Good afternoon, everyone. Welcome to day two of Citi Global Technology Conference. My name is Atif Malik. I cover U.S. semiconductor, semiconductor equipment, and communication equipment stocks here at Citi. It's my pleasure to welcome James Moylan, CFO, Ciena, as well as Patti from Investor Relations. I'll kick it off with my questions first, and then we'll take a break and see if you have any questions. If you have a question, please raise your hand, and the mic will come to you. Yeah, welcome, Jim.
Thank you. Good to be here.
Jim, just to get started, just want to give you an opportunity to recap your great last quarter and maybe touch high level on the end market dynamics.
Yeah, one of the things that we focused on in our call was something that, I think is a, a little bit of a misunderstanding with respect to how we look at our business. We look at our business sort of in a, in a stage-like approach. The fundamental demand for bandwidth has been growing at 30% a year for a long, long time. Of course, the web scalers have contributed large amounts to that growth in bandwidth, but there's a lot of other applications and use cases which are driving bandwidth demand. That has continued really without any sort of cyclical change for at least two decades and probably longer than that. But when we think about how that drives our business, we say: the first thing that we're looking at is what are our customers doing? What, what's the pipeline look like?
We have top 40 or 50 customers, which make up 90% of our revenue. We have a dedicated sales force to each one of those customers, and so there's a constant communication with that customer, and that sales team knows a lot about what's going on in that customer. That's what we talk about when we say pipeline. We're talking about the amount of engagement, the new RFPs, the new upgrades that are coming, not all next quarter and not even some this year. But we see in terms of that activity, that the pipeline is strong and growing. That's, and so we talked about that as a stage of demand. The next stage of demand is when you actually get into the, you know, you've won an RFP, and they start ordering. And ordering, orders is something that, I think people accentuate the importance.
I'm not saying they're not important. They're absolutely important. It's what's gonna drive our revenue. But we can have periods of time in which our orders are not gonna be as high as we think they might be, given what's going on with respect to underlying demand and what's going on with respect to the pipeline. Orders are an indicator. They're not the be-all and end-all, and I really think that the market places too high a degree of importance on orders. Now, understand, that's what we and everybody in our industry talks about, and we can, you know, we can somewhat quantify it, not necessarily in dollars, but at least in direction. And so I understand that's part of the reason for that focus, but I think it's a lot more going on than that.
And we've tried to say to the world, and we tried to say on that call, there's a lot more going on. We don't necessarily talk about it all the time, but we have a very healthy business, and it's growing. And then finally, we talked about backlog, and we all know backlog. So, that's, that's a, you know, a short version of what we talked about on the call, just to try to give, give the market a view as to why we feel so confident about our business. All of the elements of that, of those stages of demand are strong. Pipeline is strong, orders are finally coming back to where... they're not where we'd like them to be, but they're starting to come back to that, in that direction, and our backlog is very strong.
So that was just the content part, or the context part of answering your question, Atif. On the quarter, this year, we entered the year with a backlog of well over $4 billion. I forget the exact size of the backlog. And it's because during the supply chain situation, our lead times, as well as the lead times of our vendors, extended dramatically. Prior to that time, we were delivering, in some cases, in as short as two weeks, in many cases, four to six weeks. Backlog lead times extended to 52 weeks for a wide range of products. Not all, but a wide range of products. That forced our customers into a situation, at least our big customers, into a situation of ordering ahead of time, and that's where our backlog grew.
We've had a lot of visibility into what we were gonna deliver in a particular quarter, really for a year and a half, as a result of that. We knew that we didn't have to have much in the way of orders to hit good numbers. Now, we were constrained in some cases by the absence or inability to get some parts. What I would say is that we have actually done a little better than we've guided to in each of the last three quarters because our supply chain situation has improved sequentially each quarter. Now, I will not say that the world is the way it used to be and everything is back to normal, but I would say that we're approaching normal on the supply chain side.
