Good day, ladies and gentlemen, and welcome to the Alcatel-Lucent IP Innovation Webcast. My name is Katie, and I'll be your coordinator for today. At this time, all participants will be in a listen-only mode. We will be conducting a question-and-answer session towards the end of the conference. If at any time during the call you require assistance, please key star followed by zero, and an operator will be happy to assist you. I would like to now hand the call over to your host for today, Lindsay Newell, Vice President, Product and Solution Marketing. Please proceed.
Thank you, operator, and welcome, everyone. We're here today to announce a significant innovation in IP networking. I would like to just comment on slide two. There may be some forward-looking statements on today's webcast, so we've included the obligatory safe harbor statement to cover that part of things. You can also find supporting materials at www.alcatel-lucent.com. Supporting press release and other materials are available on the public website now. As the operator said, we will host a Q&A session at the end of today's webcast, and you can also submit questions through the webcast interface itself, as well as audio at the end of the call. With that, I'd like to introduce Basil Alwan, President of the IP Division at Alcatel-Lucent and Head of Network Strategy, to take us through the webcast today. Basil.
Thank you, Lindsay. Good morning, everyone. Thank you for spending some time with us this morning. We have a brief, but I think very informative presentation for you this morning. It's entitled "Faster, Smarter, Greener." Let me just, for a minute, back up and talk a bit about the context of our announcement this morning. Many of you who work with us, either customers or analysts, understand we've had quite a bit of success in the routing market over the past years. Since 2004, when we announced the first service router, the 7,750, or started shipping it, I should say, we've had tremendous success. In fact, we've had year-over-year market share growth consistently every year to today. Of course, that's in a market that actually has quite exceptional growth itself. We've been a benefactor of a very strong market, and we've had very good market share growth.
Here are some statistics on our latest year 2010 results. You can see we've become quite successful and ship products around the world to most of the top service providers, and the number of platforms we've now shipped is quite substantial. One of the questions, of course, that always comes up around this continued success and this secular market share gain is why? How is this happening? What is the difference? What's the change in the market that allowed this kind of growth? To answer that, I just think at a very high level, but really, I think kind of the right way to look at it is really in the business of routing, IP routing, there's been, I should say really routing, multi-protocol and IP, there's been three major waves. The first wave of routing technology was really enterprise routing.
Of course, this is where Cisco built their business initially, and multi-protocol and interconnecting sites of large corporations around the world. In the routing business, to build a new business, there has to be meaningful change. Our customers spend an awful lot of time learning these platforms and integrating the platforms into their operations. A little bit of a little bit better product is nice, but it's not typically enough to convince a mainstream pragmatist or someone who's really looking at the problem to change platforms. What has to happen in the market is a meaningful change. The next meaningful change was, of course, building the internet, which was a fairly different task. Yes, it involved internetworking, but it also was really purely focused on IP, not multi-protocol. It was about scalable, high-speed, high-stability platforms, and of course, this is where Juniper built their business.
The internet's still growing like crazy, but at the same time, in the early 2000 timeframe, we and a few others kind of took a look at the market and said, "Hey, what's really going to happen here," or what we believed was going to happen, was ultimately the move of all services onto a common IP/ MPLS platform. This was in the early days of MPLS, so it was anything but certain, and it was a pretty controversial point. One thing you could see clearly is you looked at our customers. They had a network for video, a network for voice, a network for business services, a network for consumer services, and a network for internet services. All these were separate networks. Oftentimes, there are many of these networks, and each of them sometimes had their own supporting networks, supporting metro networks and aggregation networks, access networks.
If you just look at it, the benefit of bringing this together on a common technology platform was huge. The challenge, of course, was that IP routing platforms and the technology itself was really not ready for that task. These systems were not built initially when we were building, in terms of building out the internet. They were not built initially in order to handle the kinds of SLAs, service level agreements, or more importantly, let's put it a little more succinctly, if a router decides to reconverge or needs to calculate a reconvergence, does the voice call continue to go through? Does the video session continue to stay up? These types of issues are fundamental and needed to be addressed, and that was really the foundation of what we considered service routing. The third generation of routing is really about service routing, all services on a common technology platform.
Ultimately, all services increasingly on a common network. That's not to say in a major service provider you can ultimately have just one network. That would be unrealistic, but you'll have less networks, more efficient, simpler operation, and be able to do more things with these platforms. This has been a huge change in the market, and this is really the fundamental change that has allowed us to grow and maintain leadership in this business. Our customers really are focused on how do they do two things right now. They're doing a hundred things. They're building networks as fast as they can to keep up with bandwidth growth. In a way, our customers have a superb challenge here.
Not only do they have to keep up with bandwidth growth, which is like changing the jet engines on a flying jet, but at the same time, they're trying to make their systems, as I said, more efficient, converge, bring together these various networks into a common platform. That's kind of a really complex thing. It's not purely a technology issue. In fact, it's anything but purely a technology issue. If you think about it, each network that has services on it is not just a bunch of technology, but also an entire staff. It's an OSS-BSS integration. It's what they call sometimes a network stack. It's a quite complex system. The concept of bringing these systems together or over time converging these systems is exceptional, but it's a huge task, and it's not something you do immediately.
