Fabrinet (FN) Earnings Call Transcript & Summary

Good morning. I'm Samik Chatterjee, I cover technology, hardware at JPMorgan. For the next session of the fireside chat, I have the pleasure of hosting Fabrinet. With us is Seamus Grady, CEO; and Csaba, CFO of the company. Thank you both for being here. Garo, thank you for being here as well.

Starting off, just you -- I mean, just talking a bit more about the revenue growth that you've seen in recent years. You've seen double-digit revenue growth for the last couple of years, and many of your customers expressed the outlook that they see double-digit growth for their businesses as well. When you sort of think about the current headwinds that you're facing, does that call into question sort of a double-digit growth rate for your business? And should investors really just think of this as temporary in terms of some of the headwinds? Or does it have any structural sort of implications in terms of the growth rate?

Yes. I think the growth we see in the last 5 years, we've grown at a compound annual growth rate of about 14%, that's been a little bit more in the last couple of years, 18% to 20% the last couple of years. And a lot of that growth has been fueled by increases in our Automotive business, increasing market share in our Optical business, but also some significant wins in the complete network systems space that we've been able to secure. Will we maintain that level of growth? I'm not sure. I think we generally guide 1 quarter at a time, but the goals we set for ourselves internally are we like to grow at 2x the rate of growth of the industries we serve and about 3x the rate of growth of the contract manufacturing industry because we're a contract manufacturer, so that would put us, if the Optical industry grows, at kind of mid- to high single-digit growth rates and the contract manufacturing industry grows at low to middle digit -- single-digit growth rates. We think certainly growing at more than the contract manufacturing industry is achievable and growing it a little bit more than the Optical industry, we think is achievable. But we'd be reluctant to put a specific number on it, but we're just working hard to continue to win market share and win new business, and all that growth in the last 5 years has been organic. We haven't grown through acquisition. It's all been organic growth, so we work hard at that and making sure we have a strong pipeline of new customers and new business.

Maybe just to sort of ask that question another way is, as you said, you've seen 18% growth, 20% growth recently. One of the concerns investors have is, is it more that those growth rates have been elevated just because of pull forward of demand from your customers? And as we sort of look at the next maybe 12 to 18 months or even 2 years as a period, is there going to be sort of under or below that sort of run rate -- below run rate performance in terms of growth for the next couple of years as you -- as customers digest the pull forward that they had over the last couple of years?

Yes, I think the majority of the growth that we've been able to achieve has been from new wins that we've had, so we won a significant piece of business from Infinera a few years ago. I think they're presenting here where they acquired Coriant, and we transferred all of the Coriant business from Berlin to Bangkok. We then won a complete network system, piece of business from Cisco, which we took away from one of our competitors a couple of years ago. And then more recently, we won complete network system business from DZS, so the majority of the growth has come from new business wins. That said, we are hearing about some inventory correction that's going on, which, of course, would imply. There was inventory build going on in the last while. How long will the inventory correction last? We're not sure. We believe based on what we're hearing from our customers that is probably 2 quarters. And we'd be -- this would be the first of those 2 quarters, at which point that inventory digestion period should be behind us. So the other question is, okay, is it inventory correction? Or is there actually a softening in demand? The honest answer is we don't know. We get 13 weeks visibility typically from our customers. They don't give us great visibility beyond that. But certainly, the -- we call it the mood music we're hearing from the customers is that it's an inventory correction as opposed to some large-scale reduction in demand, underlying demand seems to be quite strong.

Okay. Staying on the topic of growth and maybe you talked about sort of what your aspirations are in terms of growth, but how do you generally tie that into long-term planning for even capacity and particularly now that you're going after a lot more systems business, does that change sort of the nature of opportunities that you're going after just as a function of trying to prioritize that systems opportunity a lot more, what does that addressable market look like at your customers that you can go after?

