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

To have the team from Lumentum join us. Michael Hurlston, CEO. And I'll go through the same fireside format, my questions, but please feel free to raise your hand if you would like to bring up anything. Michael, warm welcome. Really appreciate you joining us. And thank you for the positive pre-announcement yesterday.

Thank you.

Yes, always good to get something exciting before our fireside.

We just wanted to keep your blood pressure high. That's all...

That, I think there's a lot going on, too. But maybe talk us through what led to that upgrade during the quarter? Where are you seeing the upside drivers? And the way I understand it, I think you raised guidance for June, right? Earnings guidance even stronger than that. And you expect to get to $500 million a quarter earlier than you thought before. But the $600 million target you kind of kept as was before. So maybe walk us through these dynamics.

Yes. I mean I think first in the quarter, we're seeing just broad-based strength across the business. It's not driven by any one thing, but it really is something that we can't ignore. So we are obviously -- we feel like our new guidance is at the very high end of our revenue -- our old revenue guidance at the midpoint. That falls through on both the operating income line and EPS, right? So we feel like we needed to say something and probably more importantly, how that starts to cascade, as you correctly called out, the $500 million quarter comes in to our September quarter. And then the $600 million quarter, which has gotten a lot of discussion in our one-on-ones today, that's mostly because we just don't have the visibility. We feel like we've got a ton of growth drivers that you and others have written about that really set us up super well. As we think about 2026, we have our module business that I think you understand better than anybody that now is starting to really take effect. In the first half of 2026, we have OCS that we feel good about that a lot of people, I think, underappreciate our opportunity and our positioning there. And then in the back half of 2026, we have co-packaged optics. So we sort of have this set of dominoes that we think line up almost on a 6-month cadence. And again, in the short term, it really became difficult for us to ignore. We just said, "Hey, we've got to go out and say something because this business is performing a lot better than we expected. I think for the $600 million quarter, look, we just don't know. Tariffs, there's a lot of uncertainty. If the business plays out like we expect, we probably will do better than the expectation that we've set out, but just too early to call.

Got it. Very good. So Michael, 4 months into this new role as CEO, but you have been in the industry before, right, Synaptics, Finisar before that and Broadcom, right, before that. What have been your first impressions like? Because this optical industry has been an interesting one over time, right? There's -- in Moore's Law, we have like doubling of benefits every 1 to 2 years. In optics, they get 10x the benefit, but then the industry has had trouble maintaining profitability over long periods of time. So talk us through what your first impressions have been.

Yes. Look, I think Lumentum is an incredibly technically driven company. It really is super, right? The technology drivers that we have inside the company are second to none. We have an end-to-end portfolio that I think gives us a tremendous amount of confidence. We can play in submarine networks. We can play in data center interconnect. We can obviously play inside the data center. So we do have a really, really complete portfolio. I think the one change -- significant change that's been made in the first 4 months is let's do addition by subtraction. I think that there was so much opportunity coming at the company that we -- as my mother said, your eyes are bigger than your stomach. We were really taking on a lot, and that was leading to some execution challenges. And we became our own worst enemy in a sense. And so what I've done in partnership with Wajid here is really streamline the number of projects, really streamline the set of activities that we're trying to take on. And that has led to a lot better execution, which in turn has led to today, the positive announcement. So I think we took some things off the table. We had an announcement we talked about in our last earnings call that we -- one of our big kind of initiatives that we talked about was a metrology business. We stopped that. We redeployed resources to OCS. Inside the module business, we took the number of projects that we wanted to focus on. We took that down by about 50% to give us the best chance of success, and we're starting to see the fruits. Look, it's early, but we really feel like simplifying the company has been -- is the path forward. I think as we play this thing out over the next few years, I have a system. We run the system. I learned the system at Broadcom, employed the system at Finisar. Unfortunately, we sold the company in a short period of time, really put that system in place in Synaptics. And Wajid and I are now deploying that system again here, which gives us great transparency across the businesses and will allow us to control OpEx, allow us to really focus on the gross margin line, which we want to improve greatly while we have this big revenue tailwind, right? So we think that, that will start to kick in, in the second half of this year and in 2026 and beyond. And I think that system will really, really serve us well.

