rules of engagement there. Raj, do you want to offer some thoughts on that?

Yes. So I think you captured a lot of the key points. Just to step back, third-party, the NEMs, as we call it, network equipment manufacturers, are a significant portion of Acacia's revenue even now. So that's still a fact. All the steps that Bill talked about, which is people-level separation, system-level separation and firewalling of information is critical. I think we've shared all of this with all of the executives in our customer base. And so I want to say that we haven't lost any customer because of this. In fact, some of them are continuing to grow revenue from these customers as we move forward. And we are also getting new wins, design wins for new products that are still on the road map coming from these customers. And so that is even excluding the web scale guys. The web scale guys, we are continually engaged, but even the NEMs we are doing very well. So we see this as a win-win because I think the more we talk to these other leading vendors that we supply to, and they give us a different customer base than Cisco has. We want to develop the best-in-class products. And so -- and Bill's point is very accurate. You want to -- our goal is not to -- to make their decision first of all to say, do you want to buy the best product with a Cisco logo on it? Or you want to buy an inferior product? So I think from everything that was said during the whole merger has gone extremely well. I think there is a team that is focused on internally in Acacia, the third-party NEMs are a big priority as well. And so far, we've done very well with them.

Perfect. I did not appreciate the extent of firewalls you had in place for Acacia. So it's really, really good to hear that. I know we're way past the time we had allocated for this. So maybe I'll pause here, Carol, and turn it back to Cisco or out to you just to see if there are any closing remarks or anything we did not touch on that you want to make sure investors are cognizant about when it comes to Cisco's optical business.

I think you hit the key questions, Amit. And again, thanks very much for folks that were tuning in, and hopefully, they got some additional insights about where Cisco is headed and our commitment to this space.

Likewise, thanks for the opportunity, Amit.

Perfect. Thank you, everyone. Thanks for your time.

Bye-bye.

Thanks, everyone. As a quick reminder, as you close out of your browser. Please click through and provide us with your feedback, there'll be brief survey. Hope everyone have has a great rest of your day.