Yes. We think it expands the addressable market. Historically, we have not been that strong in the system space we have been in the last few years. But prior to that, we were more of an optical component focused contract manufacturer. We think it presents a number of opportunities for us. We're typically at any point in time, we're pursuing, I would say, right now, believe it or not, 15 to 20 companies for system -- Compete Network System business. Because in our business, the sales cycles are quite long. It's mostly about relationships and convincing the customer that they should have us manufacture their systems. It takes time. So we have to have several of these opportunities that we're working on at any point in time. The cadence we've been on, we've managed to win one of these about every 18 months. That's probably not a good predictor of what would happen in the future, but that's kind of the cadence, and that's probably as fast as we could absorb. Maybe once a year, we could absorb one of these big wins, but here once every 12 to 18 months seems to be the cadence we're on. In terms of capacity, most of our manufacturing is in Thailand. We have a number of new product introduction facilities, our on-ramp facilities as we call them in high-technology locations, we've won in Silicon Valley. We call it Fabrinet West, and we have another one in Israel, Fabrinet Israel; where we try to win new products from customers, ramp them until the yield is stable and the volume is about to take off, and then we transfer the manufacturing to Thailand. That has worked very well for us, So most of our capacity is in Thailand, and we have a lot of capacity there. We just recently opened in October the newest building, it's a 1 million square foot facility with capacity for about $1 billion of revenue, so we think we'll be a few years getting to the point where that's full. And the approach we take because we own a lot of land up there, once we get to 70% utilization in that building, we'll then pull the trigger on the next building, which will also be another 1 million square foot facility. That takes about 18 months to build, and that's about $50 million of CapEx, so that's the approach we take. We don't have to worry about having hundreds of factories all over the world. We have a very compact manufacturing footprint, which makes it easier to manage and easier to kind of predict what we'd be doing in the future. But like I say, once we get to 70% utilization in a few years from now in Building 9, we'll pull the trigger on the next building.

Going back to inventory digestion and obviously, I'm sure you're getting these questions the most today as well in the meetings. You said you have visibility from your customers of a 13-week window right now. So then for your internal planning purposes, how are you sort of putting the guardrails around it of like how do you manage your business step going -- already stepping into the next quarter or thinking about the next quarter? How are you really putting guardrails about how sort of -- what the magnitude of that change can be on the revenue side? And how do you manage your cost structure then?

Yes, a good question. So our cost structure is very lean. We generally operate at just the financial model we have. Our gross margin is usually between kind of 12.5% to 13%. Our OpEx is about 2%, a little bit less, so our operating margin is about 11%. We run at about 7% fixed cost, 7% of revenue, so our fixed cost base is extremely low. What that means in practice is because we have a very strong balance sheet, we're able to really plan for any project that comes our way, we can fund it, we can grow, so we can really capitalize on the upside when there's a -- either a gradual upside or even a sudden upside. If there's some big explosion in a particular part of the market, we're generally very good at capturing that revenue, so we're able to capitalize on the upside. But because of our fixed cost, we've maintained very low fixed cost, we're able to respond very quickly as there's a downturn, so whether the current situation is a temporary inventory digestion or maybe the start of something bigger, we're ready to respond, and we're very good at moving quickly to preserve the margin effectively and to reduce our costs. So the guardrails, if you like, are kind of already built into our business model because of the strong balance sheet to make sure we capture the upside but also the low fixed cost base to make sure we guard against any downside that comes along.

Okay. Moving to just telecom and that's when -- that's an area of the business where we've heard more from some of your customers about the inventory digestion. When you sort of look at the headwinds in that business, is it focused on certain technologies or components in particular, like are -- is it more focused, the inventory headwind, on the legacy products versus the newer products, how would you characterize where the pockets of weakness are the greatest?

Yes. Well, I think first of all, the [indiscernible] in telecom more so than datacom. We've certainly seen continued strength in datacom. We have seen some inventory correction on the telecom side. Some of our customers have talked about that, our direct customers and also our indirect customers, and we're not immune from that. I mean, some of our customers who have already announced, maybe they've talked about the revenue headwind to them because of inventory corrections. And again, we're their supplier, so we're not immune from that. We haven't actually specifically called out the amount of the headwind due to these inventory corrections, but we're not immune. And we've been able to largely offset those inventory correction headwinds with growth in other parts of the business. But I think it's on the telecom side, it's on the network system side, and then that flows through again to our customers and through to our business. But I think the specific products in question, I'd probably leave it to others to talk about which particular products are causing the headwinds.

