Afternoon, welcome to Fabrinet's first investor Q&A session at OFC 2026. I'm Garo Toomajanian, head of investor relations at Fabrinet. We are joined today by Fabrinet's chairman and CEO, Seamus Grady, and Fabrinet's CFO, Csaba Sverha. We're also joined by Samik Chatterjee of JP Morgan, who will be kicking things off for us. Before we begin, let me inform you that we will be making forward-looking statements during our presentation that are based on assumptions and beliefs as of today. These statements are subject to known and unknown risks and uncertainties that could cause actual results to differ materially from those projected or implied today, in particular, those described in the Risk Factors section of our most recent 10-Q filed with the SEC. You should not rely on our forward-looking statements as predictions of future events.
We undertake no obligation to revise these statements in light of new information or future events, except as required by law. This session is being webcast and a replay of the session will be available after the completion on Fabrinet's IR website at investor.fabrinet.com. Now let me turn things over to Samik to kick things off, and then we'll be taking questions from the audience.
Thank you. Thanks to the Fabrinet team for the opportunity here. I'll kick it off with a couple of questions and we do want to make this interactive. When you open it up, please ask your questions. Seamus, Csaba, thank you. Thank you for the time. Maybe I'll start off with a much more broader question. A question that comes up often from investors is sustainability of your growth rate. You reported 36% year-over-year growth rate in the last quarter. Obviously, a lot of discussions about sort of what's driving that, but sustainability given, I think partly because you also guide one quarter at a time.
Yes.
There's always sort of questions about sustainability of that growth. Maybe share your thoughts around that. The drivers of the growth, how sustainable do you think those drivers are? How should we think about that?
Yeah. We think they're sustainable. Yeah. This is probably a time that we wish we guided further out than one quarter at a time, but we guide one quarter at a time. Our growth, you know, if you I think the best predictor of the future is sometimes, you know, what have we demonstrated in the past. If you look over a 10-year period, our compound annual growth rate is 16% over a 10-year period. In FY 2025, we grew 19%, and last quarter we grew 36%. This quarter, at the midpoint of the guidance, we will be up 35%. It's you know, the growth is quite strong and we believe sustainable. There's a few drivers for that.
Our telecom business is probably the largest category that saw growth up 59% from a year ago last quarter. Largely driven by DCI growth, which was very strong for us. Also non-DCI growth in our systems business, which is growing very strongly for us. Also, high performance compute was a very strong category for us. High performance compute is a new category for us. We haven't had that category in the past. We grew from $15 million in Q1 to $86 million in Q2. December was our Q2. A very strong growth and that our main customer there is AWS, and we're focused on, you know, continuing to grow that business and grow it by performing and executing and doing a good job for our customers.
They're probably the two biggest areas of growth. Our Datacom business was flat, largely driven by supply constraints of an EML. Aside from that, the demand remains quite strong actually in Datacom, even though our revenue was flat. You know, several growth drivers. We have several other ones as well. I'm sure you'll hear about lots of them here at the show. We're certainly very excited, I think, about the prospects in front of us. They're on several fronts. You know, we're getting visibility, I would say, from customers that's multi-year visibility. That doesn't mean we're getting multi-year orders, but we're getting visibility.
The customers are sharing with us their demand and their plans over the next few years, which is really helpful for us as it allows us to make sure we have capacity lined up to support the customer's demand. You know, it seems to be sustainable. Certainly very robust. The demand is very robust and it seems to be sustainable.
Correct. Maybe just getting into a bit more details in terms of the last quarter. Your DCI growth sequentially was much smaller than what we've been used to. Anything going on specifically there that you would call out? Again, how are you thinking about sort of that rebounding and the sustainability of strong growth on the DCI side?
Yeah. DCI, we've started to break out DCI as a separate. It's in our telecom number, but we felt it would be more helpful for people such as the people in the room here, if we broke out DCI separately. Because while it's in the telecom number, it's obviously driven by what's going on in the Datacom world. The growth for us there has been just tremendous. It's mostly, not exclusively, mostly 400ZR and 800ZR. We have significant share in that business, we think, as a contract manufacturer. You know, we support several customers, but you know, three of them seem to have the majority of the market share, and we're supporting them.
You know, we're the sole source for all of the companies that we supply in ZR and, you know, we certainly think we have a majority of the share there. You know, year-over-year growth has been great, while quarter-over-quarter growth was a little bit flat, primarily because of product transitions. As customers transition from, let's say, 400ZR to 800ZR, you know, the new product hasn't fully ramped yet. The previous product is beginning to taper off, so that was really a one quarter. I would call it an aberration as the customers go through a product transition. Overall, you know, DCI has been exceptionally strong for us.