We still have some extended lead times, but at least we're getting what we ordered, and we're getting it in the volumes that we asked for. So as we came into third quarter, we had a very good setup. We could see what the customers wanted. We had the ability to deliver, and we delivered. I'd say there's one small exception to that, and that is that we are still getting a bit of churn within customer sets about what they want and when they want it. And as a consequence, we had built some goods in anticipation of demand, which they changed. They changed their orders and changed what they wanted to order. So if you looked closely at our inventory, you would see that our inventory did not come down in the quarter.
It went up a bit, and a lot of that increase was in finished goods inventory because of this change in the mix. So it was a great quarter from an income statement point of view. We didn't have a ton of cash that was generated this quarter, and our balance sheet is still too heavily disproportionate to inventory, but that's gonna change as we move through this quarter and in the next few quarters.
Great. Jim, you talked about the need for bandwidth growing at 30%. And can you talk about the nuances between 5G, cloud, and now AI bandwidth?
Yes. Cloud, you know, there are some drivers of demand, and then there are some enablers of capacity movements. You think about cloud. That is clearly a driver of demand because what's happening is that instead of data centers being on the premise, they're up in the cloud. Data has to flow back and forth from the cloud. All of the cloud-based companies, the web-scale companies, have a basically a cloud-based business, so all of the demand for their services flow through networks. That's. Those are the two big drivers. 5G is an interesting one.
5G is sort of an enabler of this, increase in demand because it allows much, more data to be, sent around to various places no matter where you are in the world, and, requires shorter reach in terms of the antennas and enables the flow of data. It's not necessarily a driver of demand. The big drivers of demand are movement to the cloud and the cloud, cloud companies whose business is by their very nature, demand bandwidth.
Jim, on the uptick in the cloud orders that you guys talked about, I'm curious if it's due to AI or progressive improvement in inventory digestion. And like you guys said on the call, you know, first in, first out, the market that went to correction first should come out first. So if you can help us provide any color on the cloud orders.
Yes. It came from a couple of the big cloud companies that are our customers. It's very encouraging to us because we knew that our customers generally are operating their systems a little hotter than they like to. There has been some push out of demand for network products because of the overwhelming demand for compute inside the data center. So we suspected that this was going to happen. It's a fairly sizable piece of business, a couple of pieces of business. It is this year's demand. It's demand for this year. It's not a pull-in from 2024, which we think is very encouraging, and it's the same use cases that we have. It's connections between data centers, but they both, in their discussions with us about this need, talked about AI.
In one case, it was clear that AI was driving demand on their network, and in the other case, wasn't clear. They were seeing some. They didn't know how much they were gonna see, but they wanted to get ahead of the game. So those, those events were extremely encouraging to us. We felt it was inevitable it was going to happen as long as demand for bandwidth continues to grow, shows no sign of stopping. But the fact that it has come in and the underlying reasons that they cite and the fact that they cite it for 2023 demand are all very encouraging to us.
Great. Jim, I don't follow you guys closely, but I do follow the DSP providers like Marvell, and, you know, they've been seeing a pickup in orders as well. And so just trying to understand the landscape, when you talk about, you know, converged, you know, optical packets, is it the same demand that they're seeing, and you're getting the DSPs from them? Can you help us understand, like, your place in the ecosystem?
We have our own DSP in optics. We don't require... We don't use their DSP. They are also, Marvell is inside the data center in some of their business. And so I think that, I would guess, I don't know, you would know more than I, but I would guess that much of their demand is driven by the needs inside the data centers for AI and for compute. So I think that's what's driving them. It's not anything that we're doing for them, but the fact that they're seeing demand is gonna fall over to us, I mean, it, as it described, it already seems to be happening.
Okay. Any kind of color you can provide, is it 400 Gb or 800 Gb? Or what's driving the uptick in the cloud orders?
Yes. It's, I guess what I'd say is it's a, it's sort of a complex situation. There are four big web scale, mega scale companies to which we are vendors. And you would know who they are. They're all household names. In addition to that set of vendors, we have 50 or 60 other, what we consider web scale companies. It's a bit of a misnomer, but it's companies that have a web or software-based business model and build data centers. And so that's, that's the universe of what we call web scale. It is very concentrated, and the four of those are the biggest part of the equation for us.