Okay. I know your customers are really sort of driving your 13-week window and what you sort of really build capacity towards our plan for growth in the telecom business. But as you outlined, you have a certain view of what the optical industry has historically done. When you look at the drivers for the optical industry, particularly related to the telecom products, do you see some of those drivers continuing? Like when you evaluate what the optical industry probably looks in terms of growth rate for the next 3 to 5 years, what are those essential drivers, why telecom for the optical industry remains a big growth driver?

I think we really look to the industry kind of analysts and the industry experts that we look to their reports and data that's out there. That's typically what we base our plans on. The growth drivers, of course, everyone knows there's never enough speed. There's never enough bandwidth. And I think the underlying growth drivers for our customers business and hence for our business, they remain intact even if the short-term headwinds but they remain intact. So I think those drivers are very, very strong. But we look to, again, the kind of industry reports that everyone looks at the same reports, I think, as we look out to shape our thought process.

Okay. Let's move to datacom. Your revenue trend there, which has been sort of moving up sequentially in the last few quarters, that's been counter to the inventory digestion that we've heard more broadly from cloud customers in recent quarters. So maybe just help us think through what's driving the sort of almost like contrary trend to what other suppliers are seeing in the datacom side.

Yes. I think there's a couple of things. I mean, under the -- I would put it under the general banner of new products is probably the short answer. The longer version is the hyperscale companies who are, if you like, the customers of our customers. There -- some of them have talked about inventory digestion and whatnot, but they don't really move in unison. I think in general, the trend is from more bandwidth, more speed and we're the beneficiary of that. As they move from, let's say, inside the data center from 100-gig transceivers to 400-gig transceivers, inside the data center, that's certainly a driver of growth for us. And maybe the more significant one in recent times has been artificial intelligence data center growth. For a while, I suppose, we were wondering how -- what would artificial intelligence mean for us. And I think it's becoming maybe clearer now that what it means for us is more significant growth in very short range, very high speed, low latency, low power interconnects, both rack to rack, but also inside the rack. So we've seen some nice growth on particular product with the customer of ours. It's an 800-gig non-silicon photonics transceiver that's being used in data center -- AI data center applications by one of our customers who's installing it as we speak. And that has been a really good growth driver for us and has really, if you like, offset maybe some of the softness we've seen in the rest of the business.

Okay. So let's talk about 800 gig then. How big of an opportunity do you think it can be? Who are your primary customers right now? And do you see that customer list expanding beyond sort of 1 or 2 really more sort of concentrated customers? Or do you see this more being, okay, this is going to be sort of the group of 4 or 5 that are going to be really the customers and we scale with them? How do you see that evolving?

Yes. I think it will expand and evolve over time. I mean, right now, we're fortunate to have 1 particular customer who's doing very well in that space. We haven't named them as a customer. They haven't named us so we're quite cautious about naming our customers unless they name us first or unless, of course, if they become a 10% customer, we'll name them. But that business, it's almost like a new category for us because it's a new application. This interconnect is very, very short range interconnect inside the data centers. It's a completely new application. So I think it has a lot of -- it's only in its infancy, actually. This is a product that didn't exist a year ago. We won the product initially in our Israel NPI center, so it's kind of a testament to our strategy of having on-ramp NPI centers close to customers, and then we transferred it a couple of quarters ago to Bangkok. Initially, when we introduced the product, because, again, customers are quite secretive, we weren't even clear what the application was when we initially introduced the product. It only became clear as we ramped it. But I think that category, if you like, of AI-specific data center interconnect products. I think it's only in its infancy. It's only starting. So we think there are other -- certainly other companies we're talking with and working with to introduce other products that do a similar job inside the data centers.

I'm just curious on that front because we've seen different companies talk about sort of their AI products that they're doing. How do you get to the sort of end point of realizing its AI use case? Is it based on the speed? Is it based on some of the requirements that these products have? Like because different companies seem to be using different classifications of what they end up calling as AI sort of use case and some might be just doing it based on speed, et cetera. How you're doing that? And then when you're sort of doing that with this particular customer, are you seeing then more customers come in with similar requirements?

We are. So this particular product, we realized it was an AI application when the customer told us. I mean, we had an inkling, I suppose, at the start. But then when the customer told us what the application was, it became clear to us. So yes, we are seeing other customers looking to introduce products to, if you like, compete with that particular product. But -- and we think because of our reputation, thankfully, most of those companies would come to us for these complex products -- to make these complex products for them. So we're happy to produce for everybody. So yes, it's -- there are differences in how the product goes together and what the product does that make it, if you like, unique to AI applications, especially the very short reach, short range again, low latency, low power high-speed application and 800 gig seems to be the order of the day for these data center applications.

I mean we've asked the other companies this as part of more of the upfront questions, but maybe this is the right time to ask you just in terms of AI broadly, when you think about the implications for your business, you outlined sort of the new products that you're doing on that front. How do you characterize the opportunity? Is it more of a strong investment cycle and a volume opportunity for you? Or as you sort of see these products being very early stage at this point, you are probably one of the few contract manufacturers in the world that can do them. Is it more of a content opportunity, a margin opportunity for you to start with and then the volume sort of kick takes over later, how should we think about that?

Well, I think we're trying to figure that out, I suppose, as we go along. Again, if you go back a year ago, we weren't even clear what the implications of AI would be for us. Now that's becoming more clear. Again, I think these unique interconnect products, they are quite specific and unique to AI. So I think, if you like, it's a new category of interconnect products that we'll be looking to make sure we're the leader on in terms of manufacturing those products for our customers. So there's no -- in terms of how the product goes together, it's not particularly different from a regular transceiver-type product, and the optics content is quite similar. And we're -- again, we're very well positioned to bring these products to volume and to a high yield quickly, so I think we're a good choice. I think we are uniquely positioned as a contract manufacturer to kind of capitalize on those new applications that come along. We'll be looking to make sure we capture as much as we can.

Okay. Maybe just sort of then going back to some of the legacy products, what's happening on pricing relative to 400 gig and 100 gig, as 800 gig already sort of in the market, at least a customer is using it and then more customers sort of start to look at 800 gig as an option. What are you seeing on the pricing dynamics on 400 gig and 100 gig? And maybe a second part to that is, then when you think about now going from 400 gig to 800 gig in sort of -- with a few of your customers, does that cycle look different from when you went from 100 gig to 400 gig?

I think so far, the 800-gig application for us is, if you like, a different application to the traditional transition that we see going from, let's say, historically, we went from 25 gig to 100 gig, now we're in the middle of transitioning from 100 gig to 400 gig, the 800 gig is, as we said, a different application. But right now, we're, if you like, ramping 400 gig, 100 gig will begin to taper off. The volume will begin to decline on 100 gig, which is normal. That's normally what happens is the new product comes along, as it ramps, there's a kind of a race between price and volume, as the 400-gig volume goes up, the price will need to come down to justify the higher volumes. So you'll get that ramp of 400 gig, the volume will go up, the price will come down. And then the 100-gig volume will begin to taper off as time goes along. But we've participated in these changeover cycles, many times before, and we've tended to do well when they occur. And I would say we're just in the early stages right now of ramping 400 gig.

Okay. Relative to 400 gig as the price -- as the volume scale and pricing comes down, how much of an impact is that to the contract manufacturer like Fabrinet versus something that's absorbed by the OEMs themselves?

Generally, it's on a per unit basis, the unit price comes down. But in aggregate, if you look at the overall contribution for us, the revenue increases and the margin percentage stays about the same. So therefore, the margin contribution dollars increases as the revenue grows. We -- the product is designed by the customer. The components are specified by the customer. So the kind of the levers to pull, if you like, are yield, manufacturing cost, and then material cost is a big one. The material cost is typically 80% of the total cost as material. So we typically work closely with the customers to make sure we're getting the best possible material cost, but we usually drive the price down by driving the cost down. It doesn't come out of our margins. It comes out of the cost.

Okay. Okay. Let me just pause and see if any questions from the audience. Just wait for the mic, please.