All right, great. Good afternoon, everyone. Thanks for spending time here at UBS's Tech Conference. I am David Vogt, the Hardware Networking Analyst, and we're excited to have with us Michael Hurlston from Lumentum, President and Chief Executive Officer. We're going to dig into the business, but I thought maybe since you've been in the seat for 18 months, is that about right? 15 months?
Less than a year, David.
Less than a year. Yeah, but you know, maybe just, okay, so we'll go into each of the segments that I think investors are focused on in a second, but from your seat, from an outsider coming in, clearly less than a year, maybe just level set kind of how you were thinking about the business when you joined to where we are today. I think a lot has happened in this less than a year period, and we'll dig into each of those products and categories separately.
Yeah, look, it was. CEOs very rarely change jobs. You either get fired or you retire, right? Yes. And I have now been CEO of three public companies, one of which I was retired from, but then I made a switch. And I remember the board of directors sort of presenting to me their view of the forecast as they were trying to recruit me to come into the company. And I'm like, this is crazy. I mean, there's no way the forecast can look like this. Why would you possibly be wanting to make a CEO switch in the face of this forecast? And the reality has ended up much different. The forecast that they gave was understated probably by a factor of two, which was already incredible to begin with. And it's been an incredible journey.
I mean, 10 months, I think the day I joined, the stock was like $70. And obviously, we've seen quite a bit of appreciation in the share price over that period of time. But we've had to do a lot of things, a lot of work on the company. I mean, I think, the previous CEO did a great job, obviously setting up the roadmap, but the board wanted to see the ability to scale because the telecom industry, which is really where optics has been, was built on scales of thousands, right? And they could see that it was going to be semiconductor-like scales that were coming. And they wanted to get somebody who had seen a semiconductor cycle before and could scale thousands into millions and tens of millions and maybe even hundreds of millions, right? So this optics industry has never seen anything like this.
And I hit it at exactly the right time. I watched my good friend Jim Anderson, who was also a semiconductor guy, make the jump to Coherent, and he's done obviously a hell of a job over there too. So you've got two semiconductor guys now in optical companies.
Great, great time to be alive.
Yeah, not too bad.
So when we think about the forecast that you talked about being really bullish and now obviously magnitude stronger, let's talk about underpinnings for that. Let's start with sort of, I think, not to de-emphasize this business, but let's talk about the transceiver business first. So obviously, that business has seen tremendous growth. The market has grown leaps and bounds faster than I think anyone had expected. You're back in the, you're in this business because of an acquisition made back in 2023. So can we talk about kind of how you're thinking about that business long term? I know you've been pretty vocal about how you see that from a customer perspective, a volume perspective, an IP perspective. So as we sit here today, given just the demand trends that you're seeing, like maybe just refresh us on like, how do you think that plays out?
I mean, is there really a cap on that business in terms of how you're thinking about the revenue opportunity? Self-imposed cap, I should say, not a demand cap.
Yeah, David, look, I mean, we, of all the businesses that we have in the portfolio, this has been the most challenging for us. We haven't executed, I would say, particularly well in the transceiver business. We are a minority player. We are behind several of the big Chinese names in terms of being able to supply transceivers and certainly behind Coherent. So we're trailing the play. And what we've said is, given the margin profile of that business, which even in the best case, I think is sort of a mid-30s, maybe high-30s type business, we want to get our overall corporate margins. We've given a long-term target of 42%. We think we have obviously room to expand above that. But that transceiver business will always be a bit of a headwind. And so we've communicated that today that business is running about $500 million annually.
We've said, look, we have aspirations to get it to $1 billion annually to add another $500 million of incremental revenue, but we don't want it to run much higher than that, just given the margin headwinds we see. We think we can manage our business up from a margin perspective if we keep the business to about $1 billion top line. If it gets beyond that, it'll be more challenging. Now, that being said, and you sort of alluded to it in the question, we've certainly seen pressure from our largest customer and other customers to ship more, right? Because the demand right now is very, very high. And so our challenge is going to be to how to improve the margin of the business. We really need to do that.
And then how to kind of manage it probably in the face of a little bit higher than $1 billion of top line revenue over the capture period.
And that, I think you've said publicly, that's really targeted with three large existing customers that you've had a long-standing relationship with since you've gotten back into this business. And to get to that margin target, I know it's probably below that 35% today. What needs to happen? Is it vertical integration? Is it outsourcing? Like, how are you thinking about driving that margin? I know it's the least exciting piece of your business today in your view, but just maybe let's get this one out of the way.