Now, they buy, for the most part, a product that we call Waveserver, which is a simplified version of the box that we would sell to a service provider customer. The service providers require a very complicated system which allows for the addition and the dropping of capacity at various points, and allows for a very dynamic, in many cases, software-driven ability to manage demand. Whereas the service providers, for the most part, want to connect data centers, and they want as much capacity as they can get. We have been the leader in coming to market with each successive generation of optical coherent technology, from 100 Gb to 200 Gb, to 400 Gb to 800 Gb.
We will have 1.6 Tb in the market next year, and it's because of their tremendous demands for capacity that they are the first to move. I said it's complicated, though, because not all web scale companies operate their networks exactly the same way, and it makes it, it's common sense. You wouldn't want to put an 800 Gb product in a place where you only need 100 Gb, right? You'd buy a 100 Gb product, or maybe you'd buy a 200 Gb p roduct to provide for growth. So and the upshot of all of that is we introduced the 800 Gb product to market, first to market with that in October 2022, Patty? I think it was 2022. We were the only ones with a product for about a year, a little over a year.
There's one other competitor with that product. It is now the largest selling product that we have. We call it WaveLogic 5, but we still sell a lot of WaveLogic 4s, Wave and WaveLogic 3s, which take you down to 400 Gb and to 100 Gb. So it's, it's complicated, but the Web scalers are always going to be the first to grab that capacity, and they, they are the first movers.
Right. Jim, you spoke about $4 billion in backlog. In terms of converting the backlog into revenue, what are your expectations? You know, kind of is it 40% of it you could convert into revenues by the end of this year or into next year? And is that translating into share gains for you this year?
I think without a doubt, it's translating into revenue share gains this year because we're going to grow at, you know, if you just take the midpoint of our guide, 20%, and the market's not growing at 20%. So I think if, if you just look at that measure in isolation, clearly we're gaining share this year. If you go back a longer period of time, since 10 years ago, 12 years ago, we have gained roughly a percentage point of share in the optical market outside of China every year. So we've gone from a 10%, 12% market share 10 or 12 years ago. Today, at the midpoint of this year, we had, according to the analyst data, we had about 29% of share. So that's, that's the kind of share gains that we've had.
I would say that has not been a, you know, straight up into the right line. There are movements in it, but we're clearly gaining share this year. I think we're going to continue to gain share over time. We have the best optical technology in the world. We are developing a very good routing and switching capability that will become important as networks converge. And so, I believe that we will continue to gain share and grow faster than the market.
Sure. Can you talk about what you're seeing on the telco side as it appears, like, coherent, even more relevantly, the continue to see incremental weakness there?
Well, I'd say that we're not seeing a bunch of orders from them. I'd say that they were big orders early. They have a fair amount of inventory. Demands on their systems are still growing. And, you know, as clearly the drivers of demands on their system, since they're essentially the connection between the web scalers and the customers, they're driven by what's going on with the web scalers. That's why when we see things happening with web scalers, we think that it's going to trickle down into the service providers, and we think that'll happen again. We do know that they're busy thinking about the next generation and what they're going to do with our latest technology. That's not going to happen this quarter or next quarter, but it's going to happen.
But, you know, I, I'm very confident that the service providers, the Tier 1 service providers, are going to continue to be an important part of our business for the foreseeable future.
Great. Jim, there's a bit of a debate, and I know it's a technical question, but how do you see AI impacting the technology trajectory of, you know, co-packaged optics versus linear drives versus DSPs?
... Yeah, the one thing that's been a constant over a long, long time is that the need for capacity inside of data centers has grown at phenomenal rates. It's probably growing faster than 30% a year. AI is creating an enormous need for compute inside the data center. Now, what that means is that power has become the most important characteristic for cloud-based companies or data center-based companies, I should say. The access to power and the cost of power is the most important element of their business today. I mean, other than making their customers happy, I'd say that's the first. The second is the power that they need inside the data centers.