Absolutely. And one interesting aspect, Michael, is that a few years ago, I remember Lumentum had gross margins in the 50s, right, like very kind of pure-play best-of-breed component. And then I think there were a bunch of acquisitions. And like you said, the aperture got expanded and margins got into the 30s. Where is Lumentum now strategically? Do you see a realistic path, obviously, not year, 2 years, but longer term to kind of get back to its roots of focusing on kind of specialized areas and getting margins up?

Yes. I mean I think that, that point in time was driven largely by Apple, right? So we had the Face ID product, and I competed against that at Finisar, right? And we saw really great execution from Lumentum to capture that opportunity and really be sole sourced at Apple for the better part of 2.5 years. That led to margins on that piece of the business at 70% plus. And so that carried the company up over 50%, as you're correctly calling out. But frankly, that was a blip. And what -- I mean, I think the previous administration did so many things right. But one thing that wasn't done was use that tailwind to clean up parts of the portfolio. And we have a big tailwind right now. We really do in terms of our demand and our growth drivers. We are not going to let that opportunity pass us by. We're going to really focus on operationalizing underperforming parts of the business, focus on getting our gross margins up. To your question, I think we definitely have line of sight to margins in the 40s. So we're not happy with margins in the mid-30s. We're inching those up. I think if you look at the guide, we get up into the higher 30s. We're not happy there, right? So I think with all of the different levers that we have, pricing levers to a certain extent, big cost levers, we think we can get it up into the low 40s. And look, we'd like to do better than that. I mean I think I'm not happy. I'd like a return to 50s, given our portfolio might be difficult. Wajid is shooting arrows at me as I say that. But we think that we have a lot of levers. And I think that's the dynamic, too, that's changed, Vivek. In the optical industry, you asked before we began the chat, coming into this industry, I see a lot of parallels to what we saw in the semiconductor industry in 2016, 2017, where the paradigm changed a bit, and we had a little more balance between customer and supplier. The optics industry, I think, has been used to being dictated to by the customers, a handful of customers that are relatively slow moving. And the customers have changed. They're much faster moving these days. And I think we can right that balance and get some better pricing and command a little more respect because what we're delivering is amazingly complex. It's no different than the semiconductor landscape. We should be able to change this margin paradigm, I think, relatively quickly.

Right. Now the one place where a skeptic might push back is that unlike the semiconductor industry, your biggest competitor, right, if we just pick modules and optical transceivers, right, because that seems to be one of the biggest categories that you are participating in right now, that the biggest competitor is in China, right, that they have access to a cost structure, different margin expectations, et cetera, and incumbency in a lot of places. Then you have another really large competitor, right, in the U.S. as well. They had an Analyst Day recently, and they said, when it comes to optical transceivers, we are optimizing for share, right, not maximizing margins. So how do you expand margins given that competitive landscape?

Yes. I mean, look, as we said, one, the strange and painful fact is that between us and Coherent, we do exactly 0 business in China, 0, right? The Chinese hyperscalers have locked us out. They do not want U.S. content in their data centers. We believe that the same is true in the U.S. that having a domestic supplier makes a difference. I think we both command a slight premium, slight over the Chinese suppliers. And if we could supply, they would bias the deck very much toward us and Coherent, right? Right now, to be fair to them, we simply can't supply what they want. But I do think that in the asymptote, I think some of that threat comes down. The second point I'd make is the business is so big, we can't pick and choose, right? We don't have to go after everything because we have a lot of other things going on. I think perhaps similar to Coherent, they have a super broad-based business, we have a very broad-based business, too. So I don't need to chase every single module design, particularly short reach, lower price, more competitive areas, I can be a little bit more selective. So the model that I think Wajid and I have worked out together is, look, we want to grow this business. So we think there's a path to get this portion of the business up over $1 billion. But I don't want it to be a $3 billion and $4 billion business. I don't, because I think it's always going to be a margin headwind. So we'll manage it. We'll manage the opportunities that we pursue, and we'll keep our eye on that margin line. And then again, the rest of the portfolio, the 2/3, 75% of the rest of it, we're going to be very, very focused on keeping the margins up.

