Fabrinet Sees Unstoppable AI Demand as Major Hedge Funds Triple Their Stakes

Citadel and Millennium Boosted Fabrinet Stakes in Q2 Amid AI Demand - Fabrinet (NYSE:FN)

Citadel Advisors and Millennium Management made notable additions to their Fabrinet (NYSE:FN) positions during the second quarter, according to the latest Form 13F filings. The timing is striking: weeks later, the optical manufacturing specialist reported record financial results, revealed an increasingly diversified customer base led by Cisco Systems, Inc. (NASDAQ:CSCO), Nvidia Corp (NASDAQ:NVDA), Amazon.com Inc (NASDAQ:AMZN) and Nokia Corp (NYSE:NOK), and told investors there was "no end in sight" to customer demand -- highlighting why the company has become an increasingly important player in AI infrastructure. Citadel Advisors and Millennium Significantly Expanded Their Fabrinet Positions The latest Form 13F filings for the quarter ended June 30 show Citadel Advisors and Millennium Management substantially increased their exposure to Fabrinet. * Citadel Advisors made the biggest move. Its disclosed common stock position jumped from 4,963 shares in the first quarter to 151,705 shares in the second quarter, an increase of nearly 2,957%. The reported value of the holding climbed from $2.6 million to $85.3 million. * Millennium also significantly expanded its investment, increasing its common stock holdings from 61,692 shares to 164,313 shares, while the reported value of the position rose from $32.2 million to $92.4 million. * Another notable institutional investor, AQR Capital Management, also added to its position during the quarter, increasing its holdings from 71,849 shares to 79,795 shares, with the reported value rising from $37.1 million to $44.1 million. Form 13F filings disclose institutional holdings at quarter-end rather than the reasons behind investment decisions. Still, the portfolio changes coincided with one of Fabrinet's strongest operating quarters to date. Also, it's worth noting that 13F filings are a snapshot of holdings as of June 30, 2026, and don't reflect any portfolio changes that may have been made by any of these hedge funds after the quarter ended. Movers Fabrinet CEO Sees 'No End in Sight' to Data Center Demand Fabrinet (FN) beats Q4 estimates with a 68% surge in data center revenue, but stock dips on softer Q1 GAAP guidance. Read key highlights. 2 min read Read this article Fabrinet's Customer List Reflects a Broader AI Infrastructure Story The company's latest earnings suggested its AI opportunity is becoming increasingly diversified. Chief Financial Officer Csaba Sverha told investors: "In 2026, we continued to diversify our customer base, with four customers representing 10% or more of total revenue. These were Cisco at 20%, Nvidia at 16%, Nokia at 11%, and Amazon at 11% of total revenue." Latest Private Market Opportunities Join 400,000+ Investors The disclosure stands out because it shows Fabrinet generating meaningful revenue from networking, cloud and semiconductor leaders rather than relying on a single AI customer. That's our inference based on the customer mix; management simply highlighted the diversification. The company also reported that data center applications accounted for 51% of fiscal 2026 revenue, underscoring the growing importance of AI infrastructure to its business. Management Says AI Demand Shows 'No End in Sight' Beyond the financial results, management struck an unusually confident tone about the demand environment. Chief Executive Officer Seamus Grady said: "Demand from these markets continues to increase, which makes us optimistic about the long-term durability of these trends." Later, when asked about customer demand, he added: "There looks to be, you know, no end in sight to the demand from the customers." Management also pointed to continued momentum across data center interconnect (DCI), high-performance computing (HPC) and optical transceivers, suggesting AI-related demand remains broad-based rather than dependent on a single product cycle. What Fabrinet Investors Should Watch Next The latest 13F filings don't reveal why Citadel Advisors, Millennium or AQR increased their Fabrinet positions, nor do they capture any portfolio changes after June 30. What they do show is that several sophisticated investors increased exposure ahead of an earnings report that reinforced Fabrinet's role in AI infrastructure through a diversified customer base, record operating performance and confident management commentary. For investors, the next key question is whether those fundamentals continue to support the company's premium valuation as AI networking deployments scale and new customer programs move into production. Tech Cisco, Nvidia, Amazon and Nokia Fuel Fabrinet's AI Boom Fabrinet's latest earnings showed Cisco, Nvidia, Amazon and Nokia as its largest customers, highlighting the company's expanding role in AI infrastructure. 3 min read Read this article Shutterstock Market News and Data brought to you by Benzinga APIs To add Benzinga News as your preferred source on Google, click here.

