Hi, good afternoon. I'm Samik Chatterjee. I cover the hardware companies at JPMorgan. With us for the next session is Lumentum. I have the pleasure of hosting Alan Lowe, President and CEO of the company. Alan, thanks for coming to the conference. Kathy, thank you as well. What we're doing is asking three questions that are more common to a lot of the companies just to get sort of your views on these before we dive into something more company specific. The first one is macro. Obviously a big discussion point for everyone. As you look to the remainder of the year, where do you see the biggest macro risk for your business?
Great. First of all, thanks for having us, by the way. I think, you know, maybe before I look forward, I gotta look back a little bit and talk about what happened, the dynamic in our industry, such that we were over-shipping demand, and that was really a result of supply chain constraints through the pandemic and strong demand from our customers and their customers. Customers were building up inventory, and as the supply constraints became more easily managed and availability of components and semiconductors in particular became more readily available, we've seen our customers work to reduce their inventory. Now today, we are under-shipping the actual end user demand until that inventory gets corrected.
Whether that's a macro issue or a pandemic issue or a semiconductor shortage issue, I think all of the above kinda culminated into what I call a situation where we're under-shipping demand. As that balances out over the coming couple of quarters, we think that, you know, we'll return to again, strong growth. You know, putting aside any issues of macroeconomic downturns or recessionary issues that, you know, could possibly impact our business. I think the other is that we are seeing some softness in cloud growth, and as our customers that support the cloud, and our hyperscale customers ordered in expectations of a 40% annual growth rate, and they came in about half or less than that. You know, they have inventory to burn off as well.
I think that's an impact that we're seeing as well. I think fundamental growth drivers and demand for bandwidth are continuing to be strong. I'm not so concerned about the overall long-term trends for our business and our customers' demand.
Okay. I was going to ask you, as a second question, the follow-up being sort of inventory versus real demand weakness, and to the point that inventory challenges could be more short-lived. It sounds like the telecom side is just saying is inventory, the datacom is more demand softness from your customers. Maybe just let me know if that's how we should think about it? When you see 3D sensing as well as commercial lasers, where would you put those?
I think you've categorized it right in the telecom and datacom space. I'd say as we look at commercial lasers, our lasers business is very tied to overall economic conditions and individual markets. Our semiconductor lasers business is down as semiconductors are investing less in the back end. We provide lasers to go into tools for singulation and scribing and etching and things like that for semiconductor. On our 3D sensing business, that's been a story of having such an overly large share of wallet of our leading customer, and that, you know, we've been warning about share normalization happening, and it's actually happened now. We've gone through that business, and we've guided appropriately. We expect that, you know, the business is more normal.
Now our business portfolio is 90% infrastructure. Telecom, datacom, and lasers, 90% of our business, where we expect 3D sensing to be, you know, less than 10% of our outlook, which makes fluctuation in consumer demand less important to us, whereas in the past when it was in the high 20%, it was, you know, very impactful. I think we're in a good place now.
Okay. Okay. The last one on that front, really just talk, help us through AI and the impact on your business. I know you just mentioned 90% of the business will be infrastructure going forward, but when we think about tailwinds, which parts of the business really benefit?
With respect to AI?
Yeah.
Yeah. I mean, I think it's an exciting time for us, and I think even where we've seen inventory build up in the hyperscalers, a lot of that is getting consumed because of the demand for higher speed optics, optical interconnect to feed the processors that are fueling the AI processing. I'd say that's step one. Step two, which I think is even more exciting, is purpose-built optical interconnects that are low latency, less power hungry, and provide parallel optics to those processors is very interesting. We're working on several different designs with leading processor companies to really have something in calendar 2024 to address that in particular.
In the meantime, that's really driving more demand for our higher speed, EML lasers, which go into 800G and soon 1.6T transceivers, as we transition from 100G per lane laser to 200G per lane laser. It's an exciting time and exciting, really incremental, market for us that we believe over the next five years could double the size of the existing Ethernet market today.