We think it's a really good solution to the power problem that, you know, as the hyperscalers roll out these distributed clusters, DCI or ZR is a really good solution to allow them to maximize their output.
Got it. Good. Okay. Let me open it up, please. Anyone with a question, just raise your hand, and Garo has a mic. Yep.
Can you please state your name and firm name?
Sure. Kate with Balyasny . Maybe I'll just open it up with, you know, people are talking a lot about the new growth drivers as we look forward to two or three years in OCS and CPO. Can you talk about your place in these new areas of growth?
Our place will be as a contract manufacturer. You know, we don't participate in the ODM market. We don't have our own products, and we never will. I would call us a pure play contract manufacturer. We're very adamant about that we won't be in the product business, which is a little bit different to some of our competitors, but that's the path that we've chosen. Our role will be as a contract manufacturer. We think we're the leading contract manufacturer in the space. For us, what we really look at, whether it's CPO or OCS or whatever the newer technologies might be, what we look at is what are the underlying technologies, the manufacturing process technologies that underpin these processes.
You know, we try to make sure working a few years ahead of the products becoming mainstream to make sure that we're putting the process capabilities in place. We've been working, you know, diligently on that for the last several years, and we really feel we have the capabilities to be the leader in the contract manufacturing space, supporting the leading companies producing those products. We're pretty excited about both. We think they both, you know, represent new opportunities for us, but also really strong growth. The demand seems to be very strong. OCS, certainly, there's very strong demand for it. Co-packaged optics, it is newer, but, you know, it has NVIDIA very much behind it. So we think they're both. We're very excited about both.
We think they're really good opportunities for us.
Christian Say from Adage.
Hi.
Could you specify maybe what are some of those process capabilities that you're investing in and growing, you know, for these new technologies?
I could, but I won't.
Okay.
Thank you. You'll have to come and work for us to find that out. Because sometimes we're developing the process technologies as the customer is designing the products. We, you know, work very well with our customers when they develop a new product or a new concept. Sometimes we're working two or three years ahead of it ever being discussed here at, you know, OFC. We're working on those technologies in the background quietly to build the capability and to make sure we have not just the capability, but also the capability at scale so that we can produce it in volume with a, you know, a predictable yield that allows us to be cost competitive.
Are there some you can talk about that are existing that you've developed, increased capabilities that you're planning?
You know, several of the products we make for our customers today, we do a lot more than, I would say, most contract manufacturers. We do a lot of packaging at the wafer level and at the die level, which is quite unusual for a contract manufacturer, I think. By way of example, if you look at our manufacturing space, 70% of our manufacturing space is clean room space, which is quite unusual for a contract manufacturer. Most contract manufacturers have probably of the order of 10%, something like that, clean room space. We have 70% clean room space. We do a lot more, I would say, packaging, precision packaging for our customers than most other contract manufacturers.
They're, you know, those capabilities we've developed over several years. We've been doing it for well over a decade. When, you know, as the industry begins to scale, if you go back nine, 10 years ago, there was insufficient volume in optical for it to be attractive for the packaging companies. Now, of course, everybody wants to do it now, but we've been doing it for a long time. Really that capability that we have, we feel is pretty unique, the packaging capability that we provide for our customers. We, again, we do more than just product assembly. We do the packaging, and we often make the components that go into the product.
that when we're building the finished product, you know, we're producing in usually in a handful of components, maybe, you know, upwards of 60%-70% of the BOM content we're producing in-house. It means we're a valuable supplier to our customers, but it also means, you know, it makes the business quite sticky. We're also able to help the customer eliminate the margin stack that they would have to pay if they were using four or five suppliers to do what we can do. It's a very compelling proposition for the customers and also for us.
Hi, Ravi Patna, Hawk Ridge. Your customers are pretty open about their, like, margins and how they're taking a price, and you see it in their sort of financial performance. Just curious if you've reconsidered how you price to those customers to sort of take part in this, you know, margin upswing we're seeing across the board.
Well, I mean, you know, being a contract manufacturer is quite different to being a product company. When you're a product company, it's really about supply and demand. When things are in short supply, you can increase your prices. We can't really do that. If anything, we do things the other way around. As we grow the business with the customers, our pricing typically improves, so we want to make it more attractive for them to do business with us. Yeah, I'm familiar with what you're referring to, but it's not something that we will be doing. You know, we're very competitive. We want to remain very competitive.
As we grow the business, we'll be, you know, using that increase in business and improved efficiency to make sure that our prices are even more competitive for our customers. That's a bit counterintuitive, but that's how you do it as a contract manufacturer. If you start to, you know, increase prices as a contract manufacturer just because you can, you'll get punished by the customers. We tend to be very competitive and make sure we stay that way.