Yeah, no, no, I appreciate it. Look, it is, it's a good business for us. I mean, there are, I don't want to minimize it. I just think that other parts of our business have incredible growth opportunities with better overall margin profiles. Look, you guessed it right. I mean, we're operating meaningfully below the mid-30s in terms of margin. We have a couple of levers. First, we are not manufacturing at any kind of scale. Getting the business to $1 billion allows us to amortize units over our manufacturing footprint. That helps the cost. We have to improve our manufacturing. Our manufacturing is substandard right now. We brought in a new person that comes from Jabil to run our factories, and he really knows what good looks like. And I think he can improve our automation levels. He can improve our throughput levels.
He can improve our scrap and yield, all of which are contributing mightily to our problems. So I like how he's thinking about that business. And then you and I have talked about it. We also don't insource. So none of our transceivers use our own components. That can be a margin benefit by taking our lasers in particular and bringing them into our transceivers. And so we intend to do that as well, all of which I think moves us much closer to a sort of a mid-30s margin profile over the next handful of quarters. It's still, you know, going to be below sort of corporate margin targets. And we have other levers, of course, across the portfolio that I think help us on the margin line.
Okay, got it. So as an extension of transceivers, maybe touch on, let's touch on EMLs, right? So EMLs have been an incredibly strong business for you. Supply is incredibly tight. Indium phosphide is in incredibly short supply. I think you've added 40% capacity year-over-year this year. So how are we thinking about, how are you thinking about as we go into 2026? I'm pretty sure I know the answer to this, but it kind of feels like you're sold out for 2026 on EMLs, if I'm not mistaken, and you have long-term agreements with customers. Is that how investors should think about the EML business in 2026? Or is there room for incremental sort of capacity additions from your existing footprint at this point?
Yeah, let me talk a little bit about the capacity. You got that a little bit wrong. We have Kathy Ta, who's here with me today, our Vice President of Investor Relations, talked about before I joined doubling our EML capacity in a 12-month period and a backward look. That happened, right? We actually out-executed that roadmap. And then I think she and I thought we were sort of going to see very asymptotic improvements in our overall capacity. We brought in again sort of a new person to look at our fab strategy, and that person's figured out ways to squeeze our existing footprint fairly considerably. And so we gave in the last earnings call a new benchmark saying over the next three quarters, meaning our December, March, and June quarters, we expect it to add that 40%.
That's a forward-looking statement where we'd expect an increase in capacity of 40% on what already is.
Right, doubling.
A doubled number. So we're outputting a ton, and then we expect to increase that by another 40%. Then on top of that, as we look at the next six quarters, we've also talked about a strategy that's a very unique strategy that this person, our new leader for our fabs, has come up with, and that is to virtualize. We have four indium phosphide fabs, which gives us a very unique purview on the industry. One of them is largely dedicated to co-packaged optics, which I'm sure we're going to get to in a minute. But the other three, one is really stretched to a maximum capacity. That's where our EMLs come from. Two are relatively underutilized, and this guy has come up with a really strategy to use all three as sort of a virtual single instance of a fab.
In so doing, we would expect to get another step up in capacity over the ensuing six quarters. It's such a unique strategy in this virtualization. We don't know what it ultimately leads to, how big that number is. Kathy and I haven't characterized that as yet. You can imagine now if we can really take advantage of three factories as opposed to one, we should get a pretty nice step up in our output. We just haven't quantified that.
Along those lines, over the next six quarters, sounds like there's obviously additional capacity. One of your relative, you know, competitors, partners in the industry is adding 6 in capacity in indium phosphide. How do you think about, on a couple of lines, how do you think about the industry supply-demand balance longer term? I mean, is this a, let's get through 2026 into 2027 and then we'll kind of revisit, or is the demand, you know, existentially strong for, you know, multiple years where you still have to consistently look at ways to drive more capacity to meet the underlying demand that we're seeing, which seems to be, you know, unwavering?
Yeah, look, I mean, again, a really good question, David. I think we're seeing, you know, Broadcom has indium phosphide capacity. They're adding, you're referring to Coherent, they're doing a really good job adding capacity. Our Japanese competitors are adding capacity. We're adding this 40% we've stated over the next three quarters. In the face of all that, we've also said we see the supply and demand imbalance increasing. We're falling further behind. And that accounts for all this other capacity that's being built out here, there, and everywhere by our competition. So at least through 2027, we don't believe we catch up. We think we're still behind on supply given the demand levels we're seeing. We are facing a decision now, certainly in the next quarter, to maybe invest in a much larger footprint. And that's one that we're weighing.
