Coherent Corp. (COHR) Earnings Call Transcript & Summary

Hi. Good morning, everyone. I'm Samik Chatterjee. I cover hardware companies at JPMorgan. The company presenting next is Coherent, and we have with us Sanja Parthasarathi, who is the Chief Marketing Officer. Sanjai, thanks for making it to the conference.

I'll start with some of the questions we've been asking as common questions to all of our companies. And the first one is around the macro. And really just to outline your thoughts about as you look to the remainder of the year, where do you see the biggest macro risk to your business?

Yes. Thanks, Samik. When we -- macro risk when we talk this morning, so many people have -- everybody has their own view of it. I think from our perspective, we see interest rates and the geopolitical tensions as sort of the two macro factors that we are watching. And I think that's one of -- these two factors we think will leave our business at sort of moderated levels in fiscal '24. That's kind of our view right now.

Okay. And when you think about the macro led headwinds, can you maybe delineate a bit where do you see sort of the headwinds being more of a function of inventory versus a real demand slowdown in your business?

I think if you look at all of our market verticals, longer term, 4, 5 years, they're all growing at double-digit CAGR. Some markets are growing, as you know, silicon carbide as an example, is growing at 30%, 40% CAGR. Short term, we do -- I think certainly in the communications business and specifically in the telecom part of our communications business, there has been an effect due to inventory. If you look at our OEM customers, they're all announcing record quarters. Their backlogs are at record levels, but yet their inventory is 3x, 4x prior to COVID. So as supply chains kind of have been easing and they're now watching their cash flows carefully and so they're dialing down our inventory, and that's really what we see. And we believe this will play out for the next couple of quarters, and then things return back to normal levels and that's our expectation.

Okay. Switching to a longer-term topic, but we're asking all of our companies sort of maybe this is a bit early in the discussion than you want it to be, but how are you thinking about AI and the impact of that on your business? How are you thinking about disruptions as well as opportunities for the business?

Yes. I mean the AI -- it's on everybody's mind. I think everybody tries to connect with AI. We are -- our datacom business is directly connected to the AI as our -- as -- we believe 50% of our datacom market will be driven by AI in the next 5 years. Right now, it's about 10%. But AI is pervasive. I mean when we look at our phones, the amount of predictive analytics that it's doing and all of that, I mean, ultimately, it needs sensors. So every AI application awareness of the world, indoor navigation, there are just so many things that AI needs in terms of sensors. So we're going to see -- I think it's not just -- I mean there are many markets that we're exposed to are going to be driven by AI. Datacom is bang direct, but there are other things that people don't readily associate with AI, and I think sensors is one of those things.

Okay. Interesting. Maybe just starting on the datacom side then, since we are discussing AI and the direct impact on datacom. Maybe flush that out a bit more, like how do you think about the datacom portfolio, it's ability to leverage AI, particularly in terms of what I think we often get as an investor question is, is this more of a volume opportunity or is this where you have to invest R&D to move to next-generation products pretty quickly, and it's more of a pricing sort of content opportunity as you move forward?

Yes. So we've been selling into the AI market for quite some time now for maybe over a year. Today, we believe about 10% of our datacom transceiver revenues go directly into AI, supporting AI clusters, so to speak. And we think that in the next 5 years, that market 50%, as I said, will be driven by AI. And if you look at what are those applications, these applications are short reach, 30 meters or below very high speed. So they're pushing the cutting edge of transceivers. And the great advantage that we have where we are completely vertically integrated. And ultimately, every transceiver, the heart of the transceiver is the laser. And that's -- so they would depend on these next-gen lasers, 200G and beyond, which is what we are at the very leading edge of that platform and OFC you may know that our largest fiber optic conference, we announced on laser. So it's really high-speed, short distance, leading-edge performance is the AI requirement. And I think we're kind of uniquely positioned there to service that market.

And just to be clear, so the way you're defining because I think there's a lot of companies who are defining sort of what they're shipping into AI use cases differently? Are you defining what you're getting in terms of AI revenue as a certain speed that when you get -- when you're shipping those products at those speed you're assuming it's an AI infrastructure or how do you know at the customer end, what is the...

Yes. So certainly, we know certain applications are really driven by 800G and beyond today. Now but that changes in a year from now, maybe it will be 1.6 tera, which is going to AI and 800G becomes more traditional networking. So it's very hard to do a 1: 1, but in general, we do know some of these applications that AI applications, we know this class of transceivers are going into this type of application. We've got some idea of that.

Okay. And maybe the sort of following on to that, when you now think about we are at 800 gig and we move to the next-generation products, do you -- what sort of -- what's your thought process around the time line around moving to these next-generation products? Like what was that traditional timeline in going from 100 to 400? And now do you see that time line shrinking as you move to next-generation products because of the AI investment?