Hi, Seamus. Thank you for hosting this. Chris Rolland, Susquehanna. Yeah, just back to the OCS opportunity. It seems like market projections are doubling for the TAM every quarter or so. You know, some of the key guys there are talking about engaging more and more contract manufacturers, which I think you're in a very good position to address a lot of that. So I guess first of all, could you talk about maybe how some of the economics work here? Like, do they do all of the equipment to put this stuff together? You know, fund all of the CapEx and then just hire you as a contract manufacturer? How do those economics work? Then, if you could talk about, you know, where you would put this capacity. Do you have the capacity to address what seems like an ever-growing market?
The economics, especially around the CapEx, it really depends on a case-by-case basis by customer and in some cases by product. In a general sense, you know, we obviously build the factories and we are responsible for the CapEx of the, if you like, standard equipment or generic equipment that can be used across any customer. We typically ask the customer to pay for any unique equipment. If it's product specific or unique equipment, such as test equipment and the like, we ask the customer to pay for it. You know, I don't think that's dramatically different to any other contract manufacturer. It's just that in our case, there seems to be a lot more customized and unique equipment because of what we do. You know, how we're adding capacity.
If you look at our footprint today, our run rate today is about $4.6 billion-$4.8 billion, something like that. Our capacity is about $5.5 billion. That includes. There's a little bit of space we're converting in our Pinehurst campus that's almost finished. I'm including that. Our capacity today is about $5.5 billion. We're building our new building in Chonburi, Building 10, which when it's finished, will be 2 million sq ft with capacity for about an additional $3 billion. That's coming in two phases.
We're opening about 250,000 sq ft of that in July of this year, and then the balance will be open by the end of the year. Between our current capacity and what we're building in Building 10, that will get us up to a capacity of about $8.5 billion. We can build another two factories on the land that we own in Chonburi, each with capacity for about an additional $1.5 billion each. $8.5 billion + $3 billion would be $11.5 billion. We have ample capacity for the next several years. We're also looking for additional land, and we buy it if we can find suitable land that's close to our current campus.
Really, we're, you know, we're doing everything we can to position the company for not just the growth that's in front of us for the next several years, but also to position the company for the next kind of 15, 20 years. You know, we feel Thailand is a very good place to do business. It's been a very successful location for us, and we plan to continue to expand our business there. Ample capacity to grow.
How quickly could you bring that up for OCS in particular?
Well, I mean, if you look at Building 10, you know, like I said, 250,000 sq ft of it, you know, which would give us capacity for $350 million-$400 million of revenue. The balance of Building 10 would give us capacity for another $2.65 billion. Three billion between all of Building 10. You know, certainly that could go into Building 10. I mean, it's I would say because of the visibility we're getting from customers, we're quite comfortable that we're able to add capacity ahead of the customer's needs, and that we keep ahead. Our role is to keep ahead of the customers. Our mission is to make sure we keep ahead of the demand coming from the customers.
You know, the business that we're in is we're a service company. We're not a product company. You know, unlike a lot of the other players at the exhibition here, we're a service company. One of the most important services we provide to the customers is that flexibility, that upside flexibility that when they know they have a big demand coming, they can rely on us to ramp. 'Cause this is all we do. All we do is manufacturing. We don't have our own products, so we don't have any distractions. You know, we're very focused on making sure we keep ahead of the customer's need. We'll be doing that over the next several years.
Seamus, if I can just follow up on that.
Sure.
You've talked about this sort of incremental capacity coming online. Eight, up to $8.5 billion of revenue capacity that you'll have. As we now look forward with some of these new opportunities, and how should we think about in the sort of segments that you report, what are the major growth drivers? Like, what are the primary growth drivers to get you to that level? Is OCS one of them? Is sort of HPC one of them? Like, how should we think about what are the major growth drivers versus maybe what takes a bit of a back seat?
I think I don't think anything takes a back seat. I think they're all, you know, they're all very important to us. They're all very positive growth drivers. If you look at the kind of growth drivers in our business. Obviously, our telecom business is very strong. We think that's going to continue both in terms of the system business growth, but also the DCI business. We think telecom's gonna be very strong. At Datacom, you know, the business has been, I would say, flat because we've had one customer, our major customer has been constrained because of laser supply. We're working to secure a few additional sources of growth, one being we call them merchant transceiver manufacturers or other OEMs who make transceivers and also hyperscale direct.
