Infinera Corporation (INFN)
September 15, 2011 1:00 pm ET
Randy Nicklas - Chief Technology Officer
Matthew Finnie - Chief Technology Officer
Thomas Fallon - Chief Executive Officer, President and Director
Unknown Executive -
Stuart Elby -
David F. Welch - Co-Founder, Chief Marketing & Strategy Officer, Executive Vice President and Director
Unknown Analyst -
Ladies and gentlemen, please welcome, CEO, Tom Fallon.
Good morning. I'd like to welcome all of you to Infinera. I appreciate both the live audience and the audience online. It's a very exciting day today for us as we kind of unveil where Infinera has been working to take the networks. Before I get into my presentation, I need to discuss with you safe harbor, which basically says the presentation I'm about to give has forward-looking commentary that reflects what we think will happen in the future versus what will happen in the future and therefore, contains inherent risk.
In the early 1900s, Ernest Shackleton decided to take on what he considered to be the last great Antarctic adventure. He crossed the continent, sea to sea, over the polar journey. To find his crew, he placed a unconventional ad in the London Times, "Men wanted for hazardous journey, low pay, bitter cold, endless hours of darkness. Safe return? Doubtful. Honor and recognition if successful." He fielded 5,000 applicants, and he and his crew set sail on the Endurance. Partly through the journey, the Endurance got trapped in ice, and captain and crew were forced to abandon ship and faced an arduous and dangerous journey back home, where they all made it safely. Today, Shackleton is considered a hero. The journey in the pursuit of photonic integration is a bit like Shackleton's adventure. First written about at AT&T Bell Labs in 1969, the promise of the PIC was unfulfilled and almost unpursued for several decades. By 2000, conventionalism was that PICs would not be commercially viable. In 2001, Infinera went in search for its own great adventurers, offering low pay and the opportunity for honor and recognition when successful. When we introduced the DTN in 2004, the question of PIC commercial viability was put to bed. And today, I am delighted to unveil to you our next leap forward in networking from Infinera, our next generation of photonic integrated circuits and the next milestone in Infinera's journey to transform the telecom infrastructure.
The DTN-X, based on our third-generation, 500 Gig PIC. It is a next-generation multi-terabit switching platform that offers really 3 products in one. It is a DWDM transmission system, offering 500 Gig super-channels based upon 100 Gig coherent channels. It's an OTN switch that scales from 5-terabit non-blocking at ODU 0 today, to 100-terabit non-blocking in the future. It is a platform that can be upgraded to MPLS in the future, allowing our customers to deploy today and have a migration path to packet in the future.
Today we're going to talk about a lot about the DTN-X, but we're also going to talk about the fundamental technologies that enable the DTN-X. Because we believe unless you design this type of integrated platform from the electron and the photon up, you cannot do it without compromise. And while the industry is starting to recognize the value or even the necessity of having integration of DWDM and switching, only Infinera with its core technologies provides the ability to create a platform that achieves the scalability, simplicity and efficiency required by today's networks.
In the next 45 minutes, with the help of some of our industry experts, Stu and Randy, I'm going to walk through a little bit of the history of the DTN and some of the success it's created. I'm going to highlight some of the network changes that have happened since the DTN was deployed in 2004. I'm going to close with why we view the DTN-X as being uniquely positioned to satisfy these new requirements.
In 2004, we introduced the DTN with 2 fundamental innovations. First, we introduced the first 100 Gig PIC, offering brand-new economics. And we also integrated for the first time integrated switching, allowing customers to extract maximum value from the network resources. Our customers loved it. It was 100 Gig slots. And today, we have over 85 customers in 50 countries who have deployed over 8,000 chassis. They loved it because they took the complexity and the art of analog engineering out of the field, and they replaced it with a digital optical experience. That's important because they were now able to achieve scalability, simplicity and efficiency. They could roll out services faster than anyone else and grow market share. And they could do that with less CapEx and less OpEx, allowing them to maximize their profitability.
Based on the success of the DTN, we have continued to invest in that platform. And today I'm also delighted to announce our next release of the DTN. It's more scalable, it's simpler and it's more efficient. First, we're announcing our 40 Gig coherent line card that enables our customers to take fiber capacity up to 6.4 terabits, twice that of most of our competitors. It's simple. Our customers can deploy a 10 Gig line card or a 40 Gig line card in the same chassis, on the same fiber, at the same time. And we're also introducing 2 new service modules: A 40 GigE service module and 100 GigE module, allowing our customers to more efficiently interconnect with routers and carry those services over either 10 Gig waves or 40 Gig waves, whatever makes the most economic sense.