Fabrinet at Rosenblatt ai summit: growth broadens beyond nvidia By Investing.com

Tuesday, 18 August 2026 -- At Rosenblatt's 6th Annual Technology Summit: The Age of AI (Part II), Fabrinet (FN) outlined a business that is growing quickly in data center optics and widening its customer base, even as it faces lower near-term free cash flow and the risk of spending ahead of demand. Management said the company is benefiting from strong AI-related infrastructure demand, while keeping a disciplined stance on capital spending and its role as a pure-play contract manufacturer. Key Takeaways * Fabrinet said its data center interconnect business reached a $1 billion run rate by the end of fiscal 2026, up 140% from a year earlier. * Management said fiscal 2027 growth could exceed fiscal 2026's 36% pace, though it stopped short of giving a forecast. * The company is expanding capacity from $5.8 billion to $9.8 billion by early calendar Q1 2027, with more room planned after that. * Fabrinet rejected becoming an original design manufacturer, saying its contract-manufacturing model helps preserve customer relationships. * Growth is coming from several areas, including datacom, telecom, automotive, industrial lasers, low Earth orbit satellites and optical coherent switches. Financial Results Management used the conference to frame the company's recent performance and expected trajectory: * Data center interconnect, or DCI, grew 140% year over year and reached a $1 billion run rate by the end of fiscal 2026. * The communications infrastructure segment grew 40% year over year. * Full-year fiscal 2026 revenue rose 36% year over year, after 19% growth in fiscal 2025. * Fourth-quarter fiscal 2026 revenue implied an annualized run rate of about $5.3 billion. * Current capacity at the end of fiscal 2026 was about $5.8 billion. * By early calendar Q1 2027, total capacity is expected to reach about $9.8 billion, an 85% increase. Management also discussed margins, cash flow and capital spending: * CapEx is expected to remain around $250 million in fiscal 2027, similar to the prior year. * Prior-year free cash flow was about $4 million, reflecting heavy investment. The company's shares currently trade at a P/E ratio of 40.85 with a PEG ratio of 1.35, and according to InvestingPro analysis, the stock appears overvalued relative to its Fair Value -- placing it among companies on the Most Overvalued list. * Return on invested capital remains near 40%. * Q1 fiscal 2027 will face a 20 to 30 basis point expense headwind from annual merit salary increases. The company's gross profit margin of roughly 12% in the last twelve months reflects the capital-intensive nature of its contract manufacturing model -- an InvestingPro Tip notes that Fabrinet suffers from weak gross profit margins compared to industry peers, though this is partially offset by strong operational execution. * One-time other income from the prior quarter will not repeat in Q1 guidance. On capital structure, the company said it recently secured a $75 million term loan from a Thai bank in August 2026 and is expanding credit lines to support growth. It also said $169 million remains under its buyback authorization. Operational Updates Fabrinet said the strongest part of its business remains optical networking for data centers, but it is seeing growth across several other lines as well. Data center interconnect and datacom * DCI has grown from "a couple of million" in revenue to a $1 billion run rate. * Management said 400ZR, 800ZR and ZR+ products help solve a power-distribution problem in large data centers. * As power limits at substations become a bottleneck, customers are spreading data centers out and linking them with DCI equipment. * Fabrinet said it now serves multiple customers in this market, not just one hyperscale account. * The company manufactures both pluggable modules and component content, which management described as creating a "sticky" business. * Multi-rail technology is under development with several customers, but timing will be announced by customers, not by Fabrinet. Customer mix and diversification * Management said Fabrinet is not a proxy for NVIDIA, even though NVIDIA remains a very important customer. * The company said its datacom business now includes three broad customer groups: the original hyperscale customer, merchant transceiver makers and hyperscale direct customers. * A second merchant transceiver vendor is expected to begin production in the December quarter. * Initial demand is centered on 800-gig short-reach transceivers, with a move toward 1.6T products later. * Quantum compute customers were also cited as a possible growth area. Automotive and industrial * The automotive segment had a strong quarter, helped by new program wins with existing customers. * EV charging business is growing well. * Lidar is moving from a flat phase into a ramping phase as the market consolidates around a few major players. * In industrial lasers, Fabrinet said it won "a couple of really good wins" with customers it had worked with before. * Management said the industrial laser industry is slowly shifting from overflow outsourcing to more strategic outsourcing. Communications infrastructure, LEO and optical switches * Low Earth orbit satellite manufacturing was described as a "very important and fairly rapidly growing segment." * Fabrinet said it has worked in LEO for multiple years with a couple of major customers. * Because satellites eventually burn up in the atmosphere, the business requires ongoing replacement production. * Optical coherent switches were described as a strong fit and an important product category. * OCS is still small, but management expects it to become a meaningful revenue contributor over the next 12 to 18 months. * The company is working with multiple OCS customers, including one already shipping product. NPO and CPO packaging * Non-pluggable optics and coherent pluggable optics were described as higher value-add businesses than traditional transceiver work. * Management said the traditional model is roughly 70% to 80% materials and 20% value-add, while NPO reverses that mix because customers consign wafers. * The NPO process includes wafer singulation, packaging and die-level and device-level testing. * These products may carry lower average selling prices, but they can support higher margin percentages. * Management said the Raytech partnership is important and will be established on Fabrinet's Thailand campus in the coming months. * ELSFP is one product category, but the company said broader packaging opportunities exist beyond that. Capacity Expansion One of the clearest messages from the call was that Fabrinet is building ahead of demand. Management said customers are already asking for capacity and ramping with the company. * Building 10 is already under construction, with equipment installation underway on Floor 1. * Floor 3 occupancy is expected in October 2026. * Building 10 is expected to add about $3.5 billion of revenue capacity at