Rather, you set a bit of a timeline for yourself in the future, and you stepwise make decisions that allow you to bring these networks together over time. That's, in fact, what's going on in our business right now. It's, in fact, happening, absolutely happening. You know, years ago, five years ago, this was a good thought, and I think a lot of people were very interested in it. Today, it's absolutely something that's going on. Stepwise, leading service providers are making choices in metro networks, in core networks, in edge networks, and they're slowly, as they can, converging these networks onto not only a common technology platform, sometimes onto a common network, which is great gain, great efficiency for them. Today, we're here talking about the service routers then. Service routers were a fundamental redesign of routing platforms, not just silicon.
We're talking today about silicon, but also in software, OS, they had to do high availability, increasing the in-service software upgrades, very complex things to do in a routing platform. Network management changed entirely because now you're doing service level management, not just port level management and aggregate flows. When I say network management changed entirely, I mean entirely. It's really a new, very complex system, important system to be delivered with these networks, including now, of course, policy management and things like that to manage specific user profiles. Of course, platforms like the 7,750, and the platforms we build out of this technology. However, while all of those things are important, and truly, we truly did have to rethink the entire routing platform, rewrite protocols, all the things we had to do to get this platform out, the foundation of it truly is silicon.
Silicon here, we've called it the DNA of IP routers. It really defines what the system can do. Silicon development, as you know, is, as you may know, a multi-year project. It's not something you do every day. The design cycles are getting longer, more expensive, more complex. Ultimately, if you do it right, silicon defines many key things about your system, certainly platform capacity and performance, which is critical in our market, especially with the advent of video and the amount of bandwidth that's being put onto these networks. It's truly stunning. Also, services scale, scale and evolution, which is increasingly important because in a converged network, these systems have to both run exceptionally fast, but they have to be agile, agile enough to basically be at once a business service platform, maybe a mobile service platform, consumer service platform.
From a customer who wants to combine certain functions, they need to be capable of doing that. That means quite a bit of intelligence as packets are flying through at incredibly high rates. Of course, power efficiency, really important, more important than ever. Five, seven years ago, this was a thought. Now it's actually one of the key design considerations that we look at when we're building these very, very complex systems. Today, we're announcing FP3. FP3 stands for FlexPath 3, and I'm going to talk to you a little bit here for a few slides about why it's faster, smarter, and greener. It's really an engineering achievement. FP3 is our new 400 Gb routing silicon. In a minute, I'll kind of share with you some details of what that means.
FP3 ultimately gives us the ability to build super density platforms with incredible intelligence that gives us the foundation to continue to build new systems for the years ahead. FP3 is a pretty impressive product. Let me tell you why. The first slide I'm showing you here is showing our evolution of FlexPath. We've done three generations now. FlexPath encompasses two chips. It's a packet processor and a traffic manager. You can see here the latest platform is 288 processor cores in a single chip. It's a quite complex 40 nm design. You can think of network processors in the same way that you might think about a GPU or a DSP. They're general purpose, but they have very specific attributes.
When I say the general purpose, when they wake up, they don't know exactly what they are, and you give them an instruction set, in this case, many instruction sets, and they start to compute. You can basically make this thing, as I said, business services, consumer services. You can change things on the fly. It gives you a lot of agility in the forwarding plane. At the same time, the parallelism and the design approach give you tremendous speed. FP3 is giving us now 400 Gb per second single flow performance, which is really absolutely remarkable. Just three, four years ago, we announced FP2, our 100 Gb network silicon. Today, we're not only announcing FP3, but we're actually demonstrating it in the labs.
It's been in our labs for multiple months, about six months actually, and it's been verified, and we have it now on 7,750 cards and running true 400 Gb traffic through this platform. This gives us a lot of flexibility in terms of continuing to grow the performance of the 7,750, and to build out for the future. In terms of how this compares to what's out there in the market today, this graph shows our network processor evolution, along with the leading other network processors that are designed both in-house as well as by third parties. The reality is that we've maintained a fairly strong lead. You'd expect that for us to have market share gains quarter in and quarter out, there's differentiation in our platform. We've been very focused on this market.
The importance of this announcement is that FP3 gives us a very strong lead into the next few years, I think critical years, because service routing is becoming so very important to our customers. FP3 is a true 400 Gb processor today, the leading edge processor running at 100 Gb, and that's, of course, a remarkable achievement. As I noted, one of the real important things here is agility. FP3 not only runs at 400 Gb, because one of the things you can do to run things faster is obviously to reduce the number of functions that the processor is capable of performing. How many instructions can you run per packet? How many things can you do? In the case of FP3, like we've done with FP2 and FP, our original FlexPath chipset, it actually supports more services, more feature depth than ever.
FP3 is not just an increase in speed. It's an increase architecturally. It's an increase in performance on both dimensions, speed and agility, or speed and smarts. This gives us a platform to continue to build incredibly intelligent routers that can address convergence, but also can be at the very leading edge of performance. Power is more important than ever. One of the great aspects of integration is, oh, we get some power benefit, but power reduction in these chipsets is no longer just achieved by the stepwise integration or Moore's Law curves. We also are doing a lot of active work now on how do we manage power on these chips and in these systems so we can deliver more and more aggressive power specs. This is the only graph that I, with my team, look at and say I'm very happy that it's going down.
Normally, I'm asking all the graphs to go up and to the right. This one I want to see go down. We're talking here about 2 W/ Gb, which is very aggressive and very competitive in terms of power consumption. As I said, this is becoming a very important term for our customers. Power density in central offices is critical. It is important to pay close attention to this and not just write down specs on a sheet, but actually actively manage power. We're doing a lot of that. In summary, faster, smarter, and greener, and truly is. The world's fastest network processor by a very large margin. This is not a specsmanship kind of game. This is a network processor. One of the nice things about network processors is there's really only one way to spec them.
Can you or can you not carry the performance and do the feature set? This is what we're demonstrating in the labs today. It's the world's fastest network processor. It's much smarter, gives us a tremendous amount of growth on that dimension, not only for IPv4, but also to scale IPv6 where it needs to go with the growth of endpoints in the network and, of course, services. Of course, as I just mentioned, much, much lower power, which is really critical nowadays. What are we going to do with this incredible technology? It's an amazing breakthrough. Our engineers have done a fantastic job working on this platform. We have a lot of options. We can use FP3 in a number of ways. We can actually build a 400 Gb interface today if we have 400 Gb optics. 400 right now we're at 100 Gb.
Alcatel-Lucent has leadership in coherent 100 Gb optical transport networks. We're going to use a lot of the same technology to get to 400 Gb. I think 400 Gb is going to be a workhorse technology step for us and for the industry because there's a good alignment between optics and IP. We have the ability now with this chipset in a particular layout, as shown on this slide, to do a 400 Gb interface. However, right now, 100 Gb high density is probably more important because that's where people are in terms of optical interfaces. We have an ability to use this chip in two modes, in what we call a full duplex or a simplex mode. In a full duplex mode, it does 200 Gb full duplex.
It's one set of tables and traffic all 400 Gb, 200 Gb in and 200 Gb out, running through a single set of chips. We can also lay down two of these. You still have one set of tables because you have ingress and egress split. This is a bit technical, I apologize. It gives us the ability to build, as it shows on this slide, a 400 Gb card or a 400 fiber card over time. It gives us a lot of flexibility. It's the way the platform was designed. It's tremendous from an engineering point of view to give it that flexibility over time. It's very similar to what we did with FP2. FP2 is a 100 Gb network processor. The first cards we built were 50 Gb full duplex. We were the first to market with 100 Gb full duplex cards, which we're now shipping.
Today, we're also announcing our 200 Gb port SR12 line cards and a 20 port 10 Gb E.U . line card for the SR12, which is incredible density. I'll get to a density comparison here shortly. These are absolutely state-of-the-art platforms. These are not just platforms for moving bits. Sometimes in the router business, you have different cards, some cards that will just move bits and do basic lookups, and some cards that do queuing and all of the other feature set that is required for convergence. In these platforms, because of the silicon innovation, all of the ports are capable of certainly moving bits at incredibly high speeds, but also all of the ports are capable of high service features, high QoS, VPNs, video, all of the various features that customers need. Looking ahead, as I mentioned, we have now a key ingredient.
I think we have an opportunity here over the next years to continue to grow in the market and continue to put out leading platforms that kind of lead the density and performance curve, as well as the intelligence curve. I thought it would be interesting to share that the original 7,750, which was a 200 Gb full duplex router, we shipped out. Those are our state-of-the-art line cards through 2007, roughly. Between 2007 and now, what we're demonstrating, we've taken that entire chassis, 200 Gb per second, and we've integrated into a single card. Now this chassis with this particular card, fully populated, is a two terabit platform. It's a two terabit platform in a third of a rack, which is important because rack density is really part of the question here. The 7,750, which is a pretty compact product, it's good from that point of view.
Here you see a comparison of densities, leading service interface densities. The reason I say service interface densities, this is about service routing. It's about convergence. Therefore, these are the densities of ports that can actually do high service. By high service, we're not setting a bar super high for a nice comparison. We're talking about more than port level queuing, the ability to actually handle multiple flows and handle them in an intelligent way. This is a comparison between the Cisco, Juniper platforms that are their edge routing, service routing platforms and the 7,750. As you can see, the comparison is quite favorable. It's particularly favorable if you look at rack density. This is listed as chassis density, but obviously, if you look at 10 Gb, for instance, with the FP3, we're demonstrating today 200 Gb in a third of a rack.
That's a 50% benefit to the nearest competitor in density and with deeper intelligence, deeper services, and a more mature and full feature set platform. It's a really great step for us in terms of density and puts us in a very good spot in the market for service routers. One other comment I want to make, which is very, I'm going to make a very important point on this particular slide. Not only have we, I think, solidified our lead in service routing, which, as I said, is going to continue to be a growing segment, a more and more important part of the overall service provider routing landscape. Not only have we solidified our position in that market, but also, for the first time, really, service routing silicon has eclipsed or matched core routing silicon in performance. That's a really important step.
Service routing silicon has always been a little bit behind core routing silicon. It sometimes got a little bit ahead, sometimes was a little bit behind. On the whole, core routing silicon, which does much less in terms of features, is doing basic route lookups. Nowadays, with the idea of core transport, it might even do only MPLS lookups. Core routing silicon is much simpler. Therefore, in the past, it's run a bit faster. With this generation of technology, 40 nm, you can have all the feature functionality of a service router, and you can accomplish the performance of the absolute fastest core routers. In fact, if you actually compare densities of the SR12 with the densities of most of the leading core routers, you'll find it compares favorably.
What we've been able to do with this generation is not only build a chipset that gives us a foundation for building the industry's leading service routers, the most important routing platform in the evolution of our customers' networks, but also gives us, on one platform, the ability to build super dense core platforms. We're not announcing a core platform today, but the point is the technology here is quite profound. The time to technology change. You can see traditionally what the industry has done was build different silicon for services edge, sometimes edge routing, core routing, and now core transport. The reality is that you can accomplish these different performance points with a single chipset by populating different memories around the chipset and doing different service steps.
This is a very profound step, the step that I think is going to make things simpler for us, for our customers, and ultimately deliver some really good value into the end market. Of course, this technology for Alcatel-Lucent is really important across the network. Our strategy of high leverage networks is about at once building the most cost-effective networks, cost-per-bit effective networks, by pushing the limits of technology in optical, in wireless, in wireline access, and in, of course, routing and switching. As you can see here on this slide, this is a very important part of the evolution of that network because 100+ Gb us is important not only at the service edge and at the core, but it's also becoming increasingly important in aggregation and backhaul because these metro networks are carrying all of the traffic of the service provider. They're also converging.
As they converge, the amount of traffic that needs to be managed, and managed carefully, so it needs a bit of attention, service interplay, you have to make sure you have the feature set to manage multiple services in a metro network. As you can see, this is a very, very important part of the overall high leverage network strategy. The strategy is about cost-effective delivery of bits, but at the same time, increasing intelligence so our customers can leverage and monetize their networks as best as possible, given regulatory constraints and opportunities they have in the market. In delivering that kind of combination of feature functionality, the service routing platform is really at the core, and it's a very important innovation for us to deliver that entire framework. In summary, we're announcing a pretty profound shift today.
Not only is it a next speed grade, but it's a really big step forward, making 400 Gb real and giving a huge performance delta from current platforms and silicon, including our own, in the market. It's a big step in terms of performance, but also a profound step because network processors are getting powerful enough to perform all the tasks across core, edge, and metro. That's a really very fundamental technology change that's brought to you again by Moore's Law and the ability to integrate more and more functions into a single device. It's not only faster, it's smarter, and of course, importantly, it's also greener, reducing power quite substantially.
We're pretty excited about the opportunities it gives us for the upcoming years to build lots of interesting platforms and to evolve our systems to help our customers face their immense challenges because they truly have some pretty immense challenges ahead, trying to keep up with customer demands and build a business model that works in the long run. We're very pleased to have that opportunity to participate in this value chain, and the FP3 is a very exciting evolution of this technology. With that, I think I'll open it up for some Q&A. Thank you for your attention, and I look forward to having a few questions to answer here.
Thank you, ladies and gentlemen. At this time, if you'd like to ask a question, please key star followed by one on your touch-tone telephone. If that question has been answered, please press star followed by two to withdraw your question. Please press star one to begin, and please hold while we compile a list of questions. Your first question will come from the line of Mark Sue from RBC Capital Markets. Please proceed.
Thank you. Hi, Basil. Hi, Lindsay. Maybe if we can just start on the timing of the core products based on the new chipsets. Now that you have the chips and also the line cards, what might else do you need to assemble to deliver a product? Can you give us a thought on what kind of capacity and throughput we should expect for some of the new core routers?
Yeah, I can't comment, obviously, on our future developments specifically. I would just say that there are a number of additional features needed at the system level in order to satisfy what our customers need in the core. There is obviously work to be done there, but we don't have at this point an announcement around core routing timing-wise and density-wise.
Okay.
You know, look, at the end, as I say, the foundation really is silicon. It really, when you get right down to it, the long pole in the tent, the thing that really drives what density you have and when you can deliver to the market is absolutely in the silicon technology, which is the reason I brought that up.
Okay, understood. If you look at the core routing market, that's been a duopoly for as long as we can remember. It's not as feature heavy as edge routing, where you have been able to gain market share. What do you think you might be able to do to kind of wedge yourself and drive incremental market share in this market that's been somewhat static? Maybe some tactics and maybe some of the ways you might be able to increase your footprint over time?
I think the important story is, first, number one, service routing is becoming service routing, what you might call edge routing, but really is no longer just edge routing. It's really any routing position in the network where you have to have some feature functionality. As I mentioned, metro networks are getting much smarter as they're pushing the IP edge out. What's happening is this feature functionality is being required in more parts of the network. The first thing I want to say is that service routing, as a percentage of the routing business, is, in my view, going to grow. The most important part of this announcement is we're set to continue to lead in that space. In terms of the pure core, what sometimes people call the super core, which is really getting less intelligent in some ways, it's still a massive engineering project and requires massive bandwidth.
The comment I would make about that is there are places where we're very involved with customers, and it would be, I think, appreciated by them. There's an opportunity for us to make things a bit simpler because they know our systems. They know our... If they can, for our customers, there's a balance between having more than one vendor and how simple their operations are and the commonality of systems. I think by virtue of being in the network and being a good size IP player, we have a seat at the table, so to speak. I think we'll get our opportunity in two ways to participate in the core. Number one, as a service routing business grows, it naturally takes a bigger portion of the overall network. Number two, we have an opportunity also to build very high capacity true core platforms.
Let's see how that plays out, but I feel pretty confident that we have a good opportunity there.
Got it. That's helpful. Thank you and good luck, gentlemen.
Thanks.
Thank you. Your next question comes from the line of Nikhil Perodolpo from UBS. Please proceed.
Can you guys hear me?
Yep.
I did maybe just an extension, a couple of questions. You might have mentioned this earlier, but I missed it. Is this 400 Gb capability available immediately, or when would we see it on the 7,750?
We demonstrated this literally this week in the labs. We had a bunch of people through showing it. We have it up and running, and it's in very good shape. It won't ship commercially until 2012. We're not saying specifically when, but given where we are, you can imagine that it's in pretty good shape. It's 2012 commercial availability of these cards.
Okay. Maybe just a follow-up question on the core market. You know, another market that seems to be developing in the core, and you kind of hinted to it in terms of the core perhaps being less intelligent, are these new MPLS core switches that some of your competitors are rolling out? I might have missed it, but I haven't heard Alcatel-Lucent's view on that subset. Is that a market that you view as meaningful and as you think about entering the core? Something you want to consider and pursue as well, or are your actions, are your plans more tied to the traditional core router segment? Thank you.
Yeah, good question. First of all, the core is further being separated into, in some cases, depending on how you look at it, into kind of the edge of the core, which is still a full routing platform. Maybe what you might call the terabit super core, whatever you want to call it, which is basically turning into potentially an MPLS-only affair. We see customers who want to do this for sure. What they're really after, by the way, it's not so much that they, it's not that they don't want that core platform to do route lookups. It's that they want lower cost and they want higher performance. The question, I think, is one of cost and performance. It's not so much of whether an MPLS-only thing is better. In fact, quite the opposite.
It's always better to give the customer flexibility to do routing, switching of LSPs, routing of, and services, frankly. It's always better from their point of view because they have flexibility in the platform and they can decide when and where to route and when and where to label switch. It's a cost-performance issue. One of the comments I made, and the reason I said this was a bit of a profound moment, is that from our point of view, silicon that allows for very good feature depth can also be used to deliver the very highest performance platforms. The question becomes one of cost. Our view is yes, there is always an opportunity to give our customers lower cost, higher performance platforms.
Maybe it would be possible that they can have also the flexibility to use a platform in a number of different ways, if you understand what I'm saying. We see the opportunity. We believe in it certainly because there's a good opportunity to help our customers reduce their costs. We think with this technology breakthrough, we can address that as well.
Okay, thanks, Basil. Thank you.
You're welcome.
Your next question comes from the line of Simon Leopold from Morgan Keegan. Please proceed.
Thank you. A couple of things I wanted to ask about. One was if you could describe the opportunity to upgrade the existing footprint of routers that you have deployed in the network. Are there any obstacles in terms of that footprint to accepting the new cards when they come available, and how we can access that opportunity?
As we have always done, the cards we build are backwards compatible, so they'll work with the existing cards. Customers don't have to do forklift upgrades. They certainly can if they need the density. Normally, what customers have done is deployed new I/O cards, sometimes side by side with the previous generation I/O cards. That path will be available. If a customer needs absolute top performance, they may decide to replace all the line cards and put in a new higher density platform. The chassis itself has been extensively remodeled over time. Cooling, power, all the various pieces of the platform are keeping, we're keeping it very leading edge so that customers who buy the 7,750 value proposition have a very long life in front of them. That's really the key here for our customers.
When they deploy a system like this, they're not deploying it for three years, they're going to forklift. In fact, quite the opposite. With service routing in particular, or with all this convergence going on, the amount of integration with their systems is quite high. It's absolutely imperative that you give them a smooth path to increase density, performance, and features. You can imagine we put a tremendous amount of focus on making sure our customers get a chance to do that. I don't believe this is one of those things where you say, okay, now everybody's going to swap out their current products. Rather, I think what they're going to do is they're going to start, as they have done in the past, adopting the new technology to stay ahead of the performance curve.
What you'll see is a growth of these platforms, sorry, of this particular technology platform into the current installed base.
Great. In terms of looking at another adjacent opportunity, does the new product help you penetrate more of a data center market than you have in the past? I know the focus has traditionally been universal edge of service provider. I'm just wondering about looking into this new market opportunity, if there's an application there.
There is. In fact, both on the optical products at Alcatel-Lucent and the IP products, there's a very good opportunity because what you need in the data centers is, in addition to data center switching, which we have a separate set of products for, and we partner with Hewlett- Packard on, another really important requirement is high-performance data center connectivity to the access networks and to other peering points in the network. These data centers need tremendous bandwidth in and out of them. There you need a very high-performance edge, and you need some feature functionality. You certainly need VPNs. You need a variety of different features. You need, in fact, a combination of speed and intelligence. There's a great opportunity for data center interconnect. The 7,750 Service Router fits perfectly, along again with our optical platforms, which are already being used.
Both of them are already being used as data centers grow. Absolutely, there is. In addition to that, there's also still incredible growth in the access side of this business, access networks, because video is really crushing right now the network. That also is another driver for just the absolute growth of these networks.
Great, thank you very much.
Yeah.
Your next question comes from the line of Vincent Maulay from Oddo Securities. Please proceed.
Yes, hello. I would like to know what the percentage of the core routers market you are able to address, so namely the part of the service router.
That's a really interesting question. I think service routers, as they get higher, more and more performant, and they get to these densities, a good service router can play at what I called before kind of the edge of the core. I don't know that I could give you a percentage because the reality is some of our customers really demand multi-chassis and other types of features which are not in the 7,750 family at the moment. If you're asking whether the set what part of the core market 7,750 itself can address, given this very, very high density, I'm not sure that I could hazard a guess. I think the more important point is that the technology platform we now have gives us, for the next three to five years, an opportunity to address those issues and build platforms into the core.
If I had to kind of think about our business and how maybe you should think about our business, I think I would think of it in this way. Number one, because it's the biggest market, the fastest growing market, maybe the most important market in routing, it's going to solidify and maybe even accelerate our opportunity at the service edge as it grows. Number two, clearly, we have this opportunity, which we'll share more information about as we go forward to address the super core.
Thanks.
Thank you.
Your next question comes from the line of [Adami Opor]. Please proceed.
Hello.
Hello, yes.
Yes. Good afternoon, gentlemen. Could you give us a breakdown of the market service edge versus core, or maybe what's the best way to break down the market in your view?
Lindsay, do you want to make a comment on that?
Certainly, yeah. If you look at most of the industry analyst reports for 2010, the split would be about 2/3, 1/3. Edge would represent about 2/3 of the total service provider router market, and core would be about 1/3. Again, precise definitions of edge and core vary, but as a general guideline, about 2/3, 1/3 is the generally accepted split between edge and core.
Okay. This is helpful. Thank you. Just maybe a follow-up. What's your market share, estimated market share in Q1? I think maybe it was slightly lower than in Q4 last year.
I think we'd caution against tracking market share on a 90-day basis. Things can change significantly or appear to change significantly. If you look at the macro trend, we're in the over 20% range and growing. If you look at rolling four quarters, for example, growing in terms of most of the major analyst firms' latest reporting. I wouldn't necessarily look at Q1 versus Q4 market share or revenues, but if you look at rolling four quarters, we're definitely gaining share and we're in the more than 20% range.
Okay, thank you.
Your next question comes from the line of George Notter from Jefferies. Please proceed.
Hi, thanks very much, guys. I wanted to ask a question about the different facets of the edge routing market. Historically, if I'm right, the company has been successful in areas like residential broadband aggregation, business Ethernet aggregation. How about some of these new, more emerging opportunities for you guys? Evolved packet core, wireless backhaul, even caching content on the edge. Are those new markets that this new product gives you a better entry into? What kind of progress are you making there? Thanks.
Yeah, no, I mean, it's a great question, George. Thanks. The reality is that this is all connected, you know, not literally. In that, if you look at our, for instance, mobile backhaul, as you mentioned, I'll treat each of the ones you commented on. Mobile backhaul is not just about the small device, what we call the 7,705 that sits out at the cell site, though that's a very important platform and it runs a consistent operating system for us, one network management system. It's very nice. It's also about the aggregation of that into what's sometimes called the MSN, which is basically a 7,750. The density there, performance, is absolutely critical. You can't have a market-leading wireless backhaul solution if you don't have a dense MSN that's aggregating all of these platforms at these thousands and thousands of cell towers.
Further, increasingly, the trend is in the mobile backhaul space to combine the mobile backhaul network, although these have been largely separate organizations in our customers. The cooperation has not always been there. Increasingly, the economic imperative is to combine business backhaul, maybe consumer backhaul, and wireless backhaul and aggregation. When that happens, you absolutely need a platform of this performance, and you need it also to be smart. You need to be able to manage the services. It's absolutely critical to have a leading service router platform to win wireless backhaul. I'd say wireless backhaul also encompasses, in our view, one architecture across any link to connect the cell tower, whether that's a microwave link, whether it's a fiber link, a copper link, it doesn't matter. That's kind of how we're approaching wireless backhaul, and we've had very, very good success in that market, as you may know.
In the packet core, we consider as the market moves to all IP in wireless, which is just now happening at the LTE step. As that happens, it's interesting to look at the packet core and to look at what's traditionally been known as the [BRAS], now the BNG, and see the amazing parallels between those platforms. They're doing a lot of the same things. For us, there's some real benefit to us from a development point of view, but also to our customers to have some commonality there. Now, I'm not saying people are going to combine a BNG or BRAS with a packet core. Packet core is a very separate thing. It's a very sensitive position in the network. I would say we're early in that one because we're relatively new to the market.
Our thesis on packet core has been that a router-based packet core is going to be needed over time because of the sheer performance required out of the evolution of the number of UEs and the performance of the UEs. The amount of traffic that you need to pass through these platforms is going up exponentially, as you know. At the same time, there's a lot of high touch. It's a very good application of this technology, of the silicon technology, the base technology of the 7,750. From our point of view, we have a very solid new modern foundation for a packet core. Like all the other markets we've entered, it takes a number of years to build a feature set to where we have all the features. In a way, you can think of a new entrant in any of these markets.
The initial platform you build, you might be able to address some smaller percentage of the overall RFPs in the world because this number of features you have to develop is in direct proportion to the number of RFPs you can actually qualify for. The reality is that what we expect to happen here is we have a new modern high-performance platform. Over time, as we add features, and this is exactly what has happened in the BNG, we get more and more opportunities and at-bats. We're very positive about the evolution of packet core, but it's a long game. It's a big changing market. It's an opportunity that we need to stay plugged into. I would further say on the packet core, there's some real benefit to being connected with a company that has a broad set of assets in building the entire wireless service.
In terms of just understanding how to build, I'm not suggesting that you sell the entire thing as one. That happens sometimes, but the reality is there's a lot of benefit even if you're not selling it all on, even if you're selling individual piece parts into understanding the evolution of those platforms. Packet core is an exciting new area. The last thing you mentioned is CDN, and I think CDN is absolutely critical for our customers to support video on demand and the evolution to basically on-demand video for all, unicast video for all. The reality there is that CDN is intertwined with routing platforms because at some point, and we've kind of pointed in this direction, some caching can be done in those platforms. It is absolutely tied, and we're pretty involved, as you know, in the evolution of CDNs for our customers. Long answer. Sorry about that.
Thank you. Your next question comes from the line of Kai Korschelt from Deutsche Bank. Please proceed.
Yeah, thanks for taking my question. My first question was really on the timing. I think in the release you said 2012. Is it more likely the first half or second half? My second one would be on the installed base. I think you mentioned 100,000. I'm just wondering whether that referred to the actual number of 7,750, or if that is a higher level number, please. If you could maybe share with us roughly what percentage of that installed base you think is potentially upgradable. Thank you.
I apologize. Lindsay, do you want to grab that one?
Yeah, sure. I can answer those for you, Kai. If we start with the last one first, the vast majority, I mean, any of our customers for the 7,750 have the backwards compatibility to mix the new line cards based on the FP3 technology with their existing line cards. All of our installed base has that opportunity to upgrade. On the number of routers that we've shipped, yes, the number we stated at the beginning was 150,000. That's routers that run our Service Router OS, so the one operating system we have all the way from Ethernet access through metro, through edge, through core. That includes all of those routers. Within that number is a number of 7,750. Remind me, what was the first one of the three, Kai?
That was just on the timing of the ramp, i.e., first half, second half next year.
Yeah, on the timing, round about the middle of 2012 is when the line cards for the existing 7,750 Service Router will be available with the new FP3.
Great. Thank you.
Thanks.
Thank you. Your next question comes from Ehud Gelblum from Morgan Stanley . Please proceed.
Hi, thank you very much. Appreciate it. A couple of things. As you look at the speed of these edge routers and the service edge routers that you're coming out with, as you said, they're faster than core. Do you see core routing in general as a separate distinction from edge routing sort of going away and edge routing kind of creeping into the core where you'll be having a lot more services in the core over time and therefore service provider networks start moving away from having pure core routers, or do you think that there's still a room for them and the distinction will be more on cost between the two types? Just wondering how you kind of see the evolution of the two separate types of routing markets.
Yeah, I think first of all, most customers, I think, are thinking to the core as a simpler, I'm going to call it the super core here because I think there's a distinction to be made. Yes, I do believe service routers do "creep" towards the core because the core, if you further break down the core, there's the edge of the core and the core. I hate to say it that way, but that's kind of the reality of how these networks are being designed. The answer is the edge of the core, it absolutely makes perfect sense because you have higher performance and you have a complete flexibility of service functionality. That's a real benefit.
In the past, that's a trade-off you made for performance is you basically said, "I'll take the slightly less capable platform for performance." It was an absolute requirement because in the core, it is all about performance. It's about moving bits. If you come to a guy with a proposition and say, "Hey, here's a platform that does a lot of fancy stuff, but it's a little bit less performant," they don't have the luxury of that. I would say that makes sense. They're running pretty hard to keep up with the demands of the network growth. The answer is, though, if you get to a point where you can have it all, certainly it gives you immense flexibility. The edge of the core, I think, is absolutely an opportunity.
I'd imagine the cost difference. There's a cost difference, though, when you add the flexibility on the service.
That's kind of the key here. The key question is, which I think you're hitting on exactly, is that silicon has started to make that less of an issue, quite a bit less of an issue. The real question becomes how much do you populate these systems if you want to get to the technicalities of it with tables and memories and things like that. Cost comes into... There's two forms. Core routers are, number one, the biggest forwarding complex you can make is by definition going to be the best because what happens is every time you put down a network processor, let's say you have a 50 Gb network processor. If you want to do a 200 Gb line card, you're going to put down four of those.
Not only are you going to put four of them down, you're going to put down four copies of the route table, four copies of the MAC table, four copies of the ACLs, four copies of everything. You can imagine how the cost and power adds up. There's a huge benefit to integration. The second question you have to ask, if you have leading edge integration in terms of, I should say, performance, in this case, 400 Gb, how will you choose to build the systems around the network processor? We have some flexibility there. We have flexibility in terms of leading towards incredibly high services, flexibility towards leaning in the direction of the more basic things. When I talk about the edge of the core, I think there's an absolute opportunity for us to grow.
As far as the terabit super core, what we call kind of what you're talking about is the evolution of kind of the MPLS only. I think there also silicon is going to give us potentially an opportunity. If you're a customer, and I've had several meetings with Johanna, obviously a lot of meetings with customers about this, what they're after is performance and cost, of course. Actually, probably in that order.
If I could ask a separate question on your share, aside from the fourth quarter when you had your Verizon build and you shot up to 28% share and then kind of came back down to 24%, you'd been at 20% share pretty consistently for the last two, two and a half years, back until 2008 when you started gaining a lot of share. Why do you think that your share kind of leveled out at 20%? Do you think that the move with this recent build has taken you to a new level that can continue up into the 30% share plus level?
Yeah, I mean, Lindsay, I don't know if you want to comment on the first one.
Yeah, I'll be happy to take that one. I think that's a bit of a mischaracterization of our share evolution. We gained share in 2008 and 2009 because we continued to grow our revenue in the macroeconomic downturn when CapEx was being squeezed. We actually continued to grow revenue and grew share. In 2010, we took about 4 points or 5 points of market share. I think the dip you're referring to in Q1 was more of a sequential anomaly because of the phenomenal Q4 results that we had in IP. If you look at the overall trending over the last 2 years or 3 years, our share gains have been meaningful and significant.
I didn't mean to say that they weren't. On an annual basis, they absolutely were. On a quarterly basis, it looks as though you had a beautiful 20% share in late 2008, early 2009, and then it kind of sat there until the fourth quarter. It might just be timing. I don't know.
Sure. Yeah, I mean, again, I counseled earlier, don't look at market share in 90-day increments. Look at them in 12-month increments, which is probably the more meaningful way to view them. Certainly, with this technology, as Basil Owen said, we've now got the DNA that provides routing performance without any sacrifice in service capability. We have a best-of-breed platform technology now to address the full range of the service provider around the marketplace. I think it's a meaningful lead that's sustainable over an extended period of time as well.
Thank you very much.
Thanks.
Operator, I think we'll just take one more call as we're past the top of the hour now.
Okay, great. Your final question comes from the line of Divya Mathur from RBS. Please proceed.
Hello there. Can you hear me?
Yep, I can hear you.
Hey, I just had a quick question. I was looking at the number of ports you have on the line cards, and the maximum combination, the maximum speed you can get is 6 Gb x 40 Gb. That's 240 Gb. My understanding is that you have some of your competitors already have 200 Gb line cards. That's not, in terms of the increase in the speeds, that's essentially a 20% increase. My first point is, am I right in thinking, am I missing something here? Secondly, in terms of the cost savings, I was just wondering, how do the cost savings compare to the line cards that some of your competitors have?
Yeah. First of all, on density, in terms of absolute density, if you're really looking at density, you have to also factor in rack density because our system's a third of a rack, a smaller board. It takes less space in the rack. If you look at, I think you might be referring to the recent announcement, which was about 192 Gb max on using 10 Gb interfaces. We're about 50% higher based on three in a rack versus two in a rack. I would also point out that it's probably more instructive, honestly, to look at what's shipping today and then listen to all the vendors, including us, because we're showing this technology. It's not something where we're just saying we're going to have it. We're demonstrating, people are verifying it, all that kind of stuff.
Ultimately, that's about a future announcement, and we'll see that play out in 2012. We feel very confident in maintaining a significant density lead on that order, 50% +. As far as how that accrues to cost, it's absolutely a direct relationship. As I said a minute ago, the higher the increment of bandwidth you have, the less replication of logic memories that you have in your system. All of these systems that are built have many, many network processor complexes across all the line cards. The question is, if you have a choice between having 40 systems, 40 network processor complexes versus 20, there's a very big difference in terms of just raw cost, just because you're replicating everything over and over and over again. It's not exactly perfect in terms of that, but it's roughly a linear relationship in terms of cost.
Okay. Just another quick follow-up on that. I was just wondering, what's the hindrance to having more ports? Can you not choose cards with four ports, maybe 400-Gb ports, and then that'll use the whole capacity of the processor?
Yeah. If you look to exactly what we did with FP2, we lived as an exactly similar thing with 100 Gb network processor. The first cards we brought out, which absolutely led the industry, were 50 Gb. We used the processor in a full duplex mode, just basically running traffic in both directions. We did that. A few years later, we were the first guys to get 100 Gb out and remain the only guys that have a service interface, service routing interface at 100 Gb shipping in the market. The reason that we did that in time is that things change. It's not just the network processor. It's also optics. Optics take an awful lot of board space on these systems.
If you go back and look at the size of the optical interfaces, for instance, for 10 Gb, four years, five years ago, they took up a good chunk of the card. Similarly, for 100 Gb right now, if you look at our 100 Gb interface, just take a look at it. You can see on the front panel, they're taking a massive amount of not only front panel space, but also board space. We have now the core technology, obviously, to build all of these variants. As the optical technology, memory technology, various things come along, we'll instantiate more density. Leading density is about network processor, but it's also about everything that goes around it. It's incredibly difficult to build a product like this. That is the core silicon technology we're announcing today.
Once you have it, you now have the opportunity to instantiate it in all kinds of ways. Today, putting a good amount of effort into a very, very high density card, what we were able to produce is this two-port 100 Gb interface. The chips are literally running both directions of that interface, which is fantastic from a cost point of view. It gives us a tremendous advantage in cost. Chips can be reconfigured when we're ready and when we have the right component densities and the other areas of the design, they can do 400 Gb, no problem.
Perfect. Congratulations. Thank you.
Thank you.
At this time, we have no further questions. I'd like to hand the call back over to management for closing remarks.
Thank you, operator. I'd like to thank everyone for joining us today. Hopefully, it's been an informative session. I'd like to remind you that in February 2011, we announced light radio innovations that will help revolutionize wireless networks. Today, we've announced the FP3 silicon innovations, which will help revolutionize IP networks, not just in terms of performance, but also massive scale service delivery from the metro to the edge to the core across residential, business, and mobile customers. Thank you again for your time and look forward to speaking with you again soon.
Ladies and gentlemen, thank you very much for your participation in today's conversation.