Lumentum Holdings Inc. (LITE) Earnings Call Transcript & Summary

Good afternoon, everyone. Thank you for joining this session. Really delighted to have Alan Lowe, CEO of Lumentum join us this afternoon. I'm Vivek Arya from BofA Securities. And as usual, I'll go through my questions, but please feel free to raise your hand if you would like to bring anything up. But really warm welcome, Alan. Really appreciate you joining us at our conference.

Great. Thanks for having us.

And maybe let's start at the high level. Just maybe give us a state of the union, Alan, as you see in terms of end market demand across your cloud and telecom and industrial tech markets as you look into the second half.

Sure. Well, certainly, the cloud is the most exciting part of our markets and especially with our acquisition of Cloud Light and getting this into -- directly into the hyperscalers and the AI infrastructure players. I'd say that's an area that we're investing deeply in both in R&D as well as infrastructure and capital in preparation for customer diversification and ramping of new products at 200 gig per lane transceivers. And so that's a very exciting time where I'm spending a great deal of my focus and effort and the team's focus and effort to make sure we capitalize on that opportunity as the demand is very broad, not only from the AI players but also from the cloud players because they tend to lag the AI players. And so AI next-generation products are 1.60 and -- but there's a big market for -- that is emerging and growing again at 400 and 800 gig. So that's an exciting opportunity as well. On the telecom side, we're continuing to see inventory being burned off at our customers and at their customers. And eventually, it will run out, but we still haven't seen the end of that yet. But encouraged by the progress. But the bandwidth demands at telecom networks as well as in between cloud data centers is continuing to grow very rapidly. So once we get through the inventory situation, I think we'll be up and running in that space. And also the new products are seeing strong demand where there is inventory in the channel. So that's encouraging. And then industrial tech, I think that's really 2 areas: one, industrial lasers, and that is inventory issue as well, but we're encouraged by how that's getting situated; and then in our consumer market, we had a very large market share, and now we share that market with 2 other players. So that's gone from a very large customer to less than 5% of our business. So I think that headwind is behind us. And I think we look forward to keeping that at a low level of our revenue but still contributed in a meaningful way.

Got it. You mentioned we are at the start of the 1.6 terabit per second transceiver generation. Maybe walk us through, Alan, like you've been in this industry for a long time. How has this transition different from what you saw as the industry moved towards 400 and then 800. So first of all, what does it mean from an industry perspective? And then specifically, how does that impact Lumentum's prospects?

Yes. I think it's different that we've seen in the past in that the transition from 400 to 800 gig happened a couple of years ago. And now we're already going to 1.6 terabit, where from 400 to 800 probably took 4 years. Now we're doubling in 2 years. And so I think from that perspective, it's all about the fundamental technology that enables that transition to be so much more rapid. So the combination of Cloud Light and Lumentum together, where Lumentum really has the enabling technology at the chip level, both on EMLs as well as silicon photonics and CW lasers. So that positions us quite well along with Cloud Light. And so as we now are a U.S.-headquartered company, making transceivers with manufacturing outside of China, we're getting a lot of traction with respect to the -- really the leading edge hyperscalers and infrastructure providers on how do we make sure we're ready for that 1.60 ramp and those qualification samples coming out of our Thailand facility this summer are really the next key milestone to that progression. So it's an exciting opportunity for us and one that we're, again, investing deeply in.

How do you see the content change as you double the speed? I don't know whether it's quite a double of what you saw in 800 gig, but how does the content change? And then the other aspect of it is that we do see optical transceiver price compression come down. So you think with 1.60, it will kind of follow that historical compression levels that we have seen? Or do you think there will be something different about 1.60?

I think it depends. I think right now, as I look out over the next couple of years, I see capacity constraints in lasers in indium phosphide worldwide based on the customers' demands today says we're going to have a problem as an industry to be able to meet the demands especially as you look at whether it's CW lasers, which are bigger chips that take up more wafer or using 8 EML lasers on a 1.6 terabit. That is going to drive the demand for and imposed wafer fab capacity that's going to limit the ability of the demand to be met. And I think that puts us in a good position, given that we have our own capability, but we're also I think in the first-to-market perspective with respect to the 200 gig per lane EMLs and the high-power CW lasers. So I think from that perspective, having the vertical integration capability gives us a good position on 1.60 ramp.

As you mentioned, the 200 gig EMLs are ramping actually right at this time, right? So it's kind of the sweet spot of when the ramp is. How big can 1.60 be this year? Or is it more of a 2025 growth driver, you think?

Yes, it's more of a 2025 thing. I think we've qualified our -- internally qualified our EMLs today. We've tested them at our transceiver R&D team. We've sampled EMLs to all the transceiver companies gotten very positive feedback. But that's step one. Step two is providing the transceivers to our customers, and that's going to happen in the summer or our transceiver customers supplying them as well. And then it takes probably 3 to 6 months before meaningful revenue comes. And so I'd say, ramp really towards the very end of this year, but more meaningful into the March and June quarters of 2025.

Got it. One thing that you presented at OFC was a $16 billion addressable opportunity for Cloud Light. Could you help unpack, what are the big kind of moving pieces of this $16 billion by 2028.

Yes, we said that it's $16 billion for the cloud itself, not necessarily cloud. So the transceivers is a big part of that, and it's growing the 400 and above transceivers is growing very rapidly. But I'd say also that the ZR market for data center interconnect is going to grow rapidly as well as data centers need to find power they can't put them next to each other anymore. So there's going to be more demand for 400 gig and now soon to be 800-gig ZR data center interconnect as well as the optical switching inside the data sets. So we include that into that $16 billion. And I think all of those together are big opportunities are really in our wheelhouse to be able to provide lower power, lower cost infrastructure for the hyperscalers and AI applications. So today, transceivers are $4 billion to $5 billion, and that's going to grow 30-plus percent a year at the high end. And so that gives us a big chunk of that growth. And we're a small player today. And I think we have a very good opportunity given our manufacturing infrastructure and capability to really grow a significant amount of share there.

What is your share, do you think in optical trans U.S. today? And what are your aspirations?

Our share is less than 10%. Yes. And I think that there's no reason why we couldn't be one of the top 3 transceiver manufacturers in the world. And I think we're certainly positioned to be able to do that, as well as with the geopolitical challenges of U.S. and China, I think as we talk to our customers, the faster we can get out of the China manufacturing and really establish a core infrastructure in Thailand faster that we'll get the business. And that's why we're moving so rapidly on establishing our Phase I, which is well underway. And now we kicked off our Phase II, which is a new, very large 3-story building in Thailand to be able to really kind of implement incremental capacity over time.

Got it. One thing that we have seen is just that the size of these clusters continues to grow, right? What used to be 10,000 is now 20,000. We had NVIDIA present earlier and they spoke about 100,000 cluster. What does that mean for attach rate of optical transceivers, right? I would imagine that there's an exponential growth factor there.

Yes, it is. And it's -- that's what leads me to my concern about the indium phosphide capability in the world to be able to interconnect all these. Now a lot of those are so copper interconnects. So it's very, very short reach stuff, but it still doesn't diminish the need for an exponential growth rate of optical interconnects on these huge clusters and then connecting the clusters in the data centers. And so that drives ER capacity. And so it's a very exciting time, and I think one that we're positioned very well for.

Got it. Is any of your telecom output fungible to become datacom output? Like can you make that conversion over?

I would say, yes, and we're already looking at that. And we have really 3 main wafer fabs. And so one of the things we're looking at is how do we maybe share capacity on what's traditionally been our telecom wafer fab to support the datacom growth because that's really very rapid. And so maybe doing part of the wafer fabrication in our traditionally telecom fab and then transferring it to finish in our datacom fab. So we're looking at that. On the back-end side, I think with the telecom slowdown and the inventory consumption, that's actually been an opportunity for us to use these clean rooms that we had built for telecom to use for datacom. And so that's gotten us the rapid ability to make prototypes in Thailand more rapidly than we would have otherwise.

Got it. When do you see your Thailand facility become tangible to your results? Is that again something that happens next year? Or you see it happening this year?

Well, I think tangible to our results are getting the parts qualified. That's very tangible because it's a real leading indicator. And so getting first prototypes out this summer is a tangible benefit. I think in contribution from a revenue standpoint. I think that's really more late in December quarter, but more meaningfully in the first half of next calendar year or the second half of our fiscal year.

Got it. Makes sense. Is Cloud Light passed the product transition issue that it had at the large customer when you acquired the company?

Yes. I mean, I think with any new product introduction, there can be challenges and inventory changes and such. But I think for the most part, that's behind us. Now we're focused on getting the qualifications done and the capacity in place to ramp these new products, of which some are already at 200 gig per lane. So it's really an exciting time and one that will ramp through the fiscal year quite steeply.

Got it. What helps, Alan, to make optical transceivers more accretive right to the business? Because you mentioned that there are two large suppliers right now. You want to be a third supplier. So how do you make sure that it doesn't end up becoming just a pricing game because there's going to be a lot of volume required. So are the products from the 3 companies that different? How do you make sure that you are able to make this business accretive over time?

Yes. I'd say really 2 things. One of which is how do you build the product. And one thing that really attracted us to Cloud Light was their capability to design and build their own equipment to align and manufacture transceivers. And so having that capability of low-cost equipment high-precision capability, bringing in wafers on one end of the factory and transceivers coming out of the other end of the factory was very attractive to us. And so you combine that with Lumentum having the core technology at the chip level. That's very different than most of the Chinese module manufacturers who rely on people like Lumentum to sell them at market price or photodetectors or other components, which they're very good at. So don't get me wrong there, they're very, very good at that. But I think from that standpoint, vertical integration best manufacturing capability at lowest cost CapEx is a good combination for having lowest cost capability. Now we've got to get the volumes up so we can get that scale. And I think that's what we're focused on in the next 12 months.

I see. And how is the visibility? So for example, if, let's say, whether you have NVIDIA like ramping with Blackwell or you have EMD ramping with their MI products. How early do you know of how your share will progress in a given year? Do you know that like 6 months ahead, 9 months ahead, how far is your visibility into that share gain opportunity for next year?

I think it's -- it really is dependent upon getting product into their hands and feedback that it works. And then it's typically from the time we get qualification samples, and I'm not saying a specific customer into any of our customers to the time where we have to launch is typically about 6 months. And that's enough time to get our capital in place and the long lead time capital that we're placing orders at risk today because we're confident in our capabilities. But I'd say it's kind of that 6-month period of time. And given the investment that we're having to make on a capital front, we're having those longer-term discussions with customers around we're going to buy all the capital because you want us to, and we want to as well. But can you guarantee we're going to use that for some period of time to get that return on investment bank? And so those discussions will happen soon as well.

Do you think the customers are open to making kind of allocation and share decisions early or next year?

I think so. We haven't tested that water deeply yet. But given the feedback we've got and the desire to have non-China manufacturing and the fact that I think everybody is starting to realize that there's going to be an indium phosphide shortage in the next several years, puts us in a good position to partner with customers and have a win-win situation on both sides.

Makes sense. On the telco side, we have seen, not just for Lumentum, but across the industry, across your peers, this kind of consistent drip down in telco. How is the visibility right now? Like do you think that the industry or Lumentum is in a better position to predict when you are done with that inventory burn, which I know is always -- is always tricky.

When this really started, which was over a year ago, I would have thought it would be done by now, but I was wrong. I'm hesitant to predict when everything gets fixed. But I will say that bandwidth demand is really not slowing. And so almost regardless of when the inventory gets taken care of, whether it's next quarter or by the end of the calendar year, which is kind of in that time frame from my perspective, we'll start shipping into our customers, what they're shipping out and deploying. And I think we'll get back to an undepressed level of telecom shipments today, it's terrible, right? It's very, very low compared to where we were a year or 2 years ago. Two years ago, we were probably over-shipping consumption. And I think that this last 4 or 5 quarters has taken care of a lot of it, but it's still not all gone.

It's interesting to compare what we are seeing in the telco end market versus what we see in the industrial end market. There, a lot of your industrial peers are saying, well, we could have better than seasonal quarters because it's not like the industrial economy that we were under shipping demand. Do you think that whenever telco bottoms that are you undershipping to an extent that they could actually be hockey stick with a company or that is not that easy to say right now?