And finally, it's in-service upgradable, allowing our customers to maximize their investment and minimize customer disruption. While customers continue to tell us, "that's great, but we need more help getting ready for the multi-terabit age." There's 3 fundamental drivers that are causing remarkable network demand. Obviously, the advent of high-quality Video Everywhere. Second, the dominance of broadband mobility. And finally, the emergence of cloud computing. And these 3 are not independent variables, but they're actually catalysts that work together to fuel more and more Internet demand. As most of you probably already know, Netflix today, which is really a cloud service, takes up about 20% to 30% of broadband demand during peak hours. And it's not unusual for a mobile device to have a video client like Netflix launch from it and rich multimedia get distributed in the cloud with features like Apple's iCloud. This creates 2 fundamental tensions in the market. First, on user experience. As a user, I'm fully aware that the application's not running on my local device. But my expectation as a user is the experience is such that it looks like it's running on my local device. That creates huge bandwidth requirements and low latency. The second tension that's brought to the market is from Wall Street. Well their expectation on the carriers and service providers is they're going to maintain or grow their profitability, even in the need of investing more and more capital.
This is not a new debate. I've been in this industry for a long time and the discussion has been ongoing, and I think there are real issues there. However, I also believe just like in the past, innovation is going to create step function opportunities to alter network cost and create the appropriate financial rewards for those who innovate and deploy that innovation, and I'll walk you through a little bit of our view of the history.
In the 80s, SONET and SDH were deployed to -- because they fundamentally did a good economic job of carrying voice and data services. In the early 90s, DWDM was introduced, and this created a step function in carrier costs, as carriers can now take more full advantage of the fiber plan that they had deployed. Later in that decade, 10 Gig DWDN was introduced. And while that had the effect of increasing overall fiber capacity, it really did not create the economic step function that had been experienced before. That is until Infinera introduced the DTN based upon the 100 Gig PIC and 10 Gig radical economics. When we introduced this platform, it was such an economic disruption that it actually accelerated the adoption of 10 Gig in the market. One analyst, maybe in this room, called it the "Infinera effect".
The industry responded mostly through brute force. They chased low cost regions of labor, and they also introduced 40 and 100 Gig. Now 40 and 100 Gig has certainly increased the capacity of fibers. But to date, our challenge has not created the economic disruption of that as needed.
Well today, we are introducing the DTN-X based upon our 500 Gig PIC, and I'm going to challenge that we view this as the next economic disruption in the network. Our 500 Gig PIC not only offers 100 Gig coherent channels, but it offers 500 Gig super-channels. It's pretty easy to understand how very large pools of resource are more economic than smaller discrete pools of resource. What's less obvious is that growing necessity to integrate switching to tame the power of these large pools of bandwidth. I'm going to walk you through that pretty quickly.
If you look at the carrier model, one of the challenges they have is, how do you match service requirement with transmission requirement carrying those services? Historically, 10 Gig has played such a dominant role in the industry that it's masked some of that over time. Lots of services were 10 Gig or close to 10 Gig, and lots of line speed was 10 Gig. So the service line speed mismatch was pretty small. As pipes get bigger and bigger going to 100 Gig, 500 Gig, terabit, that mismatch becomes pretty pointed. In this example, you see a carrier providing a number of 10 Gig services across 100 Gig waves. It's very, very inefficient with 100 Gig waves being used at 10% fill, 40%, 30% fill. That's not a rational economic outlook. So what can you do? Why, you can put a device of the middle there and say, "Great, I'm going to go and mux up those channels and I'm going to get much more utilization out of my resource." In this case, 80% fill. That's probably a pretty good model. And in this very simple example, I think it illustrates why grooming or switching is absolutely necessary in the network. So the choice becomes, how do you want to do that? Well, there's really 3 fundamental choices. First, you can put in a muxponder, a muxponder is a DOM point-to-point device. It can carry and mux things up, but it can't respond dynamically to changing network needs and it's very manual. So it doesn't scale well at all. You can address that by attaching an external switch like a CoreDirector, and that certainly does help the network respond to changing network needs. The challenge now is you still have 2 platforms, and you have those expensive fiber interconnect of maintaining them. And you also have the CapEx associated with 2 platforms, and now you also have 2 logical elements in the network to manage. You don't have a collapsed network. The way to deal with that is integrate the DWDM in the switch. You replace all that fiber interconnect with a common backplane, eliminating that cost and that ongoing expense. You also create one virtual layer in the network to manage. We did a study looking at the various cost of these. No surprise, integrated DWDM switching was by far the best economic solution. And I think that's validated by the number of our competitors who are now introducing this type of integrated platform often for the first time. So if we agree that integrated switching in DWDM is the right approach the question is, what's the right approach to do that? What's the right architectural approach to optimize that capability? And I think Infinera has a pretty good vantage point to have an opinion as we introduced this architecture to the industry 7 years ago, when we first introduced the DTN.