We're just trying to really gauge where our customers and our demand profile is before we really go and invest in breaking new ground or investing in new clean room space that would really inflect our capacity at another level.
Well, you mentioned earlier, given your background from a semiconductor perspective and Jim's background, and you used the word cycle, it doesn't feel much like a cycle right now, right? It feels like an elongated, you know, demand backdrop. I mean, how would you characterize where we are in this particular part of said cycle versus like your historical perspective? I mean, I don't think there's an analog that I've seen in 2020, 2025 years. Just kind of love to get your perspective because we get questions from investors all the time. Do you have visibility into 2027? Visibility into 2028? Jim made reference to 2028 orders, or not orders, basically comments and potential commitments from customers into 2028. So how are you thinking about that maybe at a higher level?
Yeah, I mean, we're, you know, we have commitments certainly through 2027 in our business. I'm sure it's not dissimilar to how Jim is seeing it. You know, I lived through probably the biggest wave of demand in semiconductors, which was Wi-Fi, right? Wi-Fi was a nothing and grew to a massive technology wave. And that was very, very sustainable, obviously, right? It's a technology that continues to live on. But this dwarfs even the Wi-Fi cycle. I mean, just crazy in terms of the amount of demand that we see, the changes in the demand signal. And by the way, those changes are only in one direction, right? So it feels very sustainable. We see no slowing down. You know, we're asked the same question you are, right, on a daily basis. Well, what if, what if, what if?
At least right now, the what if is, can you please manufacture more?
Got it. All right. So let's just move to CPO since you mentioned CPO. Obviously, that's a, I think, a unique, you have a unique position in CPO from a laser perspective. The market, I think, is still kind of debating in terms of scale and scope of how big CPO from an industry perspective can be as we move through, you know, 2026, more realistically 2027 in terms of more volume. I think you said you expect your CPO-related business to start ramping in late calendar 2026 or the second half of 2026. Maybe kind of give us a sense for, you know, how you see that sort of demand curve playing out or that cadence playing out as we move through 2026 into 2027 where volume should ramp more materially.
Yeah, look, we're shipping today. I mean, the good news, you know, as you're correctly saying, I think there's been a lot of debate on CPO. We see that debate waning simply because there's enough proof points out there that it is in fact happening, and of course, we're participating in a very material way, so we're shipping today. I think we've been surprised at the robustness and the performance of the solution that we're out there with, obviously, our leading partner. They've done very well with it, but we expect an inflection point on Ethernet-based switches. That's where we see the real step up where our revenue would become more material. Today, you can't really see it because it's relatively small shipping on one switch platform.
What we've said in the 2nd half of 2026 is we'd expect a pretty big step up in the revenue, and it would become something much more material. And I think since the last time we talked, we have a lot better confidence in that timing. That timing seems to be holding very consistently and more confidence in the vector, the magnitude vector of the revenue. We expect that to continue through 2027. And then we're actually engaged with multiple customers. So it's not just the one leading customer that has made a lot of noise about co-packaged optics. We see engagements now from other switch companies, from other people that are deploying switch silicon themselves, from other GPU and CPU vendors. So it's a little more broad-based than the one customer. And obviously, that's given us pretty good confidence.
Just maybe sharp straight on this, what you're seeing. This is an opportunity largely in scale-up, right, near term. Does the Ethernet consortium or ESON a longer-term opportunity by the time they set standards, by the time protocols are determined? Like, how do you think about that as an opportunity outside of that large, you know, very vocal customer that you're working with today, particularly on Ethernet switching?
Yeah, look, it's scale-out.
I'm sorry, scale-out.
The opportunity that we're addressing today is a connection to the top-of-rack switch, so that, of course, is scale-out. We do see opportunity. ESON is really a scale-up type of technology. We see opportunities for the 1st time in a much more appreciable way on optical scale-up. I think the conversations today have really been about optical scale-up. I mean, people are starting to see signs of that. Still too early, right, to really call how optical scale-up factors in. But it certainly seems that the laws of physics are weighing now much more in our favor where CPO had been talked about for years and years. It's happening. It seems like in a similar vein now, optical scale-up is being talked about, but there's enough trials, enough activity around it that it looks like it will happen.
Whether that's in 2027 or 2028 remains to be seen. But we certainly are much more bullish on optical scale-up than we have been at long.
And maybe just touching on competition and CPO, you mentioned CPO has been talked about for a while. I think if we went back to like OFC in 2022, 2023, you know, players like Marvell were making a big stink about it. You know, Cisco's maybe a little bit less optimistic. Cisco doesn't sound as optimistic on CPO being sort of a mainstream solution that they're going to participate in. How do you see the competitive landscape today and how that's evolving and the relationship between some of the other optics players in the space right now?