Yes. We are -- 800G is the highest high-volume commercial deployment that's happening today into the hyperscalers and 1.6 is coming on board. Typically, these cycles are 3 to 4 years. We now see maybe these be like 2 to 3 years because we expect to ship 1.6 in some level of volume in fiscal '24, a year from now. And so -- but the cycles are very long. I mean 100G is still, I would say, maybe half of the market today is still 100G. So these cycles are very long. They go through 4, 5 years before they completely kind of die away in a way.

Okay. At OFC, the presentation that you had, one of the things you made pretty clear was you're still expecting pluggables to be sort of the bread and butter of the optics industry till 2030 and co-packaged optics don't really sort of show up in the road map in terms of commercialization by then, does AI change that?

Not -- there are -- okay, maybe the way I answer that is, 2 -- if you look at datacom transceivers is a very broad portfolio. And that data -- the pluggable transceiver is not going away for the next 10 years, that's our expectation. But there will be pockets of co-packaged applications including in certain clusters, et cetera. So -- and we are well positioned to serve into those. So it's not -- we often say the best application for co-packaging is in a pluggable transceiver because you can still -- you can have these, what we call chiplets close to the chip, to the switch chip. And so there are all these architectures that are emerging. But co-packaging is not an either/or. And -- but we believe from an overall market, including AI, including all the applications that we have visibility to that pluggable transceivers will stay dominant as a -- from the market perspective.

Okay. Okay. One of the questions that come up quite a bit is when you think about the datacom portfolio, does it have the opportunity to support AI clusters that are depending on NVIDIA's technology stack and InfiniBand or is it going to be more sort of contained to supporting investments around Ethernet?

So we sell into all those applications today in InfiniBand traditionally Ethernet and there'll be a combination. And there may be new protocols and new things that may come up 2 years from now that we don't know of. But from an optical transceiver perspective, protocols are don't really matter. At the end of the day, it's a speed and you're taking data, whether you're taking Ethernet payload or you're taking InfiniBand payload, it doesn't really affect the revise that we make.

True. Good. Now coming back to the more traditional sort of non-AI discussion, 200-gig-plus, I think you've said recently is 50% of Coherent's datacom revenue. But at the same time, you continue to talk about for those share gains as an opportunity. I mean, you're already one of the large players in datacom overall on a global basis. So where do you see sort of those opportunities for share gains? How are you thinking about market share near to longer term?

Right, right. Yes, it's a good question. I would say, Giovanni often says, we have 1.5 hyperscalers today. I mean that's the amount of exposure we have to hyperscalers. So we have been increasing our -- as we said on the call, 50% of our revenues last quarter came from hyperscalers. And almost by definition, it's the high end, the most advanced the high-speed product. The heels of the Finisar acquisition, our exposure to hyperscalers was probably 15%. So over the last 3 years, we've been building it up. And we believe that as the drive goes towards higher and higher speeds, the vertical integration is going to be really important. It's a big deal to our customers today. Having a diversified manufacturing base, non-China manufacturing -- look, there's a whole bunch of things that we have that our customers seem to value. And that's our confidence in growing it. But the big one is really vertical integration and the in-house laser platforms because when things get really, really complicated in terms of implementation at the higher speeds. It's great to have all the platforms under one umbrella as opposed to some of our pure-play competitors who are essentially assembling things buying bits and pieces and assembling it, we think it's going to be more challenging for them as things go higher speed because there, our customers are trying to push every little inch of performance, if you will, right. They're pushing the limit.

One of the outcomes of that and what we've seen, obviously, on the telecom side play out a bit more has been industry consolidation, right? The R&D road map is challenging sort of having a vertical integration across the portfolio is a more beneficial place to be. Now do you see that changing near term in terms of more industry consolidation on the datacom side as well? Would you be a consolidator there or how would you sort of -- how would you play that investment side?

Yes, it's a good question. In the sense that if you look at our datacom -- if you look at broadly our datacom and telecom portfolio, I'll start with datacom. We are completely vertically integrated, but we don't -- we were not making DSPs for a while. And we've been quietly investing over the last 3, 4 years on DSP, and we announced our first DSP product. It's not a datacom product, but it's a telecom product, but the skill set and the product portfolio is largely the same. So we would expect datacom products to come out as well. On the telecom side, there are always opportunities. I mean people look at the Coherent acquisition. I'm talking about legacy Coherent, really, there is no interaction with communication, but that's not really true. We were -- we did not have specialty fiber for amplifiers. Telecom amplifier as we buy fiber. Now we have -- I mean we still buy fiber, but we now have a source for fiber for specialty fibers that came from. So there are always things too, but we really don't see anything else that with that, we have kind of completed the portfolio. There's really nothing else that we need in terms of acquisition or anything from a datacom perspective.