What that means is that you would just-- I mean, if you-- I've seen some crazy predictions of how many power plants have to be built in order to meet the demands of AI, and, you know, it's not gonna happen. There will be innovation that will prevent it, but I just think it's very interesting to take current trends and extend them out farther. One of the important elements for us is that, as demands for compute raise, it is a necessity that the power consumed per unit of compute come down. It has to happen, and it will happen. Now, that means that the old way of communicating inside of data centers, which is essentially an old style of optical technology called Direct Connect, is inefficient, power-consumptive, and results in a mess of cabling inside a data center.
All those things need to be better for the data center guys, and it's pointing them toward new technologies to operate the communication systems inside the data centers. Now, there are three contenders for that. One is this, I can never think of the name of it.
Linear drive.
Yeah, linear drive. That's one. Co-packaged optics is another, and the third is to implement coherent optics inside the data center. Today, it's unclear which of those technologies will prevail. I will say that we are engaged with some of the web scale companies about how we can develop coherent technologies inside the data center that for use inside the data center. Our roadmap for our WaveLogic 6 DSP, which is coming out next year, includes a path toward and inside the data center product, and we are working with them to develop that technology. It's hard to predict which one of those will be the winning technology. I think you could see one or two of the others come into play before coherent inside the data center.
But I do believe inside the data center, coherent will become a feature of their businesses, and, it'll be a different market for us. I think we may be in the selling DSPs business into that market, but it's something that we're willing to do if we think it's gonna be profitable for us and we can serve our customers. So, I'm cautiously optimistic that there will be an opportunity for us to play there, but it's not in any of our numbers today, and it's, you know, it's a couple of years out, maybe several years out.
When it comes to new switching or routing orders improvement, are those tied to new projects or renewals?
It's a mix of things, but for the most part, it's new projects. Now, we have been in the business of selling access switching for 12 years or so, and the applications for that tend to be around cell tower communication back to the core, business enterprise services, where we install the gear at or near the enterprise. That's what, that's been our switching business for a long time. We've since significantly expanded our switching capability. We now have edge routing capability. It's not a full core router. It can't send a signal across the continent, it, but it can send a signal to the next best hop, and that's good enough in terms of an access router. So we sell that. We have two or three customers, including two big customers, for that. We're excited about that.
We consider our PON activities a routing and switching product, and we sell the switches that are in back of that plug, and that's growing, and all those are brand-new projects for us. Then finally, in the metro, which is where the most exciting kind of change is fermenting, and it has to do with a convergence between optical switching and routing products in the metro part of the network. Because demands on that part of the network are so high and the CapEx needs under the old architecture are so high, there needs to be a convergence, products that can do enough of each of those capabilities to meet the needs at a more economical price point. That's very exciting. We talked about our WaveRouter. That's what that product does.
It is a converged optical routing box, which meets the demands for the metro convergence. We, we're in market with it now. We have one order with a big company. Customers are very excited by it because the number of companies that are going to be able to do this convergence, optical and routing, are very limited. It's really basically Nokia, ourselves, and Cisco, and Huawei, of course, but Huawei can't play in a lot of countries. So it's really down to, unless we're able to be a part of that market, they're only going to have two choices, and they don't like that. So I'd say we have a good chance of winning our fair share of that business. It's a good product, customers are excited, and we'll see where it goes from here.
Let me pause here and see if there are any questions in the audience.
Yes.
Maybe one question on the DSP side, particularly for cloud and, sorry, AI. Obviously, we have seen one part of AI, which is GPU, is doing very well. And when we talk to networking companies, it seems like networking is kind of waiting for Ethernet to gain more share versus InfiniBand. I guess my question is, in terms of the infrastructure build for AI, when does DSP come in? Is it agnostic between InfiniBand and Ethernet, and maybe some attach rate as well that you can provide from DSP to GPUs?
All of the activity, essentially all of the activity for AI machine learning so far has been inside the data center, and it's all compute. I mean, of course, there's storage involved, yes, and connections inside the data center, but the compute is the most important and massive piece of the demand so far. When you think about that, we've always felt, and, and that, by the way, that involves DSPs and a whole bunch of other things. It's all kinds of things that compute, store, and connect inside the data center. That's not the business we're in as of today. I said there's a chance that we can get inside the data center, and, and I think we'll have an opportunity, but it's, it's a ways off.
Now, we've always felt intuitively that this amount of compute inside the data center is going to result in demand outside the data center. And it seems clear that they can't do all this compute in one data center. They have to do it in multiple data centers, and so they have to send data between data centers. And that is going to result in demand for our products, which are data center interconnections. We hadn't seen any proof points of that until this past quarter when we talked about it. We did say that the two big cloud providers that we received orders from, they're either directly connected or indirectly connected to AI. So we are starting to see it. It's a proof point of what we assumed would happen.
You know, one thing that many people have asked is: Is there a way to estimate, you know, for every whatever, unit of compute, how many units of connect where there'll have to be? The answer is, nobody knows the answer to that question because nobody knows how much spillover is going to occur from data being developed and brought into the data center and data being sent to wherever it's going to be sent to. So we, the fact that we've started to see it is extremely encouraging to us, but no way of estimating how much or when or anything else in that regard.
Jim?
Hey, Jim. U.S. carriers have talked about taking down CapEx next year, which would seemingly be a headwind to some of your business. Inside of that, it looks like maybe 5G spending is switching from coverage to capacity. What does that mean for your business? Like, what, what part of your business might that impact as a potential tailwind?
When you think about it, I know it's true for the big Tier 1s and the big Tier 2s. Whatever they spend in CapEx, and I know that AT&T spends $20 billion and Verizon spends something like $18 billion, or maybe it's $20 billion, but the amount that they spend on optics, on what we sell them, is a very small fraction of their CapEx spend. They spend all kind of money on fiber and laying fiber and, you know, putting up telephone poles and all this other stuff that they do. And so there can be pretty significant swings in their CapEx without affecting us. We've seen that in the past. Other times, we've seen where they have directed capital toward the RAN part of the network and away from us. We've seen that happen.
Now, in particular, AT&T and Verizon, about two years ago, said specifically that they were going to increase their CapEx for a period of time. Verizon said, I believe they were going to spend $10 billion over three years. AT&T said they were going to spend $2-$3 billion extra over the next couple of years to build out their fiber and to get ready for 5G. We saw that happen. I do believe that some of this lower CapEx from the big guys is because they finished that fiber build-out or just about finished it, and so that's a part of it. Having said all that, we do know that they still have higher inventory levels than they'd like to carry. They're always gonna carry some inventory because they don't want to go bare.
But we know that that is a headwind to us as we enter the first couple of quarters of the year. As long as the demand for bandwidth continues, and they've finished their fiber build-outs, then that should help us as we get into the second half of next year. We don't look for a lot of orders from them, from the service routers, for the first couple of quarters. Now, things change. It could be that they will come back to us in a rush, but we're confident that they're gonna have to come back. They're just running their networks too hot not to.
And then just one quick follow-up on your comment about potentially being a merchant DSP vendor. Can you just, like, what would the timing of that be?
Yeah.
I mean, is that, like, even in terms of having to sort of plan that out from the business perspective?
It's not gonna be next year, unlikely to be 2025. I think the earliest it could be a meaningful part of our business would be 2026. It does require a very different looking supply chain and a willingness to handle components in massive, at massive scale. And that's why I think we're gonna have an opportunity. I'm not sure we're gonna want the opportunity, because it is a big change for us, and it needs to be reasonably profitable for us to embark down this path. So, I mention it only because I think it's a possibility. I'm not putting any numbers in anything as we sit here today. But as I say, it is encouraging to me that the web scalers want to talk to us.
It says something about our ability to design optical DSPs.
Thanks. Questions? Yep.