If the hyperscalers, you know, have their own design that they want us to produce, we're happy to do that. We're working on both of those. The newer opportunities and technologies, you know, I've learned over the years that the new technologies, they rarely appear overnight. They take time. You know, CPO will take time, OCS will take time. But we're very excited about them because, again, we believe we have unique capabilities in the contract manufacturing industry to support those opportunities. You know, we think OCS represents a really positive development, so does Co-packaged Optics. High-performance compute, again, for us, a new category.
We didn't have high-performance computing as a category, you know, a couple of quarters ago, and then we had $15 million in Q1 and $85 million or $86 million in Q2, and it's continuing to grow. We think that that's just a starting point. We think there's other opportunities to grow that business with AWS, but also with other hyperscalers. Really, all elements of the business seem to be quite strong right now and the demand looks to be very robust for a sustainable period of time.
Okay. Any other questions? Garo behind you.
Behind you, Garo.
Hi, Pierre at Balyasny. How are you?
Hi.
Regarding the HPC vertical that, you know, you were just referencing, which has been very successful, how should we be thinking about your existing customer and, you know, ability for you to, you know, effectively launch new products with that same customer versus, new you know, ramping new customers? What would you be doing for these new customers?
With our current customer, we're a contract manufacturer. They use ODMs typically. We're not an ODM. We don't have our own products. Where the customer owns the design and they need someone to produce it, we're happy to produce it. That's what we're doing, is we're producing as a contract manufacturer. And that's the model that we would employ with both, in terms of growing the business with this customer and also with other hyperscale customers into high-performance compute. High-performance compute products, they, as time goes by, are going to need more and more networking and interconnect in addition to the compute business. We think it's a very good fit for us.
They're complex products, they're difficult to produce, and we think they're a very good fit for our capabilities. We're pretty excited about HPC as a category, and we think there's a lot of opportunity there for us.
Maybe I'll follow up on that.
Sure.
You talked about additional opportunities with other hyperscalers on that front, and maybe a two-part question. Generally, how you've talked about the primary customer on the HPC side, we sort of had you getting to about $200 million a quarter roughly around a certain point of time. Is that sort of still your expectation, or have you seen sort of your pool or your share expand with the customer beyond that? As you're looking and talking to other hyperscalers, generally, is the size of the opportunity similar, or are you seeing sort of bigger or smaller opportunities as you talk to the hyperscalers on the HPC side specifically?
I think for our you know our current customer which is AWS we think it will continue to grow and the you know the magnitude you talked about is probably about right. It won't grow in a straight line because when the customer launches new products you know the growth doesn't happen in a straight line. Right now we're in the middle of ramping a new product. New products are great because they're the future and you know once you have the new product you have it for the next several years. But in the quarter in which you introduce a new product you always have to kind of tolerate that volatility where the old product tapers off the new product gets launched.
Notwithstanding that, yeah, we think that that business has a lot of potential and can certainly ramp to a much higher number than we've had in the past. You know, other opportunities, potential hyperscale customers, it's early days yet. We haven't really engaged that deeply with the other opportunities, so I think we have, you know, we have a little bit of work to do to get to the point where we have other non-AWS HPC customers, but we're working on that.
Okay. On the transceiver opportunity with the customer as well, like, when we start to think about sizing it, and as you sort of go through your discussions with them, is the opportunity similar to what your HPC business with them will look like, or is it potentially larger? How do you think about sort of being single source versus multi-sourced on that front as well?
We think that, you know, let's say transceiver direct business has a lot of potential because the volumes that some of these, you know, hyperscale customers are sourcing are just huge. If they were to have their own product, again, very important, we won't have a product. It's not our product. We're just a contract manufacturer. If they have their own product that we can produce for them, it's really a case of them deciding what percentage of the demand they allocate to their own design. In terms of potential, it's huge. You know, even if they were just to allocate, you know, 10%, 15% of the total volume to their own product, it would be very significant.
I think it has a lot of potential, but at the same time, it, you know, it has to work and it has to operate in their network and be compatible with everything else they have going on in their network. There's a lot of, you know, interoperability testing that has to be done before we're at the point where we're ready to ramp that.
Okay. Any questions?
It's Tim Savageaux here from Northland.
Hi, Tim.
Hey, afternoon. Well, your once and potentially future 10% customer at Lumentum talked about engaging with seven contract manufacturers this morning, which is, you know, a decently large number. I wonder if you have any thoughts on that from a competitive standpoint. Then more specifically and awkwardly about the OCS opportunity, and I think you've made that wavelength-selective switch that the whole platform's based on. This should position you fairly well for that opportunity. Thanks.
