Great. Good morning, everyone. Again, thank you for joining the UBS Global Tech Conference here in Arizona. I'm David Vogt. I'm the hardware networking analyst here, and we're excited to have Lumentum. With me on the stage today is Alan Lowe, President and Chief Executive Officer. In the audience is Kathy Ta. I think many of you know from an investor relations perspective, so if you have any tough questions, you could save them for Kathy for later.
Exactly.
So again, Alan, thank you for joining.
Yeah, thanks for having us.
So, Alan, so we did a dinner last night. You've talked to a lot of investors this morning already. I think maybe it helps to start with, let's talk about your big end markets. I think most of the conversation, at least from an investor perspective that we get, is around your Cloud Light business, company that you bought almost over a year now.
Yeah.
Maybe let's talk about Cloud Light and how that fits into the portfolio and what we're seeing there, and we can jump off from there.
Sure. I mean, I think we worked on that acquisition for several months as we saw the drive for artificial intelligence, machine learning build-out really take hold. And so very happy with the acquisition and happy with the progress we've made since then. So we closed the deal in November of 2023 with a primarily large single customer, with a few smaller customers. And we've really done a concerted effort to try to broaden and diversify that customer base, as well as to grow our existing customers. So I think we're well on track to execute against that plan and making very good progress. So we're very happy with that.
Great. So maybe we can dig in a little bit about the market. So if we had this conversation a year ago around the time of the deal, maybe you had the foresight. I don't think we had the foresight to predict the strength that you're seeing from the hyperscaler community, your legacy existing customer, plus new customers. You've announced a second customer two quarters ago, a third customer on the most recent earnings call. Can you kind of share with us how we should think about, one, the deployment schedule for each of these respective customers? And two, maybe to the best that you can, talk about maybe the magnitude. I think at the time of the Cloud Light acquisition, you talked about an LTM basis of roughly $200 million for that business, primarily that one large customer, with a different mix going forward.
Obviously, I think everything's going to change, but maybe we can just start there and kind of talk about the timeline since you've owned it and where we sit today.
Yeah, sure. I think we had a hypothesis that the hyperscalers really wanted to be less dependent on Chinese-based and Chinese manufacturing companies to make their transceivers, and that's come out to be absolutely true in every case, so we've got lots of engagement with lots of hyperscalers today, whether it's the Tier 1 across the board or even the Tier 2s, so there's not been a lack of interest in U.S.-headquartered manufacturing outside of China, as we've had a large footprint already in Thailand and expanding that rapidly to handle the influx of new customers and new demand and growth in a very fast-growing market, so you're right. Two quarters ago, we announced that we had a design win. That's on track. We're going through the qualification work now, and we're still on track for the 1st half of next calendar year to start that initial ramp.
It's a large tier one. It can be as big, if not bigger, than our initial customer. And then in the most recent earnings call last month, we talked about a win qualification and production orders. And so that was a product that we had ready to go for this customer. The qualification went very smoothly. Orders are coming, and we're ramping today with more meaningful impact to our transceiver mix and customer base really throughout calendar 2025. So we're really excited with both of those announcements, and we have many more in the pipeline to be working on as well.
And so the initial premise that you had with Cloud Light and what's playing out today was that customers, particularly hyperscaler customers, preferred to have a US-domiciled business that they're working with. There still seems to be, obviously, a large number of Chinese-based competitors, US competitors. I think we're going to have one shortly thereafter your presentation. How do you think the market evolves over time in the intermediate term in terms of where sort of the customer base kind of migrates to? Can we have five, six, seven players in China, two to three players in the U.S.? Is the market growing fast enough and is large enough over the next, let's say, five years to support 2-3 different transceiver companies within different parts of the network?
I think it is. I mean, if you look at the industry estimates for growth rates at 40% CAGR for the next several years, I mean, the market is large and growing rapidly. So there's plenty of room, at least in today's environment, and I think for the next several years to accommodate all of the current existing, I like to call them customers and competitors because we are providers to most of them at the critical EML level. So I think from that perspective, the growth rate is good for us from a transceiver standpoint, but it's also good for us from enabling those customers of ours that are making transceivers with our EML chips, which is we're adding a lot of capacity there as well.
I was going to bring you into my follow-up question. When I look at the transceiver business that you bought from Cloud Light, obviously, there's an opportunity for you under your stewardship to vertically integrate, use your Thailand capacity, EML from Lumentum to structurally improve the profitability of that business. But at the same time, you're still working on selling EML to your competitors.
Our customers.
Customers, excuse me. Customers. I'll change the phrase.
Who also participate in our market.
Who also participate in your market?
Right.
So as we think about that going forward, obviously, you're adding a ton of capacity to not a ton, but $200 million of CapEx to kind of support the business. How do you see, are there any constraints from an EML perspective to meet not only your needs, but your customer needs over the next couple of years? And how should we think about sort of how that market develops?
Yeah. I mean, today, the demand for EMLs is far greater than the world supply of EMLs. And so to your point, we are adding capacity. And what we've said in the past is that we're adding from the June quarter of 2024 to the June quarter of 2025, adding 40% capacity. And then we're adding additional capacity in this back half of calendar 2025 as we convert to larger wafer sizes. And so that gives us, frees up a lot of capacity without a lot of incremental CapEx. And so I think we're still going to be behind on allocation through the calendar year. And we're not using our own EMLs in our transceiver designs today. Most of the transceivers that we design and we inherited from the acquisition are silicon photonics based.
And so we are actually buying CW lasers on the open market to save our EML capacity to support our customers, given the margin profile of those chips, and at the same time, being able to buy competitively priced CW lasers for our current product. So to your point, we are working on EML designs to design our EMLs into our transceivers, and that should give us a nice pickup in the margin for our transceivers when we ramp those up in the 2nd half of calendar 2025.
So from it, without getting overly technical, how do you see sort of the dividing line in terms of, from your perspective, a transition from silicon photonics to an EML-based transceiver? Is it customer-driven, speed-driven, your design sort of focus? What's the roadmap look like as we go from SiPho to EML?
I'd say it's all of the above. Some customers have a preference for one design type versus another. And some customers want diversity in both supply base as well as technology base. So they don't want to bet everything on EMLs or everything on silicon photonics. So it kind of depends. Historically, as I said before, the Cloud Light team that we acquired had been very silicon photonics focused. And so we've bolstered that team with EML designers for transceivers and next-generation product. And we have designs in the works today to utilize our EMLs that, again, will help drive our margins. So I think over time, it's going to be a mix of silicon photonics transceivers and EML transceivers. When we look at it, we look at what is our customer asking for, number one, but what can we provide to the market at the lowest possible cost?
And in some designs, it's a wash. And in some designs where, for instance, if you've heard of the FR4 or FR8, where there's four or eight different wavelengths of light that go through a single fiber versus a DR, where it's multiple lanes of the same wavelength over multiple fibers. In that multiple wavelength approach, EMLs work better just because you can have eight different wavelengths of EMLs that make that design much easier.
So does that then necessarily suggest that over the longer term, EML would be sort of the preferred technological sort of platform, effectively, if you will, relative to silicon photonics as we move forward?
I think so. I think that for the work that we're doing on next-generation 200 G EMLs, as well as we demonstrated a 400 G EML at ECOC this past few months ago, we think that there's a long life cycle in EMLs. Customers are familiar with how to design them in. And we're working to drive the power consumption of them down to make them even more attractive to the hyperscalers.
Got it. Along those lines, obviously, you mentioned adding EML capacity. Can we talk about other capacity, just straight-up manufacturing capacity in Thailand? You have facilities there. You own land around facilities. We talked about it a little bit last night. So maybe just to help frame kind of where your capacity sits today for optical transceivers, maybe if you can size what you can effectively sell today versus how you're thinking about the next couple of years in terms of as you add capacity.
Sure. I mean, we established a manufacturing presence in Thailand eight or nine years ago. So we have a management team, an engineering team, processes to be able to manufacture mostly our telecom and lasers products in the factory. To your point, we did buy the campus next to our existing campus to be able to expand, and six or seven months ago, we pulled the trigger on a very large infrastructure build-out of a three-story building that is well underway, and first floor is done. The existing campus, we had one and a half floors of space available. Those are now built out, and equipment is going in, so our qualification builds are going on now for that product we talked about that we announced two quarters ago in our bottom floor, but equipment for production is being put in in the top floor.
So we've got lots of capacity from a clean room and infrastructure standpoint. And we're making decisions as we make progress as to how much more test capacity and assembly capacity we need to be able to meet the demands of our customers. And right now, it's very strong. And so we are being aggressive in deploying capital to meet the needs of what our customers are telling us.
Correct me if I remember this incorrectly. I think two quarters ago, you said you were adding capacity with that win, but ahead of qualification.
Yes.
Is that the case?
You have to.
You have to, right? Okay. So as we sit here today with three live customers effectively, does your Thailand facility roadmap basically account for what you think the revenue opportunity is over the next 18-24 months? Or is there incremental capacity needed to kind of hit those revenue targets for those customers?
I would say from an infrastructure clean room standpoint, our plan has us going through the end of calendar 2025 with sufficient clean room. That doesn't mean that we have tests and CapEx that goes inside those clean rooms. We're measuring those out as we make progress and as customers give us commitments for volume.
So is it fair to say, so if I use my words, not yours, if I look at your relationship with that large customer, these other customers could be of similar size in terms of volume and revenue contribution to the transceiver business as they ramp to get to some degree of scale? Is that a reasonable approximation of the opportunity near term?
Yeah, absolutely. I think by the end of the year, there's no reason to believe they couldn't be equal size.
End of the year of 2020.
Calendar 2025.
2025, right? To be clear. Okay. And so in the context of your 2025 commentary that you've laid out at OFC and on other calls, I believe you've said run rate revenue of around total company revenue around $500 million. Have you shared, and I have it in my notes, I don't know if I heard you incorrectly or maybe I was making up the number, but what percentage of that $500 million could possibly be from the transceiver side as opposed to the more traditional telco side of the business?
Yeah. I think these three customers, they could be double our revenue for transceivers without any issue. I think we are seeing a pickup in what we call telecom business, but a lot of that's really driven by data center interconnect and the transport networks that hyperscalers are driving because they're putting data centers further and further apart. So we're not counting a lot on carrier CapEx spending increasing to get to that $500 million. But I think it's a combination of these two wins, and we don't need another win, but I expect to fully get those additional wins, a little bit of a pickup in the data center interconnect. We've already started to see, and we're actually having to deploy new capital to meet the demands of really that ZR component for data center interconnect.
I want to come back to the Datacom telco side in a second. So if I look at sort of the optical transceiver opportunity, you have three. You just mentioned you have confidence in a potential fourth win. When we move from the Big 4, maybe big five U.S. hyperscalers, how would you frame the opportunity in terms of the TAM effectively? Is there a way to frame it? Because I mean, look, the Big 5 account for the lion's share of the capital going into the ground, if you will. So is there an opportunity to add similar size scale and TAM with customers 6 through 15? Or does it drop off that dramatically from your perspective?
I don't think so. I think there's a lot of people building data centers, both in the U.S. as well as in China. So a lot of our EML chips end up going to the hyperscalers in China. So that's a great opportunity for us to grow our Datacom chip business. But I'd say that there's the tier ones that are huge, and we're still a very small player in a huge market. But there are that next level of hyperscalers that are really deploying new data centers in a meaningful way.
On traditional telco, not related to the DCI, as you mentioned, obviously, we've been in a very prolonged slump. Obviously, I know your crystal ball is not particularly great, but how do you think about where we are in the cycle? I mean, is this sort of the trough and things should be getting stronger in 2025? Networks have been swept. Assets are a bit older. Capacity needs to be upgraded. Where are we in that cycle? And I know, and Kathy will kill me for asking this, but previously, you were doing $350 million peak revenue in that broadly defined segment. What has to happen to get the business back towards that type of revenue contribution?
I think we're already seeing it in new products that weren't. Maybe I'll back up. Three years ago, we were ramping. We were shipping more than our customers were consuming, and it's taken two years plus for that inventory to burn off. I think there's still some inventory in the channel. It's getting much better level in that I think we're shipping more into our customers than they're shipping out, so over some period of time, that inventory will be depleted, but we're seeing strong pickup in higher speed coherent components and sub-systems, as well as in high-end ROADMs that weren't in production through the pandemic, so a lot of great progress on new product introductions that are driving carrier CapEx expansion for these new products.
And we're also seeing some sign of life in some of the older products where we thought that there was a lot of inventory in the channel. We're not counting on that to get to that $500 million. But I like to think that most of that challenge is behind us.
Got it. And then against that backdrop of that $500 million, obviously, I know Wajid's not here, but I'll still ask it anyway. When I think about the portfolio today versus five years ago, it's changed quite a bit, right? You had a much heavier consumer business five years ago. Now you have a much stronger optical transceiver business, really strong EML chip business. Kind of what are the puts and takes as you see the business going forward, not only a revenue perspective, but from a margin perspective? Because we get a lot of questions from investors. Will they get back into the consumer market? It was a great business for them. Now they have a great transceiver business, but the financial profile is a little bit different.
Yeah. So I can talk about all the different markets. So I maybe just give you kind of the current headwinds on transceiver margins. We are investing heavily in Thailand. We're hiring people. We're deploying capital. We're getting ready for the ramp. And until that gets filled up, that's a headwind. And you saw it in our margins last quarter. It's going to be a headwind for probably the next couple of quarters. But as we fill up that capacity, it will become much less of a headwind. You couple with that the product mix of our EML lasers. And I think you have a transceiver business that looks a lot like some of the other larger transceiver players. So I think we've got a good plan on that. I'd say that the growth in our EML business is highly accretive to our margins.
The variable margin on making another wafer is very high. So as we grow 40% and then another 40%, that incremental revenue is very profitable and drives margins. So I think what we said in OFC is at $600 million, we should get into the high teens of operating margin. I feel confident that's certainly the case, even with a heavy part of that growth coming from transceivers.
Transceivers. And then when I think about sort of the portfolio today, and I'll go back to the consumer comment, obviously, look, there are some issues with some of the other businesses macro related, like auto, industrial hasn't been a great market generically, not for Lumentum, but industry-wide. How are you thinking about your commitment to these markets, given the success in sort of the data center hyperscaler market, where maybe the auto market isn't a great market over the next three to five years? Is there an allocation of capital decision that needs to be made where this is where all the growth is for the next five years? And maybe be honest, maybe the industrial auto market is just not a particularly good market for you going forward, and consumer is not a great market going forward.
I'd say certainly we've made reallocations of R&D spend as a result of what you talked about. I mean, consumer is a fraction of what it used to be, but still generates cash for us. So it's something that as we size the OpEx for those businesses appropriately, they're still a good contributor to the profitability of the business. On the industrial side of the business, I don't think anybody's happy with how the industrial market is. We feel very confident in our outlook for the industrial business. We had our kilowatt fiber laser business, much like the telecom business, was fearful that they couldn't get the inventory and.
People bought a lot.
Yeah, lots of inventory. That is mostly behind us. And I think that in the next couple of quarters, we'll start seeing a pickup in that business. So I think while it's a tough business, it's still something that we're investing in, not to the same degree as we were when the consumer business was $500 million a year.
Got it. And then maybe just on going back to the macro, we've asked everyone this. Obviously, the economic climate's a little bit uncertain. We've got a change in the administration. You have tariffs. Can you remind us again what you learned last go-around with tariffs, given what you're putting in the ground in Thailand? Obviously, it doesn't sound like it's at risk. How are we thinking about kind of the preliminary view around tariffs and what potentially could come down the pipeline?
Yeah, I don't think our strategy's changed because of who's going to be in the White House, because we were already on the path to be less dependent on China from a manufacturing standpoint as well as from a supplier standpoint. So that effort has been going on for several years. In fact, a year ago, we closed two factories in China and bringing them up in our Thailand facility. So that strategy is going to continue. Our hyperscale customers want us to continue that with not only, well, with the Cloud Light, now our Lumentum Data com modules. As we bring up Thailand, the challenge we have is they don't want us to take down the capacity because they need the product so much. So we're focused on bringing up Thailand to a meaningful size before we even touch moving that transceiver business out of China.
Now, might that change if tariffs go up dramatically? It might accelerate things. But right now, our customers don't want us to miss a single shipment. So.
Is that an incremental margin headwind in the short run as you maintain effectively kind of double dual infrastructure effectively?
I mean, it is because we don't have volume in Thailand, but once we fill that up in Thailand.
And then you can ramp down.
We can ramp down in China, assuming that there's the ability to do that and not impact our customer deliveries.
Got it. Okay. Also, just one more on telco. If we're having a conversation 12 months from now, where do you think the growth is? You mentioned ROADMs earlier. Outside of ROADMs and some other markets, where do you think the growth has cleared from a customer inventory perspective? Where have we digested? Where is there still too much inventory out there from your seat, from your product portfolio?
There used to be a large number of tunable lasers in the market. That's all gone.
That's clear.
Gone. I'd say that the high-speed coherent components, so like the high-speed modulators, coherent modulators, there was no inventory because it's a new product. And so as customers introduce 1.2T and 1.6T coherent modems, those products are flying off the shelf. So in China, in North America, and in Europe, that's a good part of our business. I'd say there's still probably some of the slower coherent stuff that still needs to get burned off. But we're seeing good signs that that's getting taken care of as well.
Got it. And then just one question that we got here was on the transceiver business. I'll paraphrase. Anything on 1.6T that you can share with us, what you're seeing out there in the marketplace in terms of qualifications and?
Yeah. I mean, I think we're working on our EMLs for the 1.6T and very happy with the progress we've done there. We've sampled 1.6T, both silicon photonics as well as EML-based transceivers. I'd say that for us, it's probably by the end of the fiscal year or early fiscal 2026, so summertime, before we get into production on the transceivers, on the EMLs, it'll be before then.
Before then. Got it. And can you just remind us again? This came in. I missed it earlier. What percentage of Cloud Light sits at 400G versus different speeds today? And what would that look like maybe a year from now?
Yeah. When we acquired it, more than 50% was 800 G, and the rest was 400 G. We don't really do anything slower than 400 G. That mix is pretty much unchanged. As we introduce new products, there will be a shift towards more 800 G, less 400 G, and I'd say a year from now, there'll be a lot of 400 G, but it'll be a combination of the workhorse being 800 G and 1.6T becoming a more meaningful part of it.
800 G, EML, 400 G.
Multi-mode and SiPho.
Okay. Multi-mode. Okay. Got it.
So we do some VCSEL-based stuff for multi-mode, both active optical cables as well as the SR transceivers.
Got it. Final question for me on the transceiver side. Obviously, you touched on it briefly on margins. So we're going to vertically integrate. We're going to have our own EML capacity. How are you thinking about best-in-breed margins? You mentioned it briefly. You've got an assembler in China, InnoLight. You've got a vertically integrated competitor/customer here in the U.S. If you're at scale from a transceiver business, do you have a sense for what you think the business should look like from a profitability perspective that you're willing to share?
Yeah. I keep telling my ops guys to look at InnoLight's margins and why can't we be there, even without our own product in feeds, and so they have huge scale compared to us. They're very good at assembling other people's stuff, and they've designed products very well, but there's no reason we shouldn't be at least as good as them, and then when we add our in feeds, we should be better. I'm not counting on that. It's not in any of our models, and we're not giving guidance to that.
No. It's kind of the dream scenario.
Yeah. It's... well.
Aspirational. Aspirational.
Well, it's my expectations of my team, right? So I think.
We have a new team.
Yeah.
Final question on capital, cash flow priorities. Obviously, your customers are asking you to meet pretty strong demand. How are we thinking about the balance between investing to meet your customer demand on CapEx versus the traditional concern that this industry is somewhat cyclical by nature? Visibility tends to be somewhat limited over a long period of time. So how are we thinking about that? And you have a lot of cash on your balance sheet. How are we thinking about deploying that cash?
Sure. So I'd say I've been through the slowdowns before, and our biggest use of cash today is for CapEx for our wafer fabs for driving EML lasers. I believe that that business is going to continue to be on allocation for at least 18 months, and again, as I said before, we're not producing our own CW lasers. We have the capability of doing that, so if there is a slowdown in the growth rates, we will move that CW laser capacity internal, and we will produce CW lasers, so we won't have a bunch of idle wafer fab capacity there. On the transceiver side, where we're investing as well, I'm pretty comfortable with the fact that we'll fill that up.
Again, being such a small part of the business, a small part of the market today, with aspirations to grow dramatically, that growth really for the next 12 months, I'm not worried about.
Got it. And what could govern or what could be an impact on the growth in terms of customer acceptance at this point? I mean, it seems like the demand is somewhat insatiable. I'm just trying to figure out what could potentially hold back demand at this point.
I think what keeps me up at night is we have so many opportunities to design products for customers that I don't have the R&D bandwidth yet, and so spreading my R&D teams too thin and misstepping. That's what keeps me up at night. I think we've got a solid team working on our transceiver designs, and progress to date has been really, really good. I'm pretty excited.
Great. All right. I think that's it for me for now. Thank you, Alan. Thank you, Kathy.
Thanks, David.
Thank you, everyone, for joining, and good luck, and we'll speak to you soon.
Great. Thanks, David.