Yeah, look, I think, you know, it's sort of a matter of timing. I think these first high-volume shipments that we'll participate in with our customer, we think that that helps them differentiate their solution against a Cisco, against a Broadcom. They're excited about the power savings that it brings. They're excited about the cost savings that it brings. And they've obviously done very well in their networking business, right, over the last couple of quarters. They've really been able to differentiate there. So, you know, I think that depending on the success, and we're obviously super optimistic, I think they're very optimistic about the success. We would expect other people to fall in behind. I mean, Broadcom makes a lot of noise about it with their platform, and they have the ability to deliver a turnkey solution on CPO. You know, we'll see how the landscape plays out.
Got it. So since you mentioned your large customer switch portfolio, maybe it makes sense to touch on OCS. So, right, obviously this is a new TAM for you, relatively new TAM for you. Talked aggressively about it, you know, six, nine months ago. How do you think about your technological roadmap in OCS versus, you know, the incumbent roadmap? You know, you have like Google has their own technology, your MEMS-based, competitors, Liquid Crystal. Maybe talk through like why you think MEMS, your MEMS solution is, you know, technically the right solution and has won you, you know, two, three customers that are, you know, ramping as we speak and will probably ramp aggressively as we move through 2026.
Yeah, look, I mean, OCS is an incredibly exciting opportunity for us, right? I think we have not given TAM sizes. There's a 3rd party data point out there that says it's a 2 billion TAM in 2029. Based on what we see, we can say that that is way off.
Probably too low.
Way too low. We deploy, as you correctly said, a MEMS-based solution. The one existing OCS that's out there that ships in very, very high volume is also MEMS-based. MEMS has the advantage of being effectively lossless, right? Because it's a mirror. You're basically shining light on a mirror and that mirror is redirecting traffic from one port to another. Remember, OCS is really a traffic management system. It's not a packet-based switch. It's a traffic-based solution. And having no loss is really a key technical advantage. The opposing camp, there's solutions that are based on Liquid Crystal, as you correctly said. You're passing light through something, so there's inherently going to be loss. You also have a band dependency. When you're using a mirror, you could operate in any frequency band, any wavelength.
You're just moving light around, whereas a Liquid Crystal is, by its definition, going to have to have different SKUs to handle C-band, O-band, whatever it might be. And that, again, gives us a distinct technical advantage. The knock-on MEMS is the fact that there's moving parts, right? That there's a reliability concern. And the way we answer that is, look, we've shipped these things in telecom solutions for years. They've had to exist under the ground in fiber backhaul networks for seven, eight, nine, 10 years, and they haven't failed. So we think we've got that part of the solution licked, the reliability piece.
Got it. And so I know you haven't given a TAM, but I think it's my understanding, and I think you might have said this publicly, your initial foray with three large customers sort of mimics or overlaps with your three transceiver customers. Is that fair?
We have not said that, but.
Okay, so I'll throw that out there. I'll state that with, you know, maybe different rank orders in terms of maybe the revenue opportunity within that portfolio versus transceivers. How do you think about, so I mean, is the volume ramp dependent upon the speed at which sort of AI, you know, AI data is driving, you know, training environments and then ultimately inference environments where bandwidth latency, to your point, is critical? And so as we see more customers, these three customers that we think are your customers ramp, you know, 800G and then faster speeds 1.6, that is sort of the tail end to think about how quickly OCS gets deployed in those environments?
You know, I think this is a great situation. I think the limiter on OCS is ourselves, right? The demand numbers that we're seeing for OCS are numbers that we simply can't meet, and so our challenge is building out our infrastructure, our manufacturing capability, the supply chain as quickly as we possibly can. There are a number of different use cases, as you know, and each of our three customers deploys this differently. You have a TPU, it's really an optical scale-up inside a cluster. You have an optical spine switch replacement use case. You have a GPU or XPU kind of protection mechanism whereby you're trying to steer traffic away from a failing or overloaded GPU. We're shipping to all three of those use cases.
We've given a revenue roadmap, as you know, David, saying $10 million, which will be our 1st real appreciable revenue from OCS, $10 million incrementally in Q1 of this year, ramping to $100 million in Q4 of this year. Our limiter, we could do quite a bit better than that if we could actually make the stuff, right? It's really a limiter, again, on the supply chain and our manufacturing capacity.