Yeah, we think so. I think if you look at their OCS product, you're exactly right. The WSS technology that kind of underpins that is something we've been doing for a very long time. You know, I believe even though they had moved some business in-house in the past, I think we're still their biggest contract manufacturer, and we're really looking forward to growing that relationship. I think, you know, our relationship with Lumentum is, I'd say it's the best it's ever been. It was never bad, even when they were moving business back in-house, we always had a great relationship with them. We just didn't agree with what they were doing, but that was their decision, not ours.
You know, I think there's a lot of opportunity there with Lumentum and really there's a lot of synergies between the two companies. You know, we're both in Bangkok. There's a long and very successful history there. Like I say, the relationship is very good at every level, you know, at the working level, but also at the most senior levels within our two companies. We think there's a lot of kind of strategic alignment there, and we'll be looking to expand our relationship and our business with Lumentum over the next few years. Every piece of business we have, we have to earn it. You know, nothing comes because of a great relationship. We have to earn the business.
We're excited about that. I would say we're excited about that. OCS in particular looks like a really, you know, compelling product. We'd love to be in that supply chain. If you're talking to Michael, please put in a good word for us. I forgot to wish everyone a happy St. Patrick's Day, by the way.
Any other questions?
Hi. Pankaj from Fidelity. Two questions. One on the AWS. The $200 million a quarter run rate, is that sort of fully ramped beyond which growth is based on your volume growth and your customers? The second part is just maybe an update on how long do you see the supply chain constraints lasting for? Are they getting better or are they getting worse, on the laser front?
Yeah. On the AWS high-performance compute business, you know, it could be less than $200 million, it could be more than $200 million. It really depends on, you know, we have to continue to perform. So far we've done a very good job, we believe or we've been told by the customer they're very happy with what we're doing. We leave it to the customer to decide, you know, how they split the share between us and the other supplier. You know, like I said, there's no guarantees, but as you may know, we do have a warrant that we signed with AWS, so they have an incentive to give us more business. The more business they give us, the better.
Well, obviously, the better for us, but also the better for AWS. But again, we have to earn that business, and we're very focused on just executing. Execution is the best strategy. Your second question?
Supply chain constraints.
Supply chain constraints. Yeah, I mean, the biggest supply chain constraint we've been dealing with is on the EML side, and specifically the 200 Gb per lane EML constraint that we've had with our main customer, NVIDIA. It is improving, you know, and the products that we make for NVIDIA, they have approved a second source now for the laser, which is Lumentum. That's just beginning to get going this quarter. I think that's probably another. You know, we may see an improvement in supply in the June quarter, but it's more likely probably out into the September quarter 'cause it just takes time to get everything ramped up. It is improving, and the demand is still very robust.
I mean, the demand hasn't gone away. If anything, the demand has been increasing, actually. That's I think the big ramp for us on 200 Gb per lane. You know, 1.6 Tb and 800 Gb is still largely in front of us. [inaudible] .
Hey, George Notter at Wolfe Research. Thanks very much. I guess I was just curious about just the bigger picture in the environment. It seems like everyone's got this just gigantic, you know, imperative to scale as fast as they possibly can.
You've got a lot of companies that have, you know, historically been doing internal manufacturing in the space. It just seems like everybody needs help, and it seems like in many respects you're the answer. Do you see that in terms of your conversations with people around the industry, the pipeline that's developing around your business? Like, give us a sense for sort of the tenor of conversations you're having with folks, and is my assertion the right assertion? Thanks.
We do see that. You know, we've spent decades developing the capability. When optical wasn't sexy and it was low volume, we were very committed to it. We've spent a long time building up the capability. It's hard to, you know, for competitors, it's difficult to compete with that expertise that we've developed. You know, the tone of the conversations with customers is very interesting. You know, for me, I've been in the industry a long time, and usually the biggest struggle you have in contract manufacturing is, you know, figuring out where the demand is gonna come from. You know, that could be next quarter or next year, but right now it's the opposite. The customers are giving us, you know, two and three years visibility.
I don't. That's not to say they're giving us two and three years orders, but they're giving us visibility. We're able to see. Okay, what's coming down the track over the next few years? What? They're asking us to help them and put the capacity in place, which we're very happy to do. We're in the very fortunate position because of how we kinda capitalize the business and how we build out our capacity. It's a very low risk decision for us to continue to add capacity. Let me give you an example of the economics of Building 10, for example. I mentioned that Building 10, it'll be 2 million sq ft when it's fully built. It'll have capacity for
It depends on the mix, and the mix is actually going in our favor over the last while, but call it $3 billion of revenue capacity when we build Building 10. The CapEx is about $130 million, maybe $132 million, depending on what the FX is doing on any given day, but call it $130 million. Well, at full run rate, you know, about five months worth of operating margin pays for the factory. The upside potential is gigantic. The downside risk, again, because of the way we capitalize, we have a very strong balance sheet. We build, you know, we build these factories using our own cash. We don't have any debt.