Okay. Okay. Maybe as we get to more company specific questions, let's start with the R&D more roadmap as well as sort of the product roadmap. Starting with Datacom, you mentioned sort of the move to higher speeds, but how much of it has to still play out in your R&D roadmap versus it's something that you already worked on is more sort of going to be about commercialization at limited R&D, incremental R&D from here on? Just talk us through the product and R&D roadmap on the Datacom side?
Sure. There's a lot of things going on at Datacom, one of which is what I talked about already, the 200G EML. We've designed that product already. Now it's getting into qualification builds with our customers. As that becomes a reality, it's just a matter of how fast can we ramp it, because customers really see the economic value of going to the higher speed lasers to get to the 800G transceivers and 1.6T. In addition to that, there's two other things that are interesting that we're working on in our R&D, which is to develop CW or continuous-wave lasers for silicon photonics applications and transceivers, which is another alternative tool for the data centers.
That very high power CW laser for what we call external laser source, which really is that custom built laser application for AI and machine learning. Those are the things that still need R&D in collaboration with our customers to get to market. I think that's really more of a first half of calendar 2024 product realization.
I'm asking this to your primary competitor as well, earlier in the day, but is the thinking still that all of this remains eventually in a pluggable format, or do you see technology sort of taking a leap forward in terms of either, different architectures, et cetera, or even co-packaged optics, just to achieve the same sort of objective?
I think it depends which customer you're talking to. Some are real believers in pluggable modules are the way to go for the future, at least for Ethernet for sure. And some believe that co-packaged optics are the way to go, although that's kind of been talked about for five years and has never come to fruition. I think what we're working with on for AI applications could be something very different. It's not a module, but it could be not a pluggable module, but it could be a module that provides laser source for multiple different lanes of fiber optic interconnects that then can address the things that AI is really looking for, which is lower power, higher speed and parallelism and lower latency.
That's what we're working on, that's really custom built for AI.
Okay. Interesting. Switching over to telecom, at OFC, you outlined multiple new technology approaches in your telecom business. I think there were a bunch that you discussed at OFC. Maybe help us think about the rank order of those opportunities. Which ones are the most sort of big in terms of opportunity and how to think about timing?
Yeah. I think the other thing we talked about at OFC is the Shannon's limit and what that does for our business. Basically, over the past 20 years, carriers have been able to get more data through a fiber by increasing the speed of each wavelength. There's a physics limitation with the ability to just keep doubling that bandwidth. What is now happening is parallelism in fiber optics. You know, where there was one fiber, now there's multiple fibers, that means more complex ROADM architectures, which is the optical switches. We're continuing to invest in R&D around higher port count and contentionless ROADMs. We believe that's an area for strong growth, you know, as we burn down the current inventory.
That's number one. I'd say number two, as you get into multiple fibers in a single path, you need more transmission. All of our work going into coherent optics around tunable lasers that came from our NeoPhotonics acquisition last year, coherent modulators, whether that's in silicon photonics or indium phosphide or optical receivers. All of those things go into the engines that our customers build for the coherent optics. That's a very interesting area for us as we see, you know, that being really the fastest area for growth in the coming years. I'd say we're seeing a lot of demand for our tunable access modules.
As speeds at the edge of the network get faster and faster, having the ability to go to 10 gig and then shortly after going to 25 gig in a tunable small form factor is really interesting for not only the cable operators and MSOs, but also for 5G deployments for fronthaul. That's another very interesting area. It's a little bit off topic from telecom, but the ROADM architecture that I talked about earlier, putting that into a data center and having a low power, low cost way of switching fibers within a data center, where today it's a, you know, optical patch panel that is very power hungry and creates a lot of latency. We're working with some of the leading hyperscalers on how to architect that interconnect.
It looks a lot like our ROADMs, but is much higher density, and solves a lot of the problems that happen today, especially as we go to artificial intelligence in the data centers.