We start with a concept of a clean sheet design. And we hired the best engineering leaders we could find around 3 core technologies. We did not start with a DWDM system and glue on a switch. And we certainly didn't start with a switch and loosely couple somebody else's optics. We started from the electron and the photon up, because that's the only way you can make an integrated platform with no compromise. And we did it around 3 core building blocks of technology. First, PICs plus coherent. A PIC is a marvelous piece of engineering that applies Moore's Law to photonics for the first time. Moore's Law in its simplest meaning, if you integrate and make things smaller, all things get better simultaneously: space, power, cost. We use Moore's Law and photonic integrated circuit to take things from light waves and break them down into electrical circuits. Once it's in the electrical domain, we can hand that off to our coherent architecture or through digital signal processing we can optimize that optical signal. And we hand it back off to a companion PIC, which takes those electric circuits and can very efficiently turns them back into light at the same or a different wavelength. That's how efficient DWDM transmission occurs.
The other thing that happens in the PIC is it breaks open those big light waves and exposes the services at the electronic level. In the electrical level is where you need to do the adding or dropping grooming functionality. And it's this level that we hand it off to our next pillar of technology, systems ASICs. We hired a world-class system ASIC team which has built and delivered in the DTN-X, a 5-terabit non-blocking switch at ODU granularity. This cannot be bought on the merchant market. And these 2 pillars are the cornerstones of building efficient networks, but we didn't stop there. We've added GMPLS software. GMPLS is a protocol we borrowed from the router world, which really allows the network to become network aware. Because we put things in the digital domain everywhere very efficiently, our nodes can report on the resources they're using and know what is available, and they become aware of what the resources are available around them. So basically, GMPLS automates terabit scale networks. These are the 3 cornerstones of technology that we use to build an uncompromised integrated platform.
The DTN-X. It is the scalable, simple, efficient multi-PIC terabit packet optical network device. It's scalable with 500 Gig super-channels built upon 100 Gig coherent channels. It's 5-terabit switching at ODU 0 today, scaling all the way up to 100 terabits non-blocking at the ODU 0 level in the future. It's simple. It takes GMPLS automation and applies it to multi-terabit scale networks. And it allows converged DWDM and OTN today, allowing our customers to avoid CapEx and lower ongoing operating expense with a hardware upgradeable path to MPLS in the future. And it can do all of this with 50% less power and 33% less space.
I want to talk little bit now about what we mean about scalable. I'm going to start talking about it in the optical domain. With Moore's Law, we are able to take over 600 optical functions and over 200 fiber interconnects and combine them all into 2 die the size of my fingernail. That is the power of our third-generation PIC. That's what enables our DTN-X to offer 500 Gig super-channels on a single card. Unleashing this amount of DWDM capability with lower power, space and cost is the foundation of our competitive advantage. But we didn't stop there. We also designed the DTN-X with the future in mind. And each slot of the DTN-X has up to a gig of -- a terabit of capacity. When we introduce our 1 terabit PIC, our DTN-X will be able to take and scale our platform from 5 terabits today to 10 terabits in the future. And if you remember, we published the test results of our terabit PIC in a March letter at OFC. So that explains how we unleash large pipes. But to get the full value of those large pipes, you have to get to switching. With our clean sheet approach, we have built a platform that allows our customers to deploy a configuration with full DWDM capacity or full switching capacity or any mix in between with no compromise.
I'd like to walk you through a few models that compare our capability in the DTN-X to those of 3 competitors who have announced products. We've given them the benefit of the doubt and have used their spec sheet for performance. We also assume that their switch is non-blocking. Because these systems are configurable, we looked at 3 different configurations. The first configuration is 0 DWDM, all tributary or client assigned switching. The first thing to note, as we stated, now the DTN-X has 5 terabits of capacity. All the competition has less, but that's okay. Let's go and change the configuration and look at 50% DWDM and 50% client side. Two things of note. The DTN-X capacity remains the same at 5 terabit. No compromise. Each of our competitors suffers capacity degradation. Let's jump to a 100% DWDM and 0%, 3 of our client side. Two things to note again. Our platform suffers no capacity degradation, no compromise. And once again, each of our competitors' platforms loses capacity. And if you poke into that a little bit, it becomes a simple understanding. As our competitive platforms remove low-powered gray optics needed to support switching and replace them with high-powered DWDM optics to support transmission, their systems can't scale to handle the power and size that is needed. It's because they don't use PICs. And when you think about it at a different level, what this really is, is a DWDM tax on an integrated DWDM system.
In the spirit of the ongoing political debates, I'm an advocate of no new taxes. So we've talked about how we scale the system from 5 terabit to 10 terabit with the introduction of the one terabit PIC, but we didn't stop there. Our system ASIC team designed a systems ASICs that allows scalability to a 100-terabit multi-bay non-blocking infrastructure. This is once again not silicon you can buy on the merchant market. Now one of the things that I have enjoyed over the last several quarters or year is that during our quarterly conference call, analysts will invariably ask me what's the status of our new platform. And George, you in particular I think have been horribly consistent in asking me the status of our ASICs. Well today, I'm delighted for those of you in the audience that after lunch, we're going to take you through our executive briefing center where we're going to show you a DTN-X live carrying traffic today. And for those of you online, I've got Lee Hunt [ph] over in our executive briefing center, who is going to give a short online demonstration of what you'll see. Lee, could you take it away?