If something were to happen that the industry were to decide that it doesn't need Building 10, and we had to sit on it, the gross margin headwind would be 15 basis points, which is. It's not nothing, but it's negligible. The downside risk is very small, and the upside potential is tremendous. We're very conscious of, yes, we want to, of course, capture the business, capitalize on the upside, but, you know, we've also been around long enough to know that sometimes there's a downside risk, and we don't wanna get caught with the downside risk. How we capitalize, you know, and plan for the growth is really important. The second thing that's very important is we keep our costs under very tight control.
You know, OpEx, our OpEx is 1.5% of revenue, which is very small. Even for contract manufacturers, it's very small. Our fixed cost runs at about 5% of revenue. 5% fixed cost, 1.5% OpEx. You know, we have, we believe, a very you know successful financial model that allows us to again capitalize on the upside, capture the business that's there, execute very well. If something were to happen on the downside, you know, we feel kind of we're not overstretched. We'll never overextend ourselves. We'll never overstretch ourselves because you know we've all been around long enough to know what happens if you do that. We feel quite good about those prospects.
You know, the tone of the conversations with customers is very encouraging, you know? It's very heartening for me to hear customers tell us how happy they are with the job that our team does, with the execution on the business that we have with them today. You know, when you're growing at 36% year-on-year, you know, sometimes, you know, growth is not always easy. It doesn't always happen in a straight line, but it's really encouraging for me to talk to, you know, fellow CEOs and for them to be very complimentary about the work we're doing and the job we're doing executing. That's the most important thing. All the strategy in the world doesn't matter if you're not executing.
We're executing very well for the customers. We're, you know, we're saving them a lot of money, and making a little bit ourselves in the process. It's a winning formula, we think.
Hi, Jeff Hopson from Needham.
Hi.
You talked a little bit about multi-year forecasts from customers, and maybe the typical order visibility is much shorter than that.
Yes.
Maybe a quarter or two. Are you seeing that change for people putting in orders in that timeline extending at all as people lock in allocation or supply?
You know, really the customer orders in our business is really governed by whatever is the longest lead time component. Typically we'll have. It depends on the customer, but we might have 13 weeks rolling orders, and then we may place orders for longer lead time components outside of that, but not for everything. If there's one device that has a six-month lead time, you know, that doesn't mean we get orders for the finished product out six months. We just get, you know, the, let's say three months rolling of demand and we'll, of course, have forecast beyond that we'll use to position inventory.
The visibility we're getting is very encouraging, you know, even if it's. It's a little bit different in our case because, again, we don't have a foundry, so we're not having to ramp up, you know, huge foundries or anything like that. We are getting very good visibility, and I would say that firm period is, you know, it's probably stronger than it's been in a long time, as customers commit to the future. Sorry. Get to you next time.
Hi, Joseph Santos from Putnam.
Hi.
Just another way to kinda frame visibility, and the way customers might be committing to you or changing the way they're committing to you, given their forecasts for their businesses. As you think about ramping up Building 10, are you seeing a different cadence of kind of commitment from them to try to take up that capacity and invest in that facility?
Uh, I-
Earlier than you normally would.
I would say, you know, the customers know how we operate. We tend to be, you know, quite flexible with the customers. You know, they give us an indication of what they need. We make sure we have the capacity in place. We're not gonna put a gun to the customer's head and say, "Well, you told us you'd need 300,000 sq ft. You only need 200,000 sq ft. We, we'll send you a bill." We're much more flexible than that. Usually, we're working all the time. With the best will in the world, we'll size what we think the customer wants, then we start to produce, and if they need more, we'll give them more. If they need less, we'll cut back on the space, and they'll save money.
It's a bit. It's quite fluid actually, how we do it day to day, and we're adjusting all the time. You know, it's a kind of a self, I would say, a self-governing process because if the customer wants more space just in case they might need it, well, it's gonna be expensive because we have to charge them for that space, and that space we could be using to produce something else. It kind of becomes kind of self-limiting. You know, the customer, we'll allocate whatever space they need, but they then pay for that space. It also works really well from an IP protection point of view. Each customer is physically separated from the next, so there's no overspill of IP between customers.
Cory Johnson from Epistrophy.
Cory, how are you?