Okay. Interesting. On the Datacom's side, you've talked about the VCSEL opportunity that can lead to some copper replacement inside the data center. Any way of sizing that up, what that opportunity could mean?
Yeah. I mean, it's total incremental market for us in optics. Today, the very short reach interconnections within a data center are done on copper, as you said. As there's a need to go to higher speed, copper just runs out of gas. So having a low power 100 gig VCSEL and maybe even having multiple of those to replace a copper interconnect is a pretty compelling way to have lower power consumption and higher speed on those short reach connections that are, you know, there's a lot of them in a data center. So when or if that opens up, I think it's a sizable incremental market for us.
Okay. Okay, got it. Maybe another one on sort of just thinking about the opportunity. You mentioned the CW lasers for silicon photonics. There's always been a sort of discussion around what the timelines for those are, and doesn't as, I mean, as much as sort of come through Neo terms. When you're thinking about the timing, what does that timing look like? What is the trade-off? What do you get in terms of performance versus lower cost?
Yeah. Well, well, I think silicon photonics is another tool in our tool chest, and it addresses, certainly, takes advantage of the silicon manufacturing infrastructure for cost. For some opportunities that are maybe less performance driven, silicon photonics is an excellent alternative to indium phosphide. You know, as we acquired NeoPhotonics, they were already producing CW lasers for silicon photonics data comm transceivers. Now that's proliferating beyond just the products and customers that came along with the Neo acquisition, as we work with more customers on providing them multiple different power ranges on CW lasers to address silicon photonics. At the same time, as EML laser costs come down, there's a compelling argument that both may certainly coexist and one's not gonna cannibalize the other.
I think, you know, having both weapons in our tool chest, I think, give us a good opportunity with our customers to provide them whatever they want.
Okay. Let's move away from some of the discussions on the products individually. In the past, you mentioned data comm and telecom will be double-digit growth markets. You said the company can be double-digit growth outside of 3D sensing. I mean, one of the concerns we hear from investors is what's the likelihood that that double-digit growth outlook was due to some of the overshipping that you outlined to the end markets? Why should we sort of honestly believe that there's a double-digit growth when you're more shipping to the real demand?
Yeah. I think that's a fair comment and fair assessment. If you look at the combination of NeoPhotonics and Lumentum, year-over-year, the growth rate was just over 19%. There was probably certainly overshipping at the time, and now we're undershipping about, you know, similar levels. I think from that perspective, and if you look at over time, over the last 20 years, the CAGR of the optical communications market was double digits. We're not saying that every year is gonna be double digits, but certainly the long-term growth trend because of the end user demand for bandwidth continues to be unrelenting. We believe that it's certainly a 10% or more grower.
You pile on top of that AI and other kinds of applications that are driving the need for more optics instead of less optics. I think that gives us even more confidence that the long-term trend rate is at least 10%.
Okay. Okay. Makes sense. NeoPhotonics, demand, or sort of your experience in relation to demand there, particularly when you look at the current headwinds on the telecom side, is it more on the legacy Lumentum or the NeoPhotonics products?
Yeah. I don't think it's, the inventory build was on any particular product or any particular origin of those products. It's across the board, where customers were afraid they weren't gonna be able to get the parts, they ordered, assuming that it was gonna be a continued challenge to get the supply of product. I'd say that, you know, it's across the board and, you know, I think as we get through this undershipping, we'll get back to the kind of revenue levels that we have seen in the past, especially as the market is expected to grow, you know, at that double-digit rate.
Okay. Let me pause here and see if there are any questions. Any questions? Okay. Let me continue here. On the data comm side, going back to data comm, obviously with AI, and with some of the changes recently in the sort of customer base, there have been newer customers as well coming in on the data comm side. What are you seeing in relation to demand from traditional data center customers versus newer customers that are looking for something more differentiated, more direct sourcing, from Lumentum?