Unknown Analyst -
Thanks, Tom. I'm really excited to show the DTN-X and the new features on the DTN in action. Let's take a look.
So what I have here, I have a 100 GigE test set, which is physically cabled to 100 Gigabit Ethernet client port on the DTN-X, that it cross-connected through the switch fabric to the line module, which houses the 500 Gig PIC. The 500 Gig PIC then takes that signal and puts it in a 500 Gig super-channel, which is being transmitted across a 3,000-kilometer ultra long-haul link which terminates on this DTN-X on the far right. From there, it comes out of the 100 Gigabit Ethernet client. It's physically connected to the new 100 Gig modules on the DTN. It then traverses the same 3,000-kilometer link only this time, we put it over 2.5 40 Gig waves. It comes back. It's terminated on this DTN right here, physically comes out and is connected to the test set. And you can see from the test set, we are running traffic error-free. We can also take a look at the constellation diagram. We see a nice clean separation between the dots, which indicates good coherent perception. So we've seen the DTN-X, the new capabilities of the DTN transmitting a 100 Gigabit Ethernet client over 500 Gig super-channels and over 40 Gig waves. Thanks. And back to you, Tom.
Thanks, Lee [ph]. Hopefully, you're as pumped up about that as I am. I think it's a very, very exciting move forward.
So we've talked about scalability. I now want to talk a little bit about simplicity and what we mean by it. The DTN-X brings the easiest to use network to the market. We've taken what our customers loved about the DTN and the ease-of-use, and we applied it to multi-terabit scale. We do this by taking advantage of PICs and digital everywhere and marrying that with the automation and efficiency of GMPLS. So why does it matter at the end of the day? Well, it matters because it matters to our customers. They can deploy networks in days. Better, they can deploy service in minutes. In fact, you take 2 tent teams -- you apply them to the network, you point and click provision. Boom. You are creating revenue-generating services before you pay Infinera for those line cards. And with mesh protection, because of the intelligence of the network, we can provide millisecond recovery. Days, minutes, milliseconds. That's what Infinera means by "time is a weapon".
Now you've heard what my impression is. I think it's more important to hear what a customer thinks about our ease-of-use. Randy Nicklas from XO. Randy, could you please join me onstage? Thanks, Randy.
Hi Tom, how are you?
Randy, you've been a great customer for a long time. I appreciate both your customer -- you being a customer, a great customer and being here today. If you wouldn't mind telling the audience your experience. Well, first of all a little bit about XO. And second of all, your experience with the DTN. Any thoughts you might have on the DTN?
Sure, I'd be happy to do so. Thanks, Tom. So again, I'm Randy Nicklas, and I'm the Chief Technology Officer at XO Communications. XO and I have been working together with Infinera for over 6 years now, and I'm going to share some of our experiences in the next couple of slides and in this brief presentation. But I'll be here for the whole program. I'll have a more extended presentation later this morning. And then like I said, I'll be around the whole day. I hope to speak to as many of you as I can, ask all the questions that you like. Most of them will probably be answerable, okay?
So again, we've had a long relationship, the 2 companies, XO Communications and Infinera. So let me give you a little more detail on that.
First, a little bit about XO. I think most of you have heard of XO Communications. We're a US-based service provider that provides IP and data solutions to mid-market enterprise and carrier customers. So we don't have any residential customers. We just sell to businesses whether those are carriers themselves. That's our wholesale group, and we do a lot of good business there, as well as to enterprises in the so-called -- what we would call the mid-market and some larger ones as well from the Fortune 1000.
In 2010, we had revenue of about $1.5 billion. We number about 90,000 customers. Some of those are quite small, but we've also got some great marquees customers as well. And then on the carrier side, I would say both domestically and internationally, we've made an impression in terms of selling bandwidth services, wavelength services across the national fiber-optic network that you see there on the map. Okay.
So again just to summarize, we sell IP and data-based services. And Infinera has played a large part in that. I said our relationship has been more than 6 years long. Back in the first half of 2006, we built on that topology that you see on the graphic there, a national Infinera build. Prior to that, we had operated somewhat I tend to call classical long-haul DWDM systems. We still operate those today. So we are versed in both the traditional way of providing intercity wavelength services on the traditional type of approach, technical approach, as well as the Infinera approach. And in fact, it's better than that since we did a national build in a very short period of time, I might add, in 2006. We have done 2 subsequent overbuilds with subsequent generations all based on the PIC 100 on a national scale. So what started out as 40 channel, a channel being a 10 Gig wavelength, went to an 80 channel and now 160 channels. And now here we are with this big machine that you see in front of you. And we're very excited about the prospects of introducing this device into our network, okay. It's doing all the right things. The only thing I would gently admonish Tom and his team, we could have used it sooner, okay. But it's a difficult machine. We understand that, at least at a high level. Because we don't understand all the black magic that goes into making these things. We're service providers. Our job is fundamentally pretty simple. We've got to provide bandwidth pipes from A-Z at a good cost for our customers, and we've got to do it rapidly, okay. And I think we've been successful in doing that.