I'm not asking for your 2030 guidance, but it was surprising this morning when Lumentum talked about this thing's really gonna take off in 2030. They kind of talked about growing capacity and growing into a next stage of growth. I wonder if in your crystal ball, when you try to imagine where we are in five years, I mean, are we still gonna be building data centers at this pace, or is it at an increasing pace?
2030, that's a long way in the future.
It's a long way.
It's a long way. You know...
For the grandkids listening.
I think it's, you know, again, the caveat, we're just contract manufacturing guys, so. You know, it seems to me, if you look at the massive expansion explosion in compute that's been going on for the last couple of years, it almost feels like the industry forgot about the networking and the interconnect. Well, all this compute has to be attached together, has to be connected together. The data centers have to be built in a distributed way because of the power envelopes that they have to operate within. The best way you can manage that is with DCI, 400ZR, 800ZR. There's massive demand for that. Then interconnect generally to connect all the compute together.
It almost feels like there's a catch-up going on, that the optical industry, the component industry, the optical industry hasn't really been ready for these kind of volumes. There is a kind of a transformation going on, I think, in the industry that it is, you know, we are getting this transition to photonics and optics. Certainly as far ahead as we can see, it looks to be very strong and very robust. I think the thing that's different this time around versus the dreaded that which we shall not mention, the dot-com, is, you know, these are real companies. These are real companies. There's no, you know, fly-by-night companies. These are all very real, very credible companies, very well-capitalized.
The supply base, including ourselves and the other companies in the industry who are customers of ours, you know, they have sometimes scars on their back. So they're careful about how they expand and how they add capacity. If you look at Lumentum and Coherent, the $2 billion investment that they're each getting from NVIDIA, I think that's excellent. I think that's really good for those companies, and it's really good for the industry. Hopefully, it'll be good for us as well because we supply both Coherent and Lumentum.
I'm probably not answering your question, but certainly we see the demand being very robust for a very long time, and it's going to take the industry a long time to catch up and fulfill all this demand, especially at the pace at which the compute performance is growing. I mean, it's just staggering, the compute power, and all the networking and interconnect that's going to be needed to support that. We believe it's sustainable. Now, 2030, I don't know. We'll see. It's not that far away, 2030, three and a half years.
Well, Seamus, one of the questions I've got on that front from investors is Coherent, Lumentum, Fabrinet all are pretty strong partners to NVIDIA.
Yes.
Investors have been asking, would NVIDIA eventually have to invest in Fabrinet as well to help you with capacity? What are the sort of differences between sort of how you see the investment with Coherent and Lumentum versus your business model, and why or why not?
I mean, we would not refuse an investment. No, I'm joking. We're, you know, I think our business is a little bit different. If you look at Coherent and Lumentum, you know, for them to put fab capacity in place, it's a much bigger, you know, capital outlay than, you know. We build factories, and then we build other people's products in those factories. That's essentially what we do. A lot of the CapEx that goes into the factory, some we fund, some the customer funds. You know, not to be running down what we do, but I think what Lumentum and Coherent do is very different because they're developing products which have, you know, R&D cycles are very long. Then to properly capitalize those in their foundries takes a very long time and a huge amount of capital.
You know, to me, it makes sense in the case of Coherent and Lumentum. I'm not sure, I don't think it makes sense in the case of Fabrinet, but if you want to ask Jensen to send over a check, we'll take it.
I think George.
All right, thanks. Do you have a guess as to what the mix of your business comes from cloud providers? Obviously, they're not always the end customer, but what do you think that mix would be just in your mind's eye?
It's a good question. I think I would say probably the vast majority of it. You know, it seems like it's not that long ago that we used to get asked questions at meetings like this of did we see AT&T's capital budget for next year, and what do we think about it? Now that seems like a kind of a funny question almost. I mean, you know, obviously everything in the data center, in the Datacom spend is driven by the hyperscalers. A lot of what's going into our telecom spend is being driven by hyperscalers. I don't know what the exact percentage is, but it's a big number.
Got it. Okay.
It's the vast majority of it, I think.
Anything, do you see anything new or different or changing in terms of, you know, inventories and your relationships with customers in terms of procuring components, putting it on your balance sheet, putting it on the customer's balance sheets? Like anything changing there with all this new demand? Thanks.
Not really. I mean, the way we operate with the customers in terms of who carries the inventory, that's pretty well determined. I mean, sometimes we get asked from time to time by maybe newer customers if we'll carry more inventory. I mean, we do whatever the customer wants, but as we always say, we're an excellent manufacturer, but we're an absolutely terrible bank. Our inventory carrying charges are generally terrible. They're not competitive. We're not a good bank. It wouldn't be an efficient way for the customer to carry inventory, to have it on our balance sheet. We tend to stay away from that. The other thing we do, sometimes the customer will have a very expensive component, and we'll allow them to consign it.