Sorry, it's me again. Oh, maybe just on the, some of your newer products, so Wave, WaveLogic 6, and I think WaveRouter, last quarter, I believe you mentioned having your first order, first customers. Can you just maybe pull the curtain a little bit and give us more detail on how those discussions are going and those new kind of discussion with new customers for those products?
Yeah. Well, you can imagine that, we have been talking to potential customers for WaveRouter for months, and Tier 1 providers around the world have participated in the process of developing WaveRouter, and we have, they've helped us design it, so it has the most general application. All along this path, they have all been extremely, encouraging and excited about the product. One reason is that, as I said, it is a product which is custom-designed for this converged metro, play, and they like it. The cost point is something that they believe is gonna help them. The capabilities are something that meet their demands, and so they're excited from that point of view. But as I also said, they're excited because the number of companies that are gonna be able to play in this market are, are very limited, and they want choice.
You want choice when you buy things. And, so it's not surprising that they're speaking in very encouraging tones. I think the fact that we were able to get to where we got as quickly as we did surprised some people, who weren't sure that we would be able to develop that converged capability, including the routing piece of it, as quickly as we did. WaveRouter is not a finished product by any means. It's generally available, but it's in a, you know, a starting form, and there will be many, many advancements and functionalities added to it as we go through time. But I'd say without exception, the customer base is excited. They like it. They want it in a bigger form factor, more capacity, and they want it in a smaller form factor, less capacity.
I mean, they, they like what we've built. It seems to be down the middle of the road, but they also want it smaller and larger, so we'll walk down that path as well. It's very exciting.
Yeah.
Jim, can you talk about, on the supply side of the things, you know, where is Ciena, you know, doing post, like, the COVID disruption and the impact on your profitability, both gross margins and operating margins?
Yes. Yeah, I know for many companies in our space, the last three years have been very exciting and, lots of changes. But the fundamental fact was that, COVID interrupted the flow of many of our customers, slowed their ability to, implement new projects. And then when COVID, when they learned, and we all learned how to deal with COVID, they came back with a rush, which created all sorts of demands on the supply chain and supply chains in many industries that, the supply chains were not able to keep up with. So our supply chain lead times went from four weeks and six weeks to 52 weeks. They're now well below 26 weeks. I'd say 12-18 weeks is not a bad range for our lead times.
A corresponding thing has happened with respect to our component suppliers. Their lead times got out there and have come way back in. So we're approaching a state of semi-normalcy inside the supply chain. It's unclear what will happen as we move through time here, and by that I mean, if you go back three and four years, this business, as well as many other businesses in this networking space, operated their supply chains on a just-in-time basis. They took orders in a given quarter, they might deliver half of those orders or 30% of those orders in that quarter, and then the rest they delivered the next quarter. It was just in time, and we could manage our supply chain with, you know, an availability to acquire unlimited components quickly in, you know, that worked. Everything worked nicely.
COVID and the resulting supply chain has. And the clear risks that a just-in-time supply chain entails have certainly caused us to change our behavior going forward. We're not, we're not ever gonna be just-in-time again. We're gonna keep more buffer inventory that, you know, we're gonna. Our inventory is gonna come way down from what it is now, but it's never gonna be as low in terms of terms, turns as it was. So, that's one thing that's happening. It's as we speak. The other thing that's interesting is customers, for the most part, our big customers don't need just-in-time business. They can order with a quarter of lead time, and, you know, they plan their business, their networks are complicated enough that they have to plan their business over quarters and months.
So, they can operate with a quarter of lead times. Now, there are some parts of the market, places where people compete on a, on a lead time basis. Basically, the wholesale network providers, they're gonna want short lead times, but I don't know that our customers in general, the biggest part of our business, are gonna need that shorter lead times. All this will be played out as we move through the next year or so. But I think that some elements of the supply chain have changed forever. I don't think just-in-time is gonna be the prevailing way of doing business going forward.
Great. We're almost out of time. Thank you, Jim and Patti, for coming to the Citi conference.
Thank you.
Thanks for having us.