Yeah. I think, as I said earlier, I think that we've seen. By the way, each hyperscaler is not the same. There's differences of architecture and differences of the design for those modules that are going into their data centers today, and some customers that have continued to have robust demand. I wouldn't categorize hyperscalers as one oversupply issue. It's really more of a customer-specific challenges there. That said, you know, as we look at, you know, the more recent demand has been more heavily weighted towards 800G per module, so that's 8 100G lasers, which is a good thing for us. You know, it consumes a lot of lasers in one module, and so, that's pretty exciting.
More excitingly is what I talked about earlier, which is the 200 gig per lane. We expect that to ramp into production really by the end of the year and early into calendar 2024. As I said, you know, the CW lasers and the external laser source for AI applications is incremental to all of that.
Switching gears to 3D sensing, just to dive into that a bit more. How are you thinking long-term growth drivers there? I think this was probably 2 OFCs ago, so not the last one, the one before that. You were outlining sort of the headwinds that you were facing in that segment because of the higher share that you had with your primary customer, at the same time, sort of expecting to recover that in a few years, from diversified sort of opportunities of growth that you had. How do you think about now, sort of what's been the trajectory in terms of that business through share loss? How long does it take to get back to the same sort of revenue level?
Are you seeing enough sort of enough of an offset in terms of growth from, outside of that primary customer?
It's gonna take some time. I mean, I think we've been in the 3D sensing business with our main customer for six years now, of which the first four we had, you know, probably north of 90% share, always warning that that's not healthy for either us as a supplier or our customer to be that reliant on one supplier. I think what we've been saying is that share normalization would take place. It has, and it is. Having that business be, you know, less than 10% of our outlook for next fiscal year, I think is a healthy way to look at things until we really diversify that business into new customers, new markets, and new applications, of which we highlighted on the last call.
In the March quarter, we had approximately $3 million of revenue from LiDAR, from automotive LiDAR. Which, you know, is a small number, clearly, but it's a sign that, you know, it is starting. It's taking a lot longer than I would've liked. I do believe that in two to three years from now, that's gonna be a meaningful part of our business and provide us that diversification that we're looking for, in addition to applications in AR/VR that could use multiple light sources, from us to address one pair of glasses or a headset.
Okay. A bit more near-term question, but related to the 3D sensing. We've heard cautious commentary on share from both you and Coherent. How much of the concern about market share is something that you expect to see with a new supplier coming in versus actually reflecting in the order trends or sort of communicated share? We've known historically in this market, new suppliers also tend to sort of ramp slowly, slower than expected. Because both are now guiding to share loss, I'm just wondering how much of this is anticipated share loss versus something that you're seeing in the orders.
I'll tell you that we had anticipated share loss a couple years ago with a new supplier coming in, they completely failed. We got the windfall of that. I don't wanna set expectations that that's gonna be the case this time. We wanted to make sure that there were realistic expectations for our fiscal 2024 for 3D sensing. If that third supplier doesn't come in, I think, you know, we'll be in a better situation with regard to revenue and share. I'm not holding my breath. I think they'll figure it out this time, you know, I think having them be less than 10% is not a bad thing in that, you know, fluctuation in consumer markets, especially in uncertain economic times, can really swing the demand.
you know, for us, then it becomes a 1% or 2% swing as opposed to a 10% swing of our business.
Okay. The reason I ask is, does that then automatically mean the competitive landscape for AR/VR is tougher than we imagined two years ago when it seemed like much more of a slam dunk for Lumentum to really gain content on that front?
I think it comes down to, you know, the investment in the customer relationship and doing what the customer wants you to be doing with respect to technology development. We're partnering very closely with the AR/VR customers as well as the automotive customers to be that partner of choice for them so that it will be a slam dunk. Hopefully, those investments will pay off, and those relationships that we've built with those customers will pay off into, you know, a disproportional size share of wallet of their spend.