the high end. * The building cost is estimated at $132 million to $133 million for a 2 million square foot factory. * The Santa Clara facility adds about $250 million of capacity and cost about $76 million to acquire. * The Nava Nakorn facility adds another $250 million of capacity. * By early calendar Q1 2027, Fabrinet expects to have about $9.8 billion of total capacity. * Management said Buildings 11 and 12 could each add about $2.1 billion of capacity over roughly 18 months to build. * Over three years, the company sees about $4.2 billion of additional capacity from those two buildings, bringing total potential capacity to about $14 billion. * The company is also looking for more land in Bangkok for future expansion. Seamus Grady, chairman and chief executive, said the economics of the expansion are attractive. "The downside risk is tiny and the upside opportunity is immense," he said, adding that even if Building 10 were idle, the gross margin headwind would be about 15 basis points. He also said the company is already installing equipment, so the building will not sit idle. Future Outlook Management said it is possible, though not guaranteed, that fiscal 2027 growth could be faster than fiscal 2026's 36% pace. * Grady said faster growth in fiscal 2027 is "not beyond the bounds of possibility." * He said the three ingredients for stronger growth are demand, capacity and execution. * Demand is already present, according to management. * Capacity is being added in advance of need. * Execution, management said, has been Fabrinet's long-standing strength. The company said its growth outlook is supported by several product and customer trends: * DCI demand is expected to rise "dramatically over the next several years." * OCS should ramp over the next 12 to 18 months. * LEO remains a fast-growing business with recurring replacement demand. * NPO and CPO are expected to become good revenue drivers and margin enhancers over time. * Datacom diversification is continuing beyond the original hyperscale customer. Management also said it is comfortable funding growth from operating cash flow and debt, rather than equity. It described capital investment in its own growth as the best use of cash, given the company's roughly 40% return on invested capital. Q&A Highlights During questions, management returned several times to the same themes: DCI growth, customer diversification and the company's decision to stay a contract manufacturer. * On DCI, Grady said 400ZR, 800ZR and ZR+ solve a basic power problem for large data centers by allowing sites to be spread out and linked together. * On multi-rail technology, he declined to give timing, saying customers would announce it when ready. * On automotive, he said the recent strength came from new programs with existing customers, not from a wave of new customers. * On industrial lasers, he said the market is beginning to outsource more strategically, which could support further growth. * On OCS, he said the market could become meaningful over the next 12 to 18 months and that Fabrinet can supply sub-assemblies, finished products or both. * On NVIDIA, he said the company is still important, but the goal is to reduce dependence on any one customer and replace early growth with new vectors. * On market share, management said it does not focus heavily on share percentages because customer disclosures are incomplete and the real goal is to build each relationship into a large business. * On becoming an ODM, Grady said the company has chosen not to do that because many long-time customers would be upset and because the contract-manufacturing model is still producing enough growth. Grady said Fabrinet's competitive position has improved as some rivals have moved toward ODM models and, in the process, lost customers that prefer independent manufacturers. He said the company wants to keep growing with traditional customers such as Cisco, Ciena and Nokia while also serving hyperscale and merchant customers that bring their own designs. He closed with a broad message of confidence, saying Fabrinet has "really excellent growth" in front of it, a "really excellent customer base" and a long record of execution. Readers can refer to the full transcript below for more detail. Full transcript - Rosenblatt's 6th Annual Technology Summit: The Age of AI (Part II): Moderator: Hi, everybody. Good afternoon. I'm Mike Genovese, the cloud and communications infrastructure analyst at Rosenblatt Securities, and this is "The Age of AI" tech conference. I'm super happy today to be joined by the team from Fabrinet. We've got everybody. We have Seamus Grady, the Chairman and CEO, Csaba Sverha, the Chief Financial Officer, and Garo, the IR guy. Hi, nice to see you. Welcome, everybody. Nice to see everybody. Seamus Grady, Chairman and CEO, Fabrinet: Thank you. Moderator: especially after reporting earnings last night. I'm sure you've been busy speaking to the sell side, starting to speak to the buy side. Glad to have you here when we have fresh information to talk about. Let's just start with the telecom business. We'll go through the different segments, and I know you've changed your segment reporting going forward, but we'll talk about it on the segments you've been reporting up until now. The communication infrastructure up 40% year-over-year, that's the new category. But telecom up much more, data center interconnect up 140%, sorry, year-over-year. Billion-dollar run rate in DCI. You had Cisco as a 20% customer, Nokia as a 11% customer. We think that Ciena might be getting close, we don't know. But there's a lot of really strong stuff going on in telecom. Specifically on DCI scale across 800ZR, OLS, multi-rail, are we just getting started on this? Is this business going to continue to grow and get to much bigger numbers than we're seeing now? What should investors know about this telecom/DCI segment that's been so strong, but what ending are we in, and how sustainable does this look? Seamus Grady, Chairman and CEO, Fabrinet: Yeah. Thanks, Mike. It looks very strong and quite sustainable from what we see. DCI in particular, 400ZR, 800ZR, it really solves a power problem for the industry in a very elegant way. For a lot of these huge data centers, the limiting factor is often the amount of power that can be physically gotten into the data center from the substation, that you max out your power, and the best way around that is to spread the data centers around and connect them using DCI, like 400ZR, 800ZR, ZR+. It's an industry trend that we've seen coming for some time. For us, the ZR business started off as a couple of million, really, not that long ago. And now, as you said, at the end of the year, we were at a billion-dollar run rate, which is just phenomenal. Originally, DCI, when there was softness going on in telecom, DCI was offsetting that for us. But now DCI on its own is just phenomenal growth. And the nice thing about that from our point of view is it's across several customers. It's not just one