You know, sometimes it's important to the customer that, you know, their contract manufacturer doesn't take too big of a margin on a high-value component. Sometimes we say to the customer, "Okay, you can just consign it," and we'll pass that through. We're very flexible. You know, we'll do whatever the customer really wants us to do. That's one of the services we provide, flexibility.
Well, Seamus, maybe if I can ask a question on CPO. If you can just clarify sort of how we should think about what's the role Fabrinet has and maybe compare it to when initially in some of the remarks you made, you said you tend to sort of in-source 60%-70% of the BOM. Particularly with these sort of newer solutions on the CPO side, where is that tracking initially? And then where do you see the opportunity beyond NVIDIA? Like, are you seeing more customers in that pipeline engaging with you on CPO?
Yeah. We have three customer engagements on CPO, one of which is NVIDIA. You know, CPO, I think if you go back, I don't know, seven or eight years ago, CPO was two to three years away. If you go back five years ago, it was 18 months to two years if you go back three years ago, it was one to two years away. now it's like a year to 18 months away or maybe six months to a year away. It's still new and everything has to line up, and everything has to operate and work. We're very excited about CPO because again, the technology that underpins it is incredibly difficult to do.
That's really where we're focused is making sure we have the capabilities to support what the customer needs from us. You know, how big could CPO be? I think it depends on what we do in CPO, you know. If you like, our role historically has been we haven't really been in the, let's say, the switch business with NVIDIA. We've been their transceiver manufacturer. I think it really depends on, you know, where we can grow that business, whether it's just in the pure CPO packaging space and maybe LPOs or if there's a role for us outside of that.
When like we all were in presentations all today and most companies like Coherent and Lumentum are talking about material CPO revenue second half of this year and ramping for then to second half 2027. It doesn't sound like that's sort of in your order book today from what you have visibility into. That's not necessarily where NVIDIA's orders have sort of already come in given that you obviously have much shorter order duration as well.
Yeah, I think that'd be fair. I mean, you know, we don't guide beyond one quarter at a time and we're gonna continue that, although there's times I wish we did guide more than one quarter. Yeah, I guess we'll talk about that next quarter. You know, we generally don't guide by category either. We talk about whenever we have CPO revenue, we talk about it looking back rather than trying to guide it going forward, because it's very difficult for us to guide that. Go ahead.
Oh, maybe just one for sharp. Okay. Maybe we'll end with that one if you're okay. Garo.
Sure.
Thanks. Peter Law, Analog Management. A follow-up to Samik's question and your answer. A few years ago, when you moved from making just components for telecom to making systems in telecom, there's sort of an order of magnitude jump in the BOM that you could address.
Yes. Yeah.
Is CPO a similar opportunity where you're moving from just components and Datacom to system?
It is if we're able to move up the stack, which is really what we do. It sounds easy, but it's not easy. We focus down deep in the stack. We do the packaging, you know, the difficult stuff, and then we move up. CPO would be similar if we're successful, if we're able to move up to the switching part of CPO.
How would you be successful to move up?
By-
What do you have to do?
By executing on what we're being asked to do in the beginning. That's the best. Again, the best strategy is execution in our business. We find if you execute, you know, in a really excellent way for the customer, and we're very predictable in terms of delivery quality, you know, really fast responsiveness, great NPI services, and we do all that at a competitive cost, the business tends to grow. That's really our formula is just execute really well.
Could you see yourself going?
Do you wanna get the microphone?
Yeah. Sorry.
No problem.
Camron Bagheri from Holocene Advisors. Just following up on that question, could you potentially see yourself integrating in terms of like, for example, a CPO switch? Would you ever, you know, go as high as doing like the whole switch? Manufacturing the whole like networking switch if it's a CPO switch?
If the customer wants us to do that, of course. Yeah. I mean, the most difficult work that we do is the stuff that we don't ever ship it as a finished product. We use it to produce the finished products. It's the packaging, you know, the precision packaging, the photonics packaging that we do. That's the most difficult thing that we do. We don't necessarily talk about it externally. We don't really advertise those services because if you think about it, we don't do it as a standalone service. We don't do packaging and then sell the packaged devices. We do the packaging so that we can use those devices to make the product for the customers. Yeah, we're happy to go up as high in the stack as the customer wants us to.
Great. We'll wrap it up. Oh, sorry. Garo, please.
No, thank you. Thanks everybody for joining us. Our next Q&A session will be beginning at the top of the hour at 4:00 P.M. Pacific. Thank you.
Thanks very much, everyone. Thank you very much.