Okay. Okay. The market for 800-gig products, how do you think about adoption from customers? Should we think it will be a smaller base of customers restricted to hyperscalers or do you see more broad-based use cases outside of hyperscalers as well over time?

Yes. I mean hyperscalers are not -- I mean the market is, I would say, 60%, 65% of the market today is hyperscalers. So it's a big fraction of the market, but that is non-hyperscalers. Now if we go 5 years back, then hyperscalers were just getting started. So that portion of the market, smaller data centers. There are lots of Tier 2, Tier 3 data centers and then the enterprise. Whatever happens -- over time, the -- whatever happens in the data center happens in the enterprise, it just takes more time. So we see that happening.

Let me open it up here and see if anyone has a question. Any questions in the audience? Okay. Let me continue. On the datacom side, another question that keeps coming up is the pricing headwinds. It's been roughly 10% per annum in terms of pricing headwinds that you've seen. How should we think about sort of historical price declines in 200 gig or sort of 400 gig products? Are you on those sort of newer products as well running at those sort of typical 10% pricing headwinds? And does some of the inventory digestion that we're going through from the hyperscalers, is that changing the pricing dynamic?

Yes. When we look at pricing, we look at it from a dollar per bit perspective, we don't -- because ultimately, that's the value that the customer -- that our customer sees, that's the cost the customer sees. So when we go to higher speeds, the dollars per bit, almost by definition, drops for the customer. However, our best margins are at the higher speed product because that's where we get the maximum value. So you have to look at it from a margin perspective on the transceiver cost perspective from a dollar per bit. So we don't see anything -- we really -- I mean our customers can -- the high speed, they will take every transceiver we make. The demand is tremendous and it's just our supply that is holding them back. So yes.

Maybe asked another way, I mean, 100 gig went through a significant price moderation what -- maybe if I play out that scenario, as you said, like the higher speeds as where the demand is, what prevents 100 gig from going through another sort of step down in terms of pricing because that still contributes a lot of revenue for you?

Yes. You'd be surprised, Samik, if I told you, we are still shipping a lot of 1G products and sub-1G products. And the pricing with any of these products, yes, it goes through some erosion. There is first -- it starts higher and then starts to go through some erosion. Then after that, erosion stabilizes. I think 100G has probably reached a time where that erosion has stabilized because -- I mean, there are some applications -- because the market kind of migrates onto the higher speeds, then this becomes sort of a niche kind of a market and there's really no more price erosion there.

Okay. Okay. Let's switch gears to telecom. One of the questions you probably expected already would be how long do you expect the inventory digestion from your customers to last? And maybe in that relation, have you done any sort of estimates around how much over shipping was happening relative to underlying demand in the last few quarters? And then sort of what's the -- what is under shipping at this point that's going on?

Yes. So we've been -- our -- if you look at our OEM customers and those who are reporting publicly, especially those who were all optical or those who are reporting part of optical, they've all had record quarters. So record quarters for them. Their backlogs are very strong, but their inventory levels and those of them who report inventory, you can see that it's 3x, 4x before what COVID was. And so based on the -- the end market is still very strong. Yes, we'll go through some generations, but the long term is also very strong. So I think based on our analysis, we think this -- the telecom OEM side, I think it plays out for about 2 or 3 quarters before inventories return to normal and then ordering patterns also become normal. I mean the ordering patterns are -- when they couldn't get enough product, they were ordering a whole lot, right? I mean that's what -- I think that's what kind of led to the -- to where we are today. But the underlying demand is strong. If you look at telecom, you see the Broadband Act, which is $65 billion of commitments. Now how much of that actually makes it out to the components is needs to be figured out, but we believe it's definitely a growth driver for the telecom business, especially in the access where we're talking about access transceivers. And anything you do in the access will ultimately drive the need for gear at the core network.

So maybe going back to the question, what's your lead time now on most of the products you're shipping to your telecom customers? When we do eventually see an uptick in demand, like how much of a advance notice do you get because of the lead time? When do we expect to start to see that into your order pipeline?

Yes. Our lead times have kind of moderated back to pretty close to what it was prior to all the supply chain issues. So it's -- and it's hard to say because our -- we sell products at all levels of the value chain. So there is a -- some are on the off-the-shelf and some takes 12 weeks to build. So it's -- but in general, if you say, how are lead times now compared to before COVID and I think it's kind of turned back to more of a normal. And our customers are also going back to a more normal ordering pattern, whatever that is specifically.