Okay. This notion that U.S. customers, the hyperscalers, right, or other customers such as NVIDIA and others, that they should buy more from U.S. suppliers. I think that sounds good in theory, but are you actually seeing that in practice? Like are they resonating with this idea that they should be shifting, right, more of the purchases to your competitor and yourself as opposed to a Chinese supplier?

Definitely, the intent is there, right? And I would say it's hard to test the intent because we simply can't supply. So if Coherent and us could make 100% of the demand, I think we have a real test of intent. They are saying all the right things, and I do believe we get a premium, right? And I believe Coherent gets a premium. So the intent seems to be matched with action. But the reality is that we're a single-digit, low single-digit player. Coherent may be 20% of the overall market. That leaves 70-plus percent for the Chinese. And you just -- we're just not able to fulfill the demand, right? And so I think it's hard to say at this point whether that's true or not. We are ramping supply. I know Jim is ramping supply. We're going to get to a point, I think, where we'll be able to better test that.

Okay. We'll come to the constraint side. The one other interesting thing that when we attended your analyst event at OFC was optical circuit switching. And that surprised me because somehow I was only -- I was under the impression that only Google had that as part of their, I think, spine network, and it wasn't really as widely adopted across other hyperscalers. So talk to us about what is that category? How large is it today? Where can it get to? And what is the key role that Lumentum is playing in that?

Yes. I mean I think you have it right. If you look at this particular point in time, it's just Google. But I think what Google has proven with their network architecture is the fact that there's significant power savings by going to an optical circuit switch, and there's very significant cost savings. So what you see is the rest of the ecosystem blaming on to that. And any new deployment is most definitely considering OCS in the spine, right? So it's a significant shift in TAM from electrical to optical. We would calculate that out somewhere between 10% and 15% of the TAM now moving gradually to optical, very much concentrated on greenfield, right? Any new deployment, it makes a ton of sense for them to go to an OCS for the 2 reasons that I mentioned. And then I think where OCS is deployed in places like Google, I think there's a replacement because our solution is better, right? We really think that we have technical advantages in terms of insertion loss. We have technical advantages in terms of cost, quite frankly. So those things together say, okay, we can compete in existing landscapes. And most definitely, we can compete in greenfield.

Right. So just to understand, this is essentially a replacement for what would otherwise be a Broadcom-based Ethernet switch, right? And you think OCS has today the reliability and the cost structure to be competitive against a Broadcom class switch in a spine situation.

Yes, you have it right, Vivek. Yes, that's right. It's spine. You can argue, and we've seen this now some even discussed in the lead, right? So it definitely will put pressure on the Broadcoms of the world. It's early innings, right? It's still early. And really, I think it's very difficult for us to say that we'd go into existing deployments where they're able to go into every single day. But I think as you think about new deployments, this has such an advantage, right? Kathy was talking about earlier. Just think about the switches from electrical to optical, that consumes power, right? And if you can keep it in the optical domain as much as possible, that's a huge benefit. That alone is reason enough to deploy OCS.

Right. And how do you differentiate your approach from Coherent's approach? Because I think they also described their OCS offering recently.

Yes. They -- I mean, there are 2 different approaches. But in the end, I don't think that that's an issue. I mean what is an issue is we have a significant time-to-market advantage on high-radix, high port count. So you're thinking about the current deployments mostly are 300x300, 320x320, moving up, 500, 600 ports. The more ports you have, the more effective this becomes. So the vector is more, right? If you look at what Coherent offers, it's a lower port count. It's a 64x 64. They will have opportunity with that, no question. It's not a spine replacement. There's simply not as many ports. It's more of a specialized DCI type of application. We can compete for that, and we will, but we don't have a solution. We're behind them. They're significantly behind us on a higher port count. The MEMS-based -- on a technical basis, the MEMS-based solution has a distinct advantage in terms of insertion loss because we're using a mirror, they're using a translucent liquid crystal. So there's loss. There's necessary loss through that system. In a higher port count, that insertion loss makes a big difference. So there is a technical advantage we have. But at the end of the day, I'd say the big, big driver is the fact that we are first to market.

Got it. So essentially, the way to think about it is that this is an adder for first half of calendar '26 that you have kind of line of sight to wins and this becoming a decent part of the business at that point?