Okay. Okay. I'll move to a few financial questions, any questions from the audience in the meantime? Okay. Let me move to the midterm model that you outlined at OFC, $2 billion-$2.3 billion in revenue. With the losses that you've seen, I think some are, as you're outlining, more temporary, some are more of a market share driven loss. How do you think about the timeline to sort of achieve that $2.3 billion of revenue target? Is it now more delayed than what you imagined at OFC as well?
Well, I'd say, you know, we're not gonna guide multiple years into the future, but I'd say, you know, given that we are under shipping today, and once that inventory does get adjusted, we'll see an immediate bump up in our telecom and datacom business. Then you put the normal 10% CAGR on top of a normalized shipping in at the same level that our customers are shipping out. I think, you know, that's certainly in the realm of possibility for us to achieve in the next few years. I'm not, I'm not suggesting that we're not gonna make that, but I do suggest we gotta get rid of the inventory at our customers. I think fundamental end drivers of bandwidth requirements are unchanged and unrelenting.
I think at the end of the day, there's plenty of opportunity as long as we execute in R&D.
Okay. The same question if I were to ask you on margins, 3D sensing business, where share loss is a high margin business for you. Telecom is a high margin business. How do you think about the margin impact related to sort of those targets as well?
Yeah, there's a few dynamics, right? One of which was the NeoPhotonics acquisition. Basically, that when we acquired it was a break-even operating margin business. We said, we'll get $50 million of synergies as we integrate that business. At OFC, we upped that synergy target to $80 million, we think that's very doable, and we'll be getting that in the next several quarters. You know, you take a $400 million business or $350 million business that's break-even, you add $80 million of synergies to it, that's gonna help a lot in our telecom margins overall. That's one step. The other is, Datacom is a very important business for us as well.
We sell at the chip level, and those margins are quite high as well. We think that, you know, once we get rid of the Datacom inventory, we'll get back to the levels that we had before, and be able to produce, you know, substantial margin improvement as a result of that. Those are the two main things that need to happen, and then we gonna get back up to north of the $2 billion mark to get back to those margin models. You know, I think we've already talked about that. I think we're in a good place, and I think the differentiation that we provide to our customers in things like high port count ROADMs and coherent components and tunable lasers, I think best-in-class products provide us the opportunity to drive our margins.
Okay. On the operating margin, one of the longer term guides, long-term guides that you had at OFC was that you'd that you're targeting 26%-28% operating margin long term. If I remember, sort of pre-Neo acquisition, you were hitting 30% already on the operating margin. In fact, you were sort of doing some upside to that in. If I think about the difference there, which almost is at that point, sort of 300 basis points to maybe 400 basis points on the long run, it seems like some portion is the Neo, because you're saying Neo is going to be slightly less than 25% operating margin with the synergies baked in.
Right.
Is Neo the only difference that you see in terms of getting to that 30% operating margin, or are there other buckets in there, like 3D sensing, others, in terms of keeping you from that 30% margin?
Well, certainly having, you know, 20%-30% of the 3D sensing revenue, like 30%-40% of 3D sensing revenue we had in the past is a hurt, a headwind for our margins. I think we're doing a lot of other things to improve efficiencies, to drive internal manufacturing, to consolidate fabs, to do things that will really free up operating income and drive us back to those levels, even without a recovery in 3D sensing. That's why we came off of our 30 points of operating margin down to the high 20s, and I think that's totally doable even without a recovery in 3D sensing.
Okay. Okay. Got it. Maybe just switching gears here to a couple of opportunities. One, you've talked about progress in relation to the DSP business that you acquired from IPG. Maybe just give us an update on where things stand in terms of sort of your current visibility into getting it to the market, and how much of a cost sort of benefit are you looking from it? Is that embedded in the model as well, the long-term model?
Yeah. I wouldn't say we've incorporated that yet into the model. We're not counting... Well, I'm counting on it in my head, but not counting it on the external targets. Number one, we're still gonna rely on third-party DSP suppliers for the foreseeable future. We have great partners that we've worked with and we continue to work with to get new products out to market. In fact, our existing 400G ZR and ZR+ is based on a third-party DSP. We view the investment that we're making and made in the acquisition is really a significant way to drive our costs down.