So maybe I'm not as familiar with what the bottleneck is from the supply chain. Is it manufacturing capacity? Is it substrate? Like, is there a golden screw, if you will? Like, what is sort of the gating factor that's limiting how quickly you can produce, your partners can produce for you?
Yeah, I mean, I'd say the long pole is our manufacturing capability. So we're going from a standing start, right? We've never seen demand on our OCS. It's been something that we've sort of had as a science project on the back shelf. We're now making sort of limited production quantities. And to get to the levels that our customers want, we have to get to a whole different set of manufacturing capacity. So we've already spent the money on the CapEx. It's a matter of deploying it. The next limiter, there are certain components in the system that are tough to get, the MEMS, you know, back to your point. We are overdriving our MEMS supplier. They've, again, never seen volumes like this. They're not used to this kind of scale.
They're used to the telecom market.
They're used to the telecom market. There's power solutions. We have very specialized DACs that go into the product to drive the MEMS. It's a voltage-driven MEMS solution. So there's several key components that we need to have come in place as well.
Final question on OCS. So obviously, it comes up in questions, and so I'll run this by you. When you think about MEMS versus Liquid Crystal, that voltage dynamic that you mentioned, obviously, since you are actually mechanically moving. Yeah, you know, how are your customers thinking about that sort of voltage dynamic difference between Liquid Crystal? I know it's lossless versus not, you know, with loss issues, but how is that sort of reconciled by your customers? And how do you think about that dynamic?
Yeah, look, I mean, I think from a competitive standpoint, we've shown, first of all, that we have a pretty appreciable lead on a 300-radix switch. The majority of the deployments are on a High-radix switch. There are some that are 64 by 64. That I would cede to our competition. I think they're doing a better job on the lower-radix switch. The MEMS argument, as we discussed a second ago, that that is a concern that comes up over and over again, but we're able to show data that we've had these MEMS-based solutions deployed in the field for five, 10, sometimes 15 years without any reliability problem. So we're able to overcome that objection pretty quickly on the backs of good data.
Got it. All right. Just in the few minutes we have left, we'll talk about your telecom business components. Obviously, that's a bit different than sort of what it looked like three, four, five years ago. I think the outlook is considerably stronger. So when we look at the portfolio today within your component business, what are you most excited about that is not only supportive to the gross margin targets that you're aiming for, but sort of growth rates maybe not on par with what you're seeing in some of the other parts of the business, but really where you're seeing sort of a step up in demand from a component perspective in that vertical?
Yeah, I mean, we talked with you and with other people that are involved in the stock in the last earnings call. The other surprise, we had two big surprises. One has been our ability to increment the EML demand. We talked about that. The second is just the broad-based nature of the scale-across opportunity, and that's affecting all of those traditional telecom-type components, our ROADMs, our pump lasers, our narrow linewidth lasers. All of these things are seeing unprecedented levels of demand. It's now our traditional telecom customers, CNS, Cisco, Nokia. These are great customers of ours. Their customer now has gone from AT&T to Verizon.
To Google, right, to Google.
To Google, to Meta, to Amazon. And they've done a great job really capturing that scale-across opportunity, but we're selling components into that. And that really has been a catalyst for our.
I know you're the component provider, so this isn't directly kind of in your purview, but like there's a subtle difference between scale-across and traditional DCI, right? In scale-across, we're connecting effectively your products are going, your components are going into products to connect sort of dedicated links between AI clusters. Is that how you're seeing your, I mean, you're removed from it, I guess, one step?
Yeah, but you're right. I mean, I think that, you know, Kathy used the example of a Venn diagram earlier today. DCI is the large circle and scale-across is the smaller circle. The specific scale-across is I'm trying to run an inferencing model over multiple sites, right? So you're actually transferring data, actively transferring data across multiple clusters within multiple sites. And that's the part of the business that has done super, super well.
As an outsider looking in, should we view the opportunity set driven largely by kind of the regional distribution of data centers because of power considerations and other sort of NIMBYism? So the more we have, you know, more distributed data center infrastructure, that's just, you know, a fairly meaningful tailwind for your component business, particularly in scale-across.
Yeah, 100%. I mean, I think it's driven predominantly by power. I've not heard the term NIMBYism, but that is an issue, right? Where you're just not going to get these massive data centers in somebody's backyard. And that's been a great tailwind for us. I don't think that that's going to decrease anytime soon.
Great. So I think we're out of time. So I want to thank you, Michael, for.
David, thank you.
Joining us. Thank you, everyone, for joining, and enjoy the rest of your stay here.
Appreciate it. Thank you very much.