customer, it's several customers, all of whom are the leaders in that space. And we're making, obviously, pluggable modules for those customers, but we also make a lot of the component content that goes into those modules. So the business is quite sticky, and the work we do for the customers is very important to our customers and to their customers. So DCI, we see having a lot of legs. The demand is very strong and looks to be very strong for some considerable time to come. Certainly, the next several years, as far ahead as we can see, the demand looks to be just increasing dramatically over the next several years. Moderator: When do you expect to see multi-rail really start in your production? I imagine it's not so much behind us, but that would be ahead of us. Seamus Grady, Chairman and CEO, Fabrinet: Yeah. It's a function of what our customers decide to do, and we're always kind of careful of not announcing products on behalf of our customers. So we'll defer to our customers to see what they have to say publicly. But we are working on that with a number of our customers, and that will be an important growth driver as well in the coming years. But not really for us to say at this point. It's too early. Moderator: Right. And I just want to remind the audience here, if they want to ask questions, if you type a question into the upper right-hand widget on your Zoom screen, which we don't have, Seamus. I see you looking up- Seamus Grady, Chairman and CEO, Fabrinet: Okay Moderator: Right. We actually don't have that. Only the audience has it. But the questions will come directly to me, and I- Seamus Grady, Chairman and CEO, Fabrinet: Okay Moderator: will certainly ask the questions that the audience type in. So we'll get to Datacom, but let's also touch on auto and industrial. You had a very strong quarter for auto. Did something specific happen there, or is that strength that we might repeat? Seamus Grady, Chairman and CEO, Fabrinet: Yeah, there's a couple of things going on there in auto for us. One is we've had some new program wins and some nice growth in the EV charging part of our auto business. That's going very well for us. And then we are starting to see some growth in lidar. Lidar has been somewhat flat, I would say, for some time. Our strategy in lidar was really to try and win all the customers. In the early days of lidar, it wasn't quite clear to us who the winners would be. So we set out to try and win, really, all of the players. And now we're a few years into lidar now, and it's kind of shaking out. There's a few really good companies there who look to have some good volume, some good demand. So we are starting to see lidar ramp as well. Really EV charging and lidar would be the two areas that we're seeing growth in automotive. Moderator: When you say new programs, as you said with EV, does that imply additional customers, or does it mean new programs with an existing customer? Seamus Grady, Chairman and CEO, Fabrinet: New programs with an existing customer. Moderator: Yeah. Got it. Okay. But then also, we spoke last night after the call, and we talked about a lot of different things, and then as we were running out of time, you said, "Look, we haven't even talked about industrial. We haven't talked about- Seamus Grady, Chairman and CEO, Fabrinet: Yes Moderator: lasers industrial, and we're really starting to see some good things there." So now we have more time to flesh that out. But what were you referring to? Seamus Grady, Chairman and CEO, Fabrinet: Yeah. So we have a couple of really good wins, I would say, in the industrial laser space. Companies that we've done business with in the past, but that haven't really grown. So we're starting to see some nice growth there with a couple of customers in the industry laser segment or category. We've always felt that that industry is a really good fit for us. It's been slow to grow for us because that industry has been quite slow to outsource. But we're seeing that change now. We're seeing more of those companies are leaning more towards outsourcing and are looking to outsource strategically, not just for overflow manufacturing, but strategically. So we're starting to see that happen now, and again, ordinarily that would be relatively big news. It's gotten a little bit overshadowed because of all the bigger news we have going on. But we're very optimistic about that. Really all of the product categories that we play in are all going very well for us and growing. Moderator: Yep. All right. Let's talk about Datacom in the old segment first, then I will talk about data center, the new segment. If we talk about last quarter in Datacom, when I look at the numbers, Datacom overall, you beat a lot in telecom and DCI, a lot in auto and industrial. Datacom overall was just a little bit softer than we expected. It seems like there are a few customers now in the Datacom mix, as opposed to before it was all just one customer. Seamus Grady, Chairman and CEO, Fabrinet: Yes. Moderator: To the extent possible, tell us what is going on with the kind of customer mix and customer diversification within Datacom. Seamus Grady, Chairman and CEO, Fabrinet: We have always said since we started, let us say, the big ramp a few years ago with initially NVIDIA, we have always said that we wanted to diversify in that space to a couple of other areas. Namely merchant transceiver manufacturers and also hyperscale direct. We also talked at the time about if there are other, let us say, competitors of NVIDIA who need optical interconnect products and transceivers, we would be happy to make them for them. We have not made a huge amount of progress on that front because NVIDIA are still very clearly the leader. The two areas I talked about, hyperscale direct and merchant transceiver manufacturers, we are making progress there. We are actually shipping to both right now, and it is early days, but we are starting to get going. The demand, again, a little bit like the DCI conversation, the demand picture that we see looks very strong, very robust. So we- Moderator: Yeah. From my perspective, the diversification is the story in that we Seamus Grady, Chairman and CEO, Fabrinet: Yes Moderator: do not focus so much on one customer. People do want to know what happened with the original data center customer. That does seem like it was down quite a bit sequentially in the quarter. What happened, and then what is going to happen with that customer from here? Seamus Grady, Chairman and CEO, Fabrinet: That is really one of the problems with the way we have been categorizing the revenue historically. It should not be possible to determine the actual revenue with one individual customer in a quarter, and that should not be possible, and yet it is. That is one of the reasons to make the change. We have to protect our business with that customer, and we have to protect that customer's business as well. We are very happy with that relationship. If you go back to the early days of our relationship with NVIDIA, there