First product out will give us an incremental big step down in the overall 400G ZR and ZR module cost that I think will maybe even open up the market even broader. We will have products after that for 800G and 1.6T. Again, we're gonna continue to rely on third parties to get us to market first and then come in with a cost reduction effort, at least for the next few product introductions that we have. Things are going really great. The team in Brazil is fantastic, and we have a lot of DSP expertise that we brought on board since the acquisition.
It's pretty exciting time, and one that I think, you know, we'll have our first results in not too distant future, and I'll share those with you when we get them.
Good. In terms of opportunities, the other thing that sort of, as I was writing out these questions, came to mind is you have capabilities at the component level, you have component capabilities at the chip level as well. How does that position you with the current manufacturing footprint in terms of benefiting from funding plans like the CHIPS Act, do you see an opportunity to participate in that given the capabilities that you have?
Yeah. I mean, we have fabs. We have two fabs in California. We have two fabs in Japan. We have a fab in the UK. All of which countries have a equivalent kind of CHIPS Act, and we're working to try to understand them better, so that when we do need to make investments, and we are making investments in all of our fab capabilities, we may be able to tap into those types of funds. You know, we're still learning and still talking to those government officials across the globe, and what's possible and how do we apply for those funds to help us with those investments.
Okay. Got it. Any questions in the audience? Let me ask you a couple more and wrap it up. LiDAR, definitely an interesting opportunity that's ramping but slowly. Firstly, the $3 million of revenue that you have, can you just clarify who are the customers there? Are these the LiDAR players that we hear of or are these sort of OEMs? Like, what is the nature of the customer there? From what we can see, most of this market starts to ramp sort of 2024. Really, in terms of volumes, what are your thoughts in terms of timelines?
I think, you know, we're engaged with many customers from OEMs to tier ones to tier twos and other LiDAR manufacturers. The early adopters for LiDAR were in China, so most of that revenue comes from our partner that we announced publicly with them, called Hesai.
Yeah.
I think they're doing quite well. I wish they were doing even better and ramping faster, but I think it's, automotive industry is one that we're learning about. It takes a while to get from, you know, the $3 million a quarter to $30 million. So we're looking forward to that and working with Hesai and others, to get to that point. You know, I think they're winning business.
It's one of those things that as we learn about, it takes a long time to get in, but once you're in, I think it's shown to be a very sticky business and one that, you know, we're continuing to partner with who we think are gonna be the winners in that space, to really make that a meaningful part of our business in the next few years.
Okay. Last one for you before we wrap up. Obviously, there was sort of the intent to diversify the business in terms of going after the legacy coherent. You've done some amount of that organically. Where do things stand now in terms of your intent to diversify the business, either organically or inorganically? Like, if you separate out the two, how much do you want to sort of delve into each? Particularly when you think about organic sort of diversification, what are the areas you're more concerned? Because with everything happening around Datacom, AI, et cetera, you would probably get more indexed to communication than before.
I think there's also organic growth in adjacent markets for existing technologies or similar technologies. One of the things we're, you know, very proud of is our entrance into ultrafast lasers for solar cell processing. The only thing preventing us from getting more revenue there is our ability to ramp those products faster. That's an area. We're looking at EV battery welding, other kinds of EV type of applications. We're also looking at other markets that are, you know, even more far afield that could be using photonics that aren't, where photonics could be very disruptive. As we start introducing those products, we'll talk more about those on our calls. It's very interesting what photonics can do to a lot of industries.
Our organic effort is quite robust, and we're investing in new markets there to see what is possible there. We're gonna continue to look at acquisitions to diversify. You know, I think the best way to diversify for me personally is looking at organic ways of disrupting markets that can really take advantage of photonics.
Okay. Great. We are almost up on time, I'll wrap it up there. Thank you. Thank you for coming to the conference. Thank you, everyone.
Thank you.