Yes, that's right. I mean we're being asked to deploy 50, 100, 200 units a week at ASPs that we're talking about, which are nearing $100,000 each. This is a big deal. It's a big deal.

And is it one customer? Is it multiple?

Multiple.

Multiple customers. Okay. And then the next thing you suggested was the move towards co-packaged optics, where, again, there's been a lot of industry debate as to when we will see it, if we will see it. What I'm still surprised by is that I think the vendor community, whether it's yourselves, NVIDIA has mentioned it, Coherent, right, Broadcom has been like so I have yet to hear a hyperscaler actually say that I need a co-packaged optics, right? So do you see enough pull by these customers? Because they, I imagine, weighed against the operational benefits of just staying with the current pluggable. So do you -- have you heard from those customers whether they need it? Or is this a technology being pushed by the industry?

No. I mean I think the hyperscalers are most certainly talking about it. But I think where you're going to see the volume is when people deploy the turnkey NVIDIA rack, and that's going to be a significant number of people. So all the hyperscalers ultimately will have a mix of NVIDIA racks in their data centers. And to the extent that, that exists, NVIDIA seems very committed to CPO. We've seen numbers from them that are very significant. And that alone is creating they're pushing, right, into the hyperscalers, but you'll see all of the big -- certainly, the big 4 U.S. will deploy NVIDIA turnkey to some extent. They will have their own. They will have their own implementation even on an NVIDIA platform, some of them do their own configuration of the NVIDIA platform. But to the extent that this turnkey exists, it will land in every single hyperscaler. And then the hyperscalers separately are talking to us about, hey, this seems to make sense, right? No DSP, power savings, the result in power savings from that. That is something that I think is super attractive. So the conversations now on a separate implementation are also happening in the various hyperscalers.

I see. And I would imagine this is incremental, right? This is more scale up rather than scale out? Or where do you see the application?

It's scale out. So to be clear, it is scale out, right? It's attached to the switch. It's more scale out, touching on a different concept. Look, scale up is a huge. I mean this is -- we measure transceivers in hundreds of thousands. We're measuring scale out in millions, right? Scale up is hundreds of millions, to be honest with you. We're all looking at that opportunity. We think we have great technical positioning for scale up, but that's a ways out. This is all scale out.

All scale out. Okay. And any simple way to say that if you had a chance to sell an optical transceiver versus a CPU, which I guess, because you're not that big in the transceiver market, it's not like you're cannibalizing your business regardless, right?

Yes. I mean -- and that's the calculus for us. I mean we have -- right now, right, we have exclusive position on CPO. By virtue of the fact, which you appreciate probably more than most, it comes from our ram and pump heritage, right? We have an extremely high market share, bordering on triple digit of undersea pump lasers that have to have the reliability and power that -- I mean, if I plop something down in the middle of the ocean, I don't want that to fail, right? And so we figured out how to get fits of basically 0. And so the ram and pump laser is -- the UHP for co-packaged optics is a derivative of that. And so we believe we have a multiyear advantage by virtue of the heritage on Ram and pump lasers that is going to give us a moat. And so from a market share perspective, we're talking about low single digits on transceivers at multiple hundred dollars, right, versus a co-packaged optics opportunity that's in the tens of dollars. I mean the ASP is actually surprisingly high at very high market share. So the math works out probably on the top line, roughly even. But if you look at profit dollars, it's super good for us.

Right. Makes sense. On the transceiver side, where are you in terms of executing on the Cloud Light, right, transaction that you did? Do you think you are kind of past the product transition issue they had at their customer. You mentioned low single-digit kind of market share. When do you see that becoming double digit or more?

Yes. I mean back to a couple of points that I made. We definitely had some execution problems on our module business, and we tried to take things off the plate in order to improve engineering execution, which we're seeing fall through. Again, early innings, but we're actually seeing improvement in terms of the execution. And we probably never want to see this business unless we get to a state where we talked about earlier that we can supply the world supply of transceivers, and we see a balance in the gross margin equation, we probably never want to run this business up too high, right? Again, we want to be selective about the opportunities that we pull on. So we have a lot of margin room here, unfortunately, right? We're operating this well below InnoLight, well below Eopto that you can see. Frankly, we're operating well below Coherent. We think we can get the business into the mid-30s gross margin. We have a lot of room right now to get there. And I think we have levers because, again, you've written about this. We don't use our own internal components today. So we're planning to shift to our own internal components. And as we do that, margins get better. Just the utilization of the factory, margins get better. We are learning. We're producing in volumes that we never have before in our Thailand factory. Again, that learning margins get better. So we definitely have a track to get the margins up into the 30s next year.

Got it. The telecom business, where is that in kind of its stage of cyclical recovery?

Yes. I mean, look, we've come down from historic highs, right? I mean I think if you look at the math, we were probably doing mid-300s on the telecom business. And now it's $200 million, $230 million, $240 million. So it's come down significantly. Part of that is Huawei, right? So we had a big bite that will never return. But if you look at the current base, it's actually growing surprisingly well, right, driven by, again, the hyperscalers, right? So our data center interconnect business, our pump laser business, is not going to the operators anymore. It's going to hyperscalers. So that business has done surprisingly well. It's shown growth. We're definitely on an upward trajectory. We've come off the bottom. Will we ever get it back into the $350 million range? Probably not, just given the delta that Huawei represents to the overall business, but we certainly will get it up $50 million, $60 million, $70 million more incrementally over the next couple of quarters.

Got it. And then finally, we discussed the path towards higher gross margins. Talk to us about the operating leverage in the model. How do you simultaneously kind of stay competitive working in all these different dimensions, but still get operating margins right into the 20s at some point?

Yes. I mean one of the issues we have is, I think, on our engineering leverage. We definitely, again, are probably doing too many things. And so by cleaning the plate, I expect to actually operate the business much more efficiently while driving top line. I mean we just -- we were taking on a lot of projects that definitely didn't have ROI. So we're trying to institute a lot more discipline. And in so doing, I think you end up having a lot more leverage. The other thing is and Wajid just talked about this in a couple of meetings earlier today, I think our -- kind of our G&A bloat, I would call it around the company, can be streamlined. I think there's cost opportunity there. We're going to start to work on that. We enter our annual planning process actually tomorrow. And we'll see, can we actually take some of the cost out of the business and trade it for R&D dollars that certainly are more productive from an ROI perspective. So we do think that we can operate the business, plus or minus in the same OpEx envelope while ramping revenue pretty significantly.

Got it. And then maybe one final question on the EML lasers. Lumentum is a well-recognized leader in that business. Where are you right now from a supply-demand perspective? At what point do you think you will have enough capacity to kind of help build your own transceiver business? And then let me ask that final part C of this, which is Coherent is also investing a lot in that business, right, move to 6-inch, they think that gives them a cost advantage. So where are you supply-demand, your transceiver -- feeding your transceiver business and then this competition?

Yes, we're behind. I mean the headline is we've added -- we've doubled the capacity. I think that figure was given out by Wajid and Kathy a couple of times. We doubled our capacity over the last year, still way behind. We have a path to probably increase the capacity another 40%, 50% over the next 4, 5 months -- 4 or 5 quarters. So we still have plenty of headroom on the capacity. I still think we're going to be chasing demand. I mean the EML demand, we were just reviewing e-mails in between breaks, right, just the number of opportunities that are coming in from other transceiver manufacturers are significant. It's really the volume seems to be exploding. Our transceivers are all CW. So interestingly enough, we've allocated all of our capacity to serving external customers and all of our capacity on EMLs. So one of the decisions that we talked about in the last earnings call was allocating a bit of the capacity to CW laser, getting our toe in the water. As we grow overall capacity, we'll continue to increment EML, but we'll probably grow CW a bit faster. One, there's an immense CW opportunity. That's also very underfunded. We think we can do things on pricing to make that equation really work at our advantage. And then two, most importantly, is serve our own modules, right? So part of our margin challenge is that we're using external components. We'd love to get that situation rectified, use our own components in our own modules. And I think that's a 5 or 6 percentage bump in margin that we didn't fully appreciate. So that's something that we're intently focused on, and we'll execute, I think, in the first half of next year.

Great. With that, Michael, thank you so much.

Thank you. Pleasure to be here.

Super helpful.