really was not any particular capacity out there for 400G and 800G transceivers in the volumes that were required to support the growth in their business. NVIDIA, as we know, acquired Mellanox, designed their own transceivers, and in the early days, Fabrinet was the only manufacturer of their 400G and then 800G transceivers. Naturally, they have diversified over the years and have multiple sources now for 800G transceivers, let us say, including their own design that we make, but also other designs that they source from other suppliers. That has always been the plan, I suppose, and we're not at all surprised at that. Our focus has been on obviously doing everything we need to do for NVIDIA and doing what they need us to do for them, but also diversifying, as I said, into these other areas, merchant and hyperscale direct. NVIDIA is a very important customer for us and will continue to be a very important customer for us. The outsized growth that we got with NVIDIA in the early days, that just was never going to continue at that pace because that blistering pace of growth we had, we always knew we would have to replace some of that growth with growth in other areas. And that's what we're off to do right now. Moderator: Just for modeling purposes, should we think about NVIDIA at the quarterly revenue level that they were at kind of being at a similar for the next four quarters? Seamus Grady, Chairman and CEO, Fabrinet: I can't help you with the model, Mike. They're still a very important customer for us, but our focus is on servicing their needs and, as I say, bringing on these other growth vectors as well. And that's really all I have to say. The revenue with NVIDIA or any other customer, we'll disclose once a year when we disclose the 10% customers. That's all we can do, really. Moderator: Okay. Well, talk about just the Datacom opportunity in terms of being a contract manufacturer and not an ODM and- not a designer of products, because it sounds like that customer basically said, "We don't want to only design our own transceivers. That's not our business. Seamus Grady, Chairman and CEO, Fabrinet: Yeah. Moderator: But if they come to you and they say, "Design a transceiver for us," you say, "Well, that's not what we do." So maybe talk about that a little bit more. But I guess more importantly, you do now have a large hyperscale transceiver customer who brought you a design, which they probably had partners in working with. They probably didn't all do it themselves, but they're bringing you the design. You then have merchant guys who just need help with capacity to make- what they've designed, and they have demand for. Seamus Grady, Chairman and CEO, Fabrinet: Yeah. Moderator: So, the growth opportunities in Datacom as a CM as opposed to an ODM, could you just flesh that out a little more? Seamus Grady, Chairman and CEO, Fabrinet: Yeah, certainly, if we had been prepared to become an ODM, there would be products we could be building and shipping right now, but that's not something we're going to do. We've decided, for strategic reasons, that that's not something we're going to do. It's very important. We have a very varied and vast customer base, some of whom would be quite happy for us to be an ODM. But a lot of the customers who got us to where we are, we want to continue to grow with, and they would be deeply upset if we were to become a product company. Companies like Ciena, Cisco, Nokia, and the like, if we were to become a product company, that would be problematic for our relationship with them. We've decided rather than get into that space, we just won't ever be a product company. It does present a little bit of a challenge, for example, when we want to do business with a hyperscaler, we can't own the IP. It's the same with if it's NVIDIA or anybody else, we just won't own the IP. We're happy to do whatever they need us to do. We're happy to produce any product that they want us to produce, but we won't own the IP. We're a pure-play contract manufacturer. We're not going down the path of ODM. Moderator: Okay. Oh, one other question. It sounds like in the quarter you just reported, you had at least three Datacom customers, right? And there was a merchant, and there was a hyperscaler, and then there's the original customer. Are you saying that there's going to be another merchant coming in in the September quarter or the December quarter? I've had some questions on what the timing of that- Seamus Grady, Chairman and CEO, Fabrinet: Yes. We're working on another one that we hope to start producing towards, I think, what did we say, Csaba, on the call? Was it towards the end of the year or in the Csaba Sverha, Chief Financial Officer, Fabrinet: Towards the end of the year. The merchant will start in the December quarter. Moderator: But that's a second merchant vendor. Seamus Grady, Chairman and CEO, Fabrinet: Yes. Moderator: Yeah. Seamus Grady, Chairman and CEO, Fabrinet: That's correct. Moderator: All right. Seamus Grady, Chairman and CEO, Fabrinet: Back to your question about the, we believe there's ample growth opportunities for us as a contract manufacturer. There could be more or less. I think the short term, there could be more growth opportunities if we decided to become an ODM, but we think long term, we would damage ourselves, actually, and damage our growth with the traditional customer base if we became an ODM, and that's why we choose to stay away from it. The proof of that is we have been picking up business and winning business with some of our customers, not just because, but in some cases because some of our competitors have chosen to go the route of being ODMs, and some of our customers don't like that. That's the path we've chosen. We believe there's more than enough growth for us to work on as a pure play contract manufacturer. Moderator: Great. So the new data center category going forward, just to make sure I understand, that's going to include DCI in that category as well as- Seamus Grady, Chairman and CEO, Fabrinet: Yes Moderator: The hyperscale transceivers, the merchant transceivers, and the high-performance computing? Am I getting that- Seamus Grady, Chairman and CEO, Fabrinet: That's correct. Moderator: correct? Seamus Grady, Chairman and CEO, Fabrinet: Exactly. Moderator: That's everything that's in the new category. Seamus Grady, Chairman and CEO, Fabrinet: Exactly. Moderator: Right. Is there any chance of additional high-performance computing customers? As you work with the one, does that attract other customers over time? Seamus Grady, Chairman and CEO, Fabrinet: We think so. I think there's other high performance, let's say, hyperscale type customers. There's also some quantum compute customers that we feel that we're working with that could be very good, and that would go into that category as well. So there's both. There's hyperscale and then there's quantum compute customers that we're working on. Especially as in order to move the workloads around, those type of products need