I think it's typical, right? And when we've seen a slowdown in the past, it's like the pendulum shifts the other direction and then all of a sudden, you get that, what do you mean you don't have capacity and people to build stuff that I need? And so I'm half kidding by that, but I've seen it too many times, and I would expect that's probably going to come -- I don't know when it's going to come, but I -- we're already seeing that on new products. So products where there is an inventory in the channel, new higher-speed coherent components, new integrated C+ L-band ROADMs where the bandwidth for a given fiber is reaching channel limit. So you need to be able to do more wavelengths over a fiber. And so those new products were a year ago, we weren't shipping them. The demand is quite strong for those. So those kinds of things are good indicators that bandwidth demand is still strong. We just got to get rid of some of the older inventory.

And last question on telco. Is there a specific recovery catalyst? Like in the past, we could have said, oh, there is 5G or there is some broadband deployment. Is there some specific catalysts that you think you're excited about that can help drive recovery there?

Yes. Well, I'd say AI, right? I mean everything is AI and data centers that are having to be built for their part means we're seeing strong demand for submarine amplifiers. And so we sell pumps to people laying cables at the bottom of the ocean, that is very strong right now. We used to call that telecom, but it really is datacom, right? It's satisfying datacom. And so those kinds of things that ZR, 800-gig ZR, those kinds of things are going to be very strong catalysts for what we traditionally had called telecom it really is driven by datacom. I'd say on -- and on traditional telecom, I don't know yet. And I'd say that the cable MSO market is also slow. And so I think we'll see what drives that. But I think, again, it's telecom bandwidth is not slowing. We should see a recovery back to the consumption level when that inventory has gone and I think in calendar '25, that's for sure going to happen.

Got it. I said last one, actually, I have one more on telco, which is kind of the non-U.S. demand. China in the past had been a big buyer of optical components for their wired rollouts. Where is China demand standing right now?

It's actually quite strong. Except we're restricted from who we can sell to. So we used to have a large customer in China that was a teens customer of ours that is no longer a teens customer of ours at all. And so we are selling to everyone else. And that demand is quite good and high-speed coherent components coherent modules, integrated CPL ROADMs are being deployed by the 3 major carriers in China. And so that's a good growth catalyst for us. But again, we're not participating with the largest Chinese network equipment manufacturer today.

Understood. On NeoPhotonics, what did NeoPhotonics add in terms of value to Lumentum so far? And then in terms of the cost synergies, where are you in that? I think you have achieved about $70 million, right? So when do you plan to achieve the remaining $30 million or so?

Yes, I see that NeoPhotonics came with the absolute best narrow line with tunable laser in the world that has a very large share of both the ZR market as well as the embedded market. So for the longer haul, submarine, all of those regional and metro, it's by far the best, came with the wafer fab in Japan that made CW lasers. And so those are going into the silicon photonics-based transceivers and into, again, our product in feed into the cloud light transceivers. So that's been very successful. ZR got us into the ZR market in a meaningful way as well. So you're right. We've been -- we're about $70 million into our $100 million synergy target. We've closed a couple of China factories. We built up a bunch of inventory to be able to solution the transition to our new factories. And as we bring up the new factories or production in Thailand, that will help us with that synergy as well as the fab, we're consolidating the fabs in Japan, and that will get rid of some fixed costs. And so I'd say that by the second half of our fiscal year, we should see that $100 million of synergies from that deal.

Got it. Makes sense. On gross margins, right now, low to mid-30s, right, or so. What is the plan to get it back to the 40%? What is the role of mix versus the role of just fab utilization, right? What are the different building blocks of getting gross margin done?

Yes, I'd say that we used to be higher than 40%. And that was when we have had a very heavy mix of chip level revenue. We're starting to see the chip level revenue coming back, but EMLs and CW lasers. And I think that could be a significant contributor to gross margin improvements. What we said at the OFC -- our Investor Day at OFC was getting revenue back to $425 million on average for a quarter will get us into the teens. Operating margin and we're longer term getting to, say, $600 million a quarter could get us to that 40%-20% gross and operating margin target. And I think we're -- as we said on our last earnings call, we should exit calendar '25 at $500 million. So we'll be in the teens if that happens or when that happens. And then growth from there will get us back to the model that we'd like to get to, which is 40% gross margin and 20% operating margin.

Got it. One question I had about your EML output. Do you plan to keep it mostly for your own transceivers? Or do you think that you will have it available on a merchant basis also?

Most of our EMLs go to the merchant market today. And so as we look at 1.6 terabit, we're making both silicon photonics perceive as well as EML transceivers and customers will choose and costs will dictate what is the best solution. So we're continuing to sell to the merchant market and that's a big part of our datacom business today, given the chips that we sell today and kind of the record output we had last quarter. We're going to continue to grow that output. But as I look forward and as I talk to my customers, I see that, again, that's going to be a challenge for us. And so we're trying to go to bigger wafers and add capacity, but I think we're still going to have some challenges keeping up with the overall market demand, both external merchant market as well as internal. So we have some very, very strong partners that we have to take care of and make sure they continue get the EMLs they need. And I think we can inform some long-term partnerships with them to make sure that the capacity we've put in place that does get consumed in that they have the capacity they need to satisfy their customers.

All right. You mentioned, Alan, exiting next year at the $500 million run rate. What are the underlying assumptions about the datacom versus telecom mix in that and industrial tech, of course?

Yes, I'd say industrial tech, it's going to grow slowly and 3D sensing, we're not counting on any growth there. So I think as lasers recover and we get into new applications and EV, display and solar applications, our ultra best business should grow rapidly. But that's a small part of going from where we are today at just over $300 million to $500 million. I'd say that chip business, where we should grow that, both from a ASP standpoint at 200 gig per lane those ASPs are higher, but we're also going to increase output. So that should have a considerable step-up. And then telecom, just getting back to selling into what they're selling out, that should have a meaningful impact. And then you add all that together and you don't need much more datacom transceivers. So we land one new datacom transceiver company, our customer, that's well within our reach of getting to $500 million. And if we went to the whole different story.

Who makes the decision on whose datacom transceiver will be used? Is it something that's specified in a reference architecture by these companies who are supplying accelerator or is it done by the end cloud customer?

It depends, but I'd say that most of them are done individually by the cloud customers. Now there's a bunch of cloud customers that are buying the entire cluster with everything that they could get. But -- so those are determined by what we call the AI infrastructure companies, but there's also a large cloud business. So it's non-AI, it's just now the .400 gig and suit to deploy 800 gig. That's a big market and one that had slowed last year, but I think it's starting to really take off and we'll consume a lot of our transceiver capacity at 400 gig and 800 gig. And those are new designs for those customers as well.

Got it. And then finally, Alan, M&A has been a key part of your strategy. The company still has a very strong balance sheet. So how do you look at -- does this industry downturn kind of open up opportunities for you to look at? Like do you think that STNG focused on just the comms market, it by the datacom or telecom? Is that the right strategy for Lumentum. Because your competitors have kind of branched out into many things, right? So do you plan to stay focused? Or do you plan to be more diversified? What is the strategic way of thinking about that?

Yes. I mean from an M&A standpoint, I'd say -- or from a capital allocation standpoint, I'd say we're focused on making what we have successful. And that means investing in Thailand, investing in infrastructure, investing capital, investing in R&D so that we can get the return on the investments we've made to date. And that's primary -- our primary focus. That said, that doesn't -- that doesn't eliminate our ability to look at things. But I'd say that our primary focus today is to make the transceiver business very successful, grow our e-mail and CW laser capacity and then get our margins up to where they need to be before we take on anything of meaningful size.

Got it. So stay focused. And then finally, I think you did mention that the goal is to get to the mid-teens EBIT by the time you get to the $500 million. So just final question on operating expenses. How are you managing expenses as the company comes out of this downturn?

Yes. I think we -- we're trying to stay around $100 million of quarterly operating expenses, and that will grow slightly as we grow our top line because we need to continue to invest in R&D. So SG&A will grow slower and revenue will grow faster than our overall OpEx. That said, I'd say as we get to, say, a $600 million quarter revenue, OpEx should stay at that 20%, so $120 million. So as we go from where we are today to $600 million, I think we're going to have some growth, mostly in R&D.

Interesting. Great. With that, Alan, thank you so much for your time. Really appreciate it.

Great. Thanks for having us.