So I think I have one more slide. So why Infinera? A lot of the points that Dave and Tom have touched upon, I can reinforce, okay. This truly is a simple to operate system. And that's important because prior to 2006, certainly on a national scale, we didn't operate long-haul DWDM systems. We don't have an XO research or an XO labs or anything like that. We're a relatively small, lean company, okay. So I like to make the analogy with the airlines. Most airlines probably really don't understand computational fluid dynamics nor do they need to. They just have to fly their airplane safety and deliver good service to their customers and let Boeing or Airbus or whomever build those things. So that's the same kind of relationship that we enjoy with Infinera. They suppress all of this incredible technology for us and give us easy-to-operate systems. And I'll reinforce that again by pointing out that we do operate the so-called classical systems. And from a network planning perspective and from a service delivery perspective, they are harder to work with. They're not impossibly hard, obviously, but they are harder. So they add a bit of friction in that sense, okay.
So Infinera has helped us grow revenue because we weren't in the inner-city wavelength business prior to 2006, okay. Take market share, this is added to our top line, okay. Reduce operating cost. It really is very simple to run our network, the Infinera part of our network. And I'd be embarrassed to share with you, maybe I can be talked into it, how many people we have in the planning or in the operations or in the sustaining engineering teams. It's quite modest, and yet we have a substantial network, and we have a great set of customers, okay, that are very demanding of us, by the way.
So I said earlier that we want bigger machines. We absolutely do. The PIC 100, the 2 generations of the PIC 100 and the DLM or the DTN-based systems that we have deployed have been very good to us. But we foresee and we've already started to plan based on what we understand of the DTN-X today our next national network based on the system. We can foresee at least in some markets or some fiber junctions 3, 4, 5-way junctions, you have tremendous amounts of bandwidth coming in across the various fiber payers into a junction. In my utopia, the ability to cost-effectively groom and switch that traffic, okay, in a complex of machines that would look something like this is a very powerful idea. Of course, the jury is still out. We got to see what the economics will be like. We got to see that actually get the vertical scaling with a PIC 1000 I'll call it, as well as the horizontal scaling with a multi-bay machine that's just starting with what you see here before. We're quite excited by that. We see no slackening off in the bandwidth demand, okay. So we need to continue to scale. We need to continue to implement services rapidly and those services need to be reliable for our customers. And of course, our customers are very demanding. They want lower unit costs. Now there's a floor to all of that. Every service provider will tell you, maybe that's kind of our fervent hope, I do think that that's the case. But certainly, Infinera has helped us get to where we are today, and we think the DTN-X is going to help us take our game to the next level.
So thanks for your attention. I look forward to interacting with all of you individually.
Thanks, Randy. It warms my heart to hear that we actually are helping customers be successful in their markets, and I'm going to make sure that every one of our engineers watch this videocast to hear that they want it sooner. So you and I are on the exact same page.
In addition to making easier-to-use networks, Infinera is also focusing on simplifying networks through convergence. If you look at networks today, in a simple way, there's really kind of 3 layers. There's a router layer, a switching layer and a transmission layer. One of the things that DTN customers enjoy today and DTN-X customers will enjoy in the future is the convergence of OTN and transmission. They can do that and avoid CapEx. They can do that and avoid ongoing operating expenses. And they can do that through just network simplification, which makes their networks more efficient and more powerful. The second thing that we see happening in the industry is really being driven by large service providers. Specifically, pushing the router guys to split out MPLS from IP. IP is a very rich services layer, but it's very, very expensive and it's prone to complexity. In the core of the network, MPLS is used to transmit or carry services in the core, not create them or provide them. So you need a really much more simplified level of MPLS than you do the richness and cost and burden of IP. We see a second phase that's going to occur then which as I split out the MPLS functionality converging that further into a converged MPLS-OTN-DWDM platform. The DTN-X was designed to allow the convergence of OTN and DWDM today and an easy migration path to MPLS in the future, where our system can perform all 3 transmission functions of transmission, OTN and MPLS, no compromise in any mix the customers want. One of the people in the industry, a leader in the industry who has joined us today, has been talking about this architecture for a period of time. And I'm delighted that Stu will be -- from Verizon is here. And please, Stu, join me on stage?
Now Stu if you wouldn't mind, I would love for you to share with both my team and the audience what are you looking for? What does Verizon need from a overall convergence perspective and a step function in cost, power and space?