optics. It's not just electronics. They will also need optics in the future. So that's why it's such a good fit for us. Moderator: Mm-hmm. Okay. My next set of questions, and some of them come from the audience that I am adding in here because we are getting good questions from the audience. It is more talking about, I am going to ask you about sort of technologies and product categories, and the question is really the timing of when it could impact you positively with revenues and the kind of scale of what that could be, to whatever extent you can answer. So sort of what does X mean for you? There is going to be three or four of these. So the first one is OCS. Is this an important product category for you? When could we expect revenues and if Coherent is saying it is a $4 billion and Lumentum is saying it is a $10 billion TAM by 2030, it is somewhere in the middle, and how much roughly flows to you guys? Seamus Grady, Chairman and CEO, Fabrinet: Well, for OCS, yeah, it is a really good fit for us. It is an important product category. We think it will be a significant revenue contributor in the future. It is quite small today. We have shipped some, but it is quite small. We are happy to do whatever the customer wants us to do. So if the customer would like us to do sub-assemblies and then they do the finished product, or do sub-assemblies, they do the finished product up to a certain volume, and then we start to produce the finished product at a point in time. Whatever the customer wants us to do, really, we will do. For Coherent, I think Coherent have a lot of their own in-house capacity, both at the system level, but also at the wafer level. So they do a lot of the in-house silicon themselves. I believe they will continue to manufacture in-house. But yeah, there is a number of companies we are working with, one of whom we are actually shipping product. The others are early days. They are more startup-type companies, but it does seem like it has a lot of potential, and we are looking forward to really beginning to ramp that over the next, I would say, 12 to 18 months, something like that. Moderator: That would also go into the Datacom category, correct? Seamus Grady, Chairman and CEO, Fabrinet: Yes. Moderator: Right. Well, this one I wasn't going to ask, but it's from the audience, which is, how large is low Earth orbit, LEO, and how fast- Seamus Grady, Chairman and CEO, Fabrinet: Yeah Moderator: is that growing? So satellites. Seamus Grady, Chairman and CEO, Fabrinet: We haven't actually quantified it. It's in our communications infrastructure category. But it's a very important and fairly rapidly growing segment for us. We've been producing products in that space for several years. We have a couple of major customers there. So it's significant. It's a significant and important category for us, and it's one that's growing fairly rapidly as well. And again, it's right in our kind of wheelhouse of capability and, of course, the nice thing about the low Earth orbit products is they're in low Earth orbit, so they eventually get pulled into the Earth's atmosphere and disintegrate and have to be manufactured again. So it's a really good fit for us, and we have a couple of really good customers there that we're growing nicely with. Moderator: Yep. I am just going to read the next question that directly comes from the audience, and these are not exactly the way that I would ask them, but I am just going to ask it and just get your answer. Seamus Grady, Chairman and CEO, Fabrinet: No problem. Moderator: What does NPO mean for you? Are you just assembling the ELSFP? What are the different revenue streams for NPO, and can this increase the gross margins? I do not think about you having kind of mixed gross margin, right? I feel like you have a fixed gross margin, but I should let you answer. Seamus Grady, Chairman and CEO, Fabrinet: No, I mean, NPO would be much better gross margin than our traditional business because it is much higher value-add content. Moderator: Okay. Seamus Grady, Chairman and CEO, Fabrinet: If you take the typical COGS on a product we make, it depends on the product, but let's assume if the COGS is 100%, let's assume that's 70%-80% material, 20% value add. For NPO, it's much higher value add content and much lower material content because the material, in many cases, is a wafer which would be consigned by the customer. Then we take the wafer, we singulate it, we do the packaging, and the testing at the die level and at the device level. So it's a much higher value add content, probably lower ASP, but a higher margin. So it's very good business for us. It's difficult for us to talk with any kind of credibility about what we'd actually be doing because we'd be disclosing what's going on with individual customers. NPO, we believe should be a very good revenue driver for us in years to come. And like I say, it's a good margin enhancer. Our relationship with Raytech, we think will be an important development in that regard. Raytech, they're a very good company. They will be setting up an operation on one of our campuses in Thailand in the coming months. So that's a critical and important part of our ability to provide these essentially kind of precision packaging services to our customers in regard to NPO and also CPO. Moderator: Yeah, so I- Seamus Grady, Chairman and CEO, Fabrinet: ELSFP, of course, it's a product category that's somewhat like a traditional transceiver. They're not that complicated. ELSFPs are not that difficult, but there's a lot more opportunity for us in NPO than just ELSFPs. Moderator: Such as? Seamus Grady, Chairman and CEO, Fabrinet: The packaging of the devices that I talked about. Moderator: Yep. Could we have just substituted NPO and CPO there, or is there anything to add on CPO? Seamus Grady, Chairman and CEO, Fabrinet: Yeah, somewhat. They're quite similar. I think NPO is probably a little bit more straightforward than CPO. But we're pretty excited about both opportunities and our ability to kind of win and grow with the customers there. Moderator: Great. The stock is down today, there might be other reasons in the market, interest rates, and things like that, but I simply think that we're just getting over this fixation with NVIDIA. Seamus Grady, Chairman and CEO, Fabrinet: Right Moderator: this, right? We'll move past this. The growth here is incredible, and I think in response to a question, you said it's not beyond the bounds of possibility that FY 2027 could grow faster than Seamus Grady, Chairman and CEO, Fabrinet: Yes Moderator: FY 2026. We're really early in 2027, so what would we have to see to know whether it could be even faster, and is there a capacity issue with growing even faster than last year? Would we run up against capacity issues? I guess I'm asking too many questions at once, but now going to an audience question, exiting fiscal 2027 with building 10 and the capacity expansion, what's the potential revenue run rate exiting the fiscal year? Seamus Grady, Chairman and CEO, Fabrinet: So maybe I'll separate capacity from the revenue run rate, because we don't forecast the revenue run rate a year from now. We only forecast one quarter at a time. Our forecast for Q1 is whatever we said in our guidance. The answer I gave was actually in response to a question from Tim during the call around is it beyond the bounds of possibility that, or something along those lines, that our growth in FY 2027 could be even higher than our growth in FY 2026. Our growth in FY 2026 year-on-year was, I think, 36%, Csaba, is that right? Csaba Sverha, Chief Financial Officer, Fabrinet: Correct. 36%. Correct. Seamus Grady, Chairman and CEO, Fabrinet: It is not beyond the bounds of possibility that we could grow at a faster pace in FY 2027. That's not a forecast, that's not guidance. Certainly from a demand point of view, the growth looks to be very strong. In order for the growth to happen, a few ingredients have to be in place. One is the demand has to be there. That's for sure, the demand is there. Secondly, we have to have the capacity, and we have the capacity, and we will have the capacity, and I'll walk through in a moment the capacity additions that we're making. Then we have to execute. The products have to work, and we have to be able to get components and all that stuff as well. Usually the biggest challenge is demand. If we have the demand, our history has been if we have the demand, we execute, and we can deliver on that. So the demand looks to be there. We just have to make sure we can get the components and we can execute. The really important message, and I think you're exactly right, is we want to get away from this obsession with NVIDIA. The growth is staggering, and the diversification is also excellent. The financial performance of the company is excellent. Everything is excellent. Yet the industry seems obsessed with NVIDIA. We're not a proxy for NVIDIA. We never have been. We never claimed to be. In fact, we always say we're not a proxy for NVIDIA. But I think the more we say we're not a proxy for NVIDIA, the more people think we are, and we're not. We are a contract manufacturer who serves multiple customers, one of whom is NVIDIA. They're a very important customer for us, but we have several other customers as well. In terms of capacity, right now our run rate at the end of Q4 was $5.3 billion. If you take our Q4 revenue times four, 5.3. Our capacity right now is for about $5.8 billion, which is our run rate at the end of Q4 plus the additional capacity we added in Pinehurst by converting offices into manufacturing space. So our capacity right now is about 5.8. Building 10, when that's fully available, that'll give us capacity for about another, at the high end, call it about $3.5 billion of capacity. Nava, the new factory in Nava Nakorn, adds capacity for about $250 million, and then our Santa Clara operation that we just acquired will add capacity for about another $250 million. If you add all those together, $5.8 billion plus $3.5 billion plus $250 million and another $250 million, that gets you to $9.8 billion of capacity in calendar Q1, let's say, in the March quarter. Early in the March quarter, actually. That's an 85% increase versus where we ended the year in terms of capacity. We certainly have the capacity to continue to grow and scale. Then beyond that, building 11, which we haven't announced building 11 yet, but building 11, were we to pull the trigger on building 11, will take probably 18 months to build, and that will add capacity for another $2.1 billion of revenue, approximately. The same for building 12. If we were to move ahead with those capacity adds, they would take about a year and a half, about 18 months each. Over the next three years, on top of the $9.8 billion of capacity that we'll have at the end of this year, we would be able to add another $4.2 billion, which would take us to approximately $14 billion of capacity over the next few years. Of capacity, not a revenue forecast. That's the first step for us is make sure we have the capacity ahead of the demand. It's not a case of there's no field of dreams. It's not like if you build it, they will come. They are already here. The customers are here. They want the capacity. They want us to ramp with them. We have the demand in front of us, we just have to make sure we keep the capacity coming on ahead of the demand and make sure we ramp it appropriately. Even with all of that, we're still looking for more land in Bangkok, in Thailand, and looking to expand beyond our current footprint. Moderator: When you say that the customers are here, the demand is here, but how much of that is actually orders versus forecasts? Is there any kind of prepayment? I guess the real question is, as the CEO and as you sit here, you're already executing on building 10, and it's very unusual for you to talk about 11 and 12. In old days, that would never happen, right? Seamus Grady, Chairman and CEO, Fabrinet: Correct. Moderator: That shows we are in a new environment that you are even discussing 11 and 12 right now. What are you watching in making these decisions? Seamus Grady, Chairman and CEO, Fabrinet: Well, it is actually a really straightforward kind of calculus for us because the downside risk is tiny and the upside opportunity is immense. If you take the economics of Building 10, for example, and you can then extrapolate that for Building 11 and Building 12. Building 10, and Csaba, please stop me if I get any of these numbers wrong, Building 10, it is about a 2 million sq ft factory. It has revenue capacity for, I think we said about $3.5 billion of revenue. So the upside opportunity is the profit, the operating profit, that we are able to generate when we ramp up in Building 10. So it is immense. The downside risk, even if Building 10 were to sit idle, which it won't, because we are actually already starting to install equipment. We are starting to get going on Floor 1 of Building 10 already, and we will start to occupy Floor 3 in October. Anyway, even if Building 10 were to have sat idle, the gross margin headwind is about 15 basis points. Moderator: Yeah. Seamus Grady, Chairman and CEO, Fabrinet: The downside risk is negligible, and the upside opportunity, just to put it in context, just over five months' worth of operating profit at full run rate would pay for the whole factory. So from an investment point of view, it is a really excellent use of the company's cash, leave aside the fact that we have to have this capacity for our customers. So the downside risk is tiny and the upside opportunity is immense. The downside risk, if we don't put this capacity in place, customers will have to go somewhere else, and we lose out on the opportunity. So, it is a very straightforward decision for us. Yet ordinarily, we don't talk about buildings until we are ready to pull the trigger, but I think these are very different times. The demand is so huge. We do feel it's important to make clear that we will keep the capacity coming on