Great. Thank you. All right. I won't describe Verizon. I think many of you are familiar with Verizon. Some of you probably call our service center and complain at times, and we'll try to do better. But I think it's been said, and it's very important to focus on the bottom of this graphic, which I call the carrier's dilemma. And it's already shown up in a couple of the charts today, but bandwidth consumption is going up and I'll touch more on that in detail in a longer talk later on this morning. But more importantly for us as a service provider, the revenue per bit is going down. It's going down gently, but it's going down. And obviously, we have to focus on serving our customers' needs, giving them good service, but we also have to deliver shareholder value and keep our profitability up. So we need to continually -- it's not a single-step function, but continually drive our cost structure down, so that our cost structure is doing better than that revenue per bit. And it was mentioned before, but the problem is our network is not following Moore's Law, and it needs to because certainly demand is at least following Moore's Law. There's too many boxes, too many interconnects between those boxes. That was described, that's part of it, that's part of the CapEx problem. But it's also part of the power consumption problem, right? And that's where integrated circuits come in and PICs come in. The need to drive down the cost per bit. So we talked about CapEx cost per bit, and I think Dave mentioned that. But it's also another metric that wasn't mentioned which is really picojoules per bit. How much energy is consumed per bit? Because when you look at the total cost of ownership now at Verizon's network, the after-purchase OpEx costs actually outweigh the initial purchase cost, and much of that is power consumption. So we need to drive towards integration in the future, and that really is the next big step function.
So what do we mean by transport? And I want to lay this in the vein of what a service provider like Verizon thinks of. Well, we generate revenue by providing something that customers see and feel. So a simple example, here some kids may be in New York accessing, I don't know, a website out on the West Coast here in Palo Alto, and there's a service there. There's revenue being generated by that service, which is running over an IP network. And anything that's not part of this picture is really just cost, right? It's not directly generating the revenue. It's adding to the cost structure of that service, and so we consider it transport. So there's a little build here. Transport includes the fiber and the DWDM. It includes the OTN switching. It includes services that can write on OTN like TDM. There's still customers out there that buy T1s and DS3s, they don't go away, SONET services and Ethernet and MPLS services. All of this to us is transport. Today, we do that in 4 or 5 or 6 different boxes or traditionally have done that in 4 or 5 and 6 different boxes in the central office, all lined up with lots of jumper cables between them, lots of electro optics between them driving cost and power. So when we talk about something called packet optical transport platform, this is a made-up name not from marketing. It's the -- what we call it as engineers in Verizon technology. It's the concept of combining all of those functions into a single platform so they are not doing jumper cables or interconnects, and they're getting the integration of a backplane to do those interconnected services. So that's the driver. That's the first step, and that's something that Verizon's been deploying at its local networks, and we are looking to move into our long-haul networks. But to look at the overall cost structure, you have to optimize across not only those packet OTP platforms, but the IP network it connects. And I think that was the message of the last slide that Tom put up actually, is that we have to simplify the IP network and have the right relationship between the IP network that's delivering the services and the underlying transport, which means things like next-generation MPLSTP for instance moving into packet optical transport so that we can simplify how routers interconnect to these transport boxes. So it's an overall optimization of cost at the IP layer and down to the transport layer. That's been a big focus of my team at Verizon over the last half decade. We're focusing very hard on our ultra long-haul global network and how we might find that next step function to move into that space.
Thank you very much. I am delighted and not surprised at the direction that Verizon says they need to go is the direction we're following, so thank you for your ongoing coaching. I've talked about scalable. We talked about simple. And I'd like to talk a little bit about efficient. To look at efficiency, we actually think it's important to look at real networks. So we took a Pan North American network, and we worked with a couple of university professors. One from George Washington University, one from the University of Arizona. And we asked them to model real traffic patterns on a real network design. And they did this across multiple architectures. But for today, I want to only share the data that is really comparing the integrated switching DWDM platform. And when we looked at this comparison, it really reinforced what we talked about when a clean sheet design with no compromise brings the best economics because it comes down a lot to economics.
The first thing you see is that the total number of modules are field replaceable units to satisfy the exact same network requirements is distinctly different for DTN-X versus what we consider our closest competitor. They take almost 10,000 line cards, and we use about 3,000 line cards. Part of that is because we are offering 500 Gig super-channels on a single card. They're limited to offering 100 Gig on a card. Along with the fewer number of cards, there's a drastic reduction in the number of fiber patches that are required. That's not surprising. But when you think about it, each one of these fiber patches has to be manually installed, manually tracked and manually maintained. This is a very significant ongoing operational cost. And it's also a nontrivial part of network quality. The less fiber patches you have, the more robust your network tends to be. So less cards, less fiber patches. Once again, no great surprise significantly fewer number of chassis to hold these cards, in this case going a 67% decrease. Now to be fair, our DTN-X, as you can see, is a full bay and our competition is only a half bay. So to normalize that, we looked at number of chassis. And even in base -- even in the bay case, we reduced the overall number of bays by 33%. So no compromise, clean sheet design. We end up providing the same fundamental transport capability with drastically reduced CapEx and drastically reduced ongoing expense. And as Stu mentioned it's not just about space, but power has become a significant industry issue not only the cost of power, but the availability in power and the environmental impacts of power. Because our system is PIC-based, our system uses 50% less power than any of our competing solutions. And it's because once again, it's the economics of Moore's Law applied to integrating lots of devices into fewer devices.
We had one more customer that wanted to be here today. Matt Finnie, CTO of Interoute. Matt could not make the trip unfortunately. But he did send us a video to use, and I'd like to share that real quickly.
The PIC-based DWDM solution for Interoute gives us the operational flexibility and simplicity we require at the scale that we need. Interoute aggregates all the major operators in Europe and many of the global operators in their need for the Internet growth and growth of their networks. And to do that, we need a platform that is stable, is efficient and can scale as our customers' needs grow. I think for Interoute this new development is really building on the, if you like, the tradition of Infinera, if Infinera can have a tradition. It's building on the same very sort of structured base, where it's thinking about the manufacturing of the platform, which has given us that simplicity, which has given us that operational efficiency. And for us it's very simple. It's the next level of scale. We absolutely need to move from 100 Gig to that 500 Gig PIC now. What I like about the new DTN-X is it's the next evolution of something we've become very comfortable with, and something that we rely on to drive down our cost curves. So we've been looking for the next phase, the next step function and the new DTN-X is that next shift down in terms of how we drive down the cost of our DWDM.
So Matt, if you're watching online, thank you very much. It brings me great personal satisfaction that we have so many customers and people from the industry who are reflective that Infinera works hard to make the customer successful.
So in closing, the DTN-X is a next-generation multi-terabit packet optical transport platform -- transmission platform. We started with a clean sheet design, we hired the best engineers on the planet, and we had them invest in 3 core differentiating technologies: PICs plus coherent, systems ASICs and GMPLS software. At the end of the day, it delivered a scalable, simple and efficient platform to the industry. It's scalable, providing 500 Gig super-channels based upon 100 Gig coherent capacity. It provides a 5-terabit non-blocking switch at ODU 0 today, scaling to 100 terabits in the future. It uses GMPLS technology to automate multi-terabit scale networks, and it allows a convergence of DWDM and OTN today with a path forward to further network simplification and CapEx reduction with MPLS in the future. And it does all of this with 50% less power and 33% less space. Most importantly, it carries forward the Infinera tradition of working very hard to provide tools that allow our customers to be successful in their markets and hopefully, maximize their profitability.
I'd like to thank the people in the audience, both live and online. I'd certainly like to thank Stu and Randy and Matt for joining us today. I think it's important that you have the opportunity to hear from other people with Infinera about I think the value we're creating. We have a few minutes left, so what I like to do is open this up to any questions you might have. We're going to start because there's limited time from online people, we're going to start with online. And I think Dave and Stu and Randy, if you could join me, that'd be great.
We've been actually getting clear questions throughout the presentation directly to all of you guys. We'll start off with Tom. Tom, the 500 Gig PIC is impressive. How much of a lead do you believe your current generation PIC provides and how long do you think you can sustain that lead?
I assume they mean 500 Gig PIC in comparison to somebody else doing PIC technology, so I'm going to answer it that way. We're the only people today doing large-scale photonic integrated circuits, so it's difficult to say how much of a lead we have because people have announced they're going to pursue that path and to date, they have not been successful. We're actually on our third generation of photonic integrated circuit, so we have I think a lot of learning that occurs. We've been designing this capability for about 10 years, shipping it for 7 years, and we've had 0 PIC failures in the field. And that comes from really intense understanding about the technology and the process. I think that my assertion would be -- you know what, conservatively, I would say at least a 2- to 4-year lead if somebody was going to become serious about it, but it takes a lot of money and a lot of expertise to become serious about it. But I also think, while it's an important question, it misses some of the importance because it's not just about PICs, it's about how PICs marry up with the other core pillars of technology we have, PICs with coherent. We're leaders in PICs and coherent and then pairing that with integrated switching. It's the combination of, I think, all of those things that really allows Infinera to have a very substantive leave.
Okay, great. I've got another question here on, what exactly is a 500 Gig super-channel?
David F. Welch
So a 500 Gig super-channel is a unit that can be deployed with 500 Gigabits at a time. It's made up of 5 100-Gigabit coherent signals channels from that, and it's a -- it's all integrated onto a common PIC with a common set of electronics to drive that.
Okay, great. The next question is, and I'm going to direct this one back to Tom. When do you expect to start shipping the DTN-X as well as 40 Gig on the DTN?
Okay. That's another frequent question. We've -- our stated plan for the DTN-X is that we'll have lab trials available in Q1, and our plan is to have volume shipments in Q2. So we're not altering what we have previously said. We're obviously making great progress. We obviously reserve the right to ship to customers who might be need it earlier if in fact, it's ready, and we're going to ship it as soon as we can. But we've -- it's very important to us that we maintain our reputation as being an extraordinarily high quality company, and we're going to introduce this technology as soon as possible. For the 40 Gig, we have also stated that we had started taking orders. We actually have a 40 Gig in backlog now. We've had some orders, and we anticipate shipping and recognizing revenue for some of that in Q4.
Great. Next question is actually directed towards Stu. With the current transmission technology at about 100 Gig, do we really need 500 Gig of capacity?
Yes, that's easy. Yes, we do. We started building 100 Gig in our commercial network back at the end of '09, and we've already laid another 100 Gig on top of that route. All right. That's 200 Gig right now between Paris and Frankfurt. Building out the U.S. now 100 Gig. So yes, we need it. We've already done a field trial, I call it a lab experiment that extended across Texas, that we reported back at OFC in March. It was using components, not integrated circuits just to see how it could really handle super channels and could we take that concept into the existing fiber plan, and we ran 100 Gig, 400 Gig and terabit and it does work, so we're pretty enthusiastic, and we need it now. It's ready.
That's great. Next question is directed towards Randy. Can someone like XO really use 100 terabits of non-blocking switching? We've heard that 2 to 3 terabits of switching is plenty for most networks.
I would say those lower set of figures you just gave we are exceeding today. Will we really need 100 terabits? First of all, that won't be available for a couple of years I would conjecture. And if growth demands continue, I would say, yes. We could very well need that. I'll use our own IP network as an example in the United States. So already, we'd need 100 Gig channels between router pairs in that network today. Verizon's already there on their network. There are a finite number of such large networks to be sure, but I'll just keep on going back to what's driving the demand. It's not XO that's driving the demand, by the way. It's we're trying to respond to it and keep up with it. So yes, I want very much for Infinera and frankly for others to continue to strive to build larger scalable systems with a better unit cost basis. Will we ever need them? I certainly hope so. I hope all of that work won't go unrewarded. So that's how I would answer the question.
Okay. That's great. Next question I think is targeted towards Dave here. How much value do you think your customers will really place on the integrated switching aspects of the DTN-X?
David F. Welch
I think the integrated switching makes your networks more efficient. We've come out of an era where the service speeds and the wavelength speeds are a little closely -- more closely simulated. As you move forward, that's not going to hold, right? There's going to be a disparate number of -- types of services that going into these 100 Gig or 500 Gig super-channel structures. Switching is a must. Otherwise, the efficiency of the network degrades quickly, and the economics of the network also degrade quickly. Integrated switching is critical going forward.
Okay, great. Got another question here, I'll give it to Dave. Why do you believe it's so critical to have such a large non-blocking switch router?
David F. Welch
Again, if you can envision a node in your network where I've got multi-directions coming and multi-terabits coming in, in those different directions, I've got to be able to switch that capacity freely, I need to be able to switch to that at an ODU type of granularity so I can maximize the bandwidth of that and be able to supply the services to that. Being able to drive the capacity up, where I have again these multi-directions and multi-fibers in each direction coming in is important to be able to manage that traffic.
Okay, great. And I've got one last question here, and I'm going to direct it to Tom. Do you believe your competitors will be able to compete effectively without PICs?
Yes. I think -- the audience said no. I appreciate that quite frankly very much for those of you online. I think PICs provide a couple of fundamental things that are important to the network. First of all, they apply Moore's Law, and Moore's Law allows the cost curve of optics to keep up with the cost curve of electronics and that's a nontrivial capability, particularly as the cost and speeds get higher. Second of all, power and space are becoming almost as important as cost. Without PICs, you cannot provide the power density required to continue to scale the networks. You heard it from Stu. I mean, it's a major driver of your business decisions, and there is no way to really eliminate the power requirement. Certainly as silicon gets finer and finer granularity, you can do that. But the optics needs to keep up or accelerate beyond that, and the optics and the electronics are inseparable from power consumption. You have to use the photonic integrated circuits to keep that. I also think one of the things that we talked about, it's more than handing an electrical signal to switching. It's the magic of breaking open those optical pipes very efficiently, seeing the electronic services in there, and hooking that very efficiently both to coherent and to switching. I think that you have to have those integrated in the future if you're going to have, I said earlier, scalable, simple and efficient. I don't see another alternate around this. So I think that we certainly have a significant competitive advantage in there. Will other people survive? Of course, they will.
I think we are being called off. For more of the -- at least the audience online, is that correct?
Yes. Actually, we're out of time for the webcast. So I want to thank you all for joining us today. And I'd like to thank the online audience for joining us. You can go to the Infinera website and learn more about the DTN-X at www.infinera.com/go/dtn-x. Thanks, Tom.
Thank you, sir. Thanks, all.