ahead of the demand. I think at some point there was a concern that were we building Building 10 too late. Initially, was it too soon? Then the question was it too late? I think we'll find it was probably just at the right time, maybe just ahead of the demand curve. We'll do the same with Building 11 and Building 12. Moderator: Generally, what's the CapEx for this? The buildings are different sizes, so CapEx per building, I don't know if that makes sense. It just seems like the dollar return per CapEx dollar spent is. The investment here is obviously way lower than if you were building indium phosphide lasers. Seamus Grady, Chairman and CEO, Fabrinet: Yes Moderator: And had to build those kind of fabs, right? You're building manufacturing space, right? That's what this is. Seamus Grady, Chairman and CEO, Fabrinet: We build the. Moderator: How do we think about the CapEx? Seamus Grady, Chairman and CEO, Fabrinet: We build the building, and for Building 10, it is about $132 million, $133 million. But then the fit-out of the factories within the factory, that takes place in concert with the customer. Typically, if it is generic equipment or standard equipment, we pay for it. If it is product-specific or unique equipment, we ask the customer to pay for it. It is generally a fairly capital light business that we have. Maybe Csaba, if you want to talk about the CapEx for Building 10, and then what Building 11 or 12 would look like as well. Csaba Sverha, Chief Financial Officer, Fabrinet: Again, our CapEx, again, I am going to cover the entire spend of CapEx for the last year was about $250 million. We anticipate that to continue in the next fiscal year. Obviously, we are putting the capacity in place ahead of the demand. So we anticipate that these elevated CapEx levels to continue in 2027. As you know that we have just acquired a Santa Clara facility for $76 million. We are also finishing off Building 10, and obviously as the fit-outs and all those equipments are going in, we anticipate that the CapEx spend in the next fiscal year to continue to be in this $250 million range. Obviously, we also have to be mindful about the ROIC that we generate. By adding this CapEx and financing the growth through our own cash, we have been still able to generate very reasonable ROIC. I think it is in close to the 40% mark. We continue to believe that investing in our own growth is the best use of cash, as long as it generates this high returns of capital. We continue to expect, and we are comfortable financing it from our own free cash flow, actually. Moderator: Okay. Super helpful. Csaba, let me ask you just about the first quarter EPS guide, right? Strong revenue guide, good EPS guide, but you talked about the EPS doesn't grow as much sequentially as the revenue does, and you said usual first quarter expense seasonality. Could you flesh that out more what that means? Csaba Sverha, Chief Financial Officer, Fabrinet: Yeah. Typically in our first quarter, we increase our staff's salaries, and those merit increases are captured in our first quarter results. That has been the case for the past several years. So those seasonality, expense seasonality, will be adding probably about 20, 30 basis point headwinds in the quarter. So that's the biggest change from sequential basis. We did have some one-time other income in the prior quarter, so sequentially, those will not be included in our guidance. But fundamentally, that's the biggest element there, which is a temporary headwind. We anticipate that to make up with efficiencies throughout the year. So this temporary 20, 30 basis points are baked in our first quarter guidance. Moderator: And the investments in the buildings don't really have anything to do with it? Csaba Sverha, Chief Financial Officer, Fabrinet: The investments in buildings, obviously, as we are shipping revenue from there, it doesn't provide any headwinds in our numbers. If the building were to sit idle, that would probably result a small headwinds in gross margin. But since we will be shipping products right away, it's not going to give any headwinds from investments. Moderator: And just in terms of sort of free cash flow, and you have a lot of cash on the balance sheet, but with CapEx is elevated like this, there is not a lot of excess cash being generated. Are you comfortable with the balance sheet? Csaba Sverha, Chief Financial Officer, Fabrinet: Yeah. Moderator: Is there any need to raise money? Csaba Sverha, Chief Financial Officer, Fabrinet: We are very comfortable with the balance sheet. We still have ample of cash, and we also still have a lot of our cash is invested. Nevertheless, we took a small $75 million loan, term loan, just recently in August. We signed off with a Thai bank, obviously to finance our growth in Thailand. So we are mindful about expanding our credit lines as well to make sure that we have ample of credit to support the growth and the business. But again, we are very comfortable with the balance sheet. And as you look at the last year free cash flow generation, it was somewhat $4 million. So again, it goes back to our strategy of reinvesting in a growth and financing it from our own capital, so we remain very comfortable with the balance sheet. Moderator: Okay, great. What about just thoughts on buybacks here? Csaba Sverha, Chief Financial Officer, Fabrinet: We have a structured buyback program in place. It has two legs. We have a 10b5-1 plan whereby we fund the plan with the surplus cash that we generate from operating cash flow. Obviously, throughout the last year, that number hasn't been significant. We also have an opportunistic buyback program. We have, I think, $169 million left in our authorization, but we are committed to return the surplus cash through buybacks and opportunistically as we see fit. Moderator: Yeah. The discussion earlier that NPO, CPO could actually be a gross margin tailwind is very interesting and not even something I had considered, because I just thought your gross margins and operating margins would always be in a super tight range, because that's the price that you charge, right? There's very little OpEx, and it's kind of like we charge a markup, and this is it. Without giving guidance, so it's hard to do, but could we get higher margins in the future than we have now? If so, would it be a meaningful impact, or would we still be close to where we are now? Seamus Grady, Chairman and CEO, Fabrinet: I think if you look at kind of what drives the margin, as the material percentage goes up in a particular product for any given product, the gross margin percentage typically comes down. Moderator: Okay. Seamus Grady, Chairman and CEO, Fabrinet: Even if the gross profit dollars goes up. So, if you have a $10,000 product versus a $1,000 product, and they have the same amount of value add in them or transformation, then the 10,000 material content product will have obviously more profit dollars, but a lower profit percentage, if that makes sense. Moderator: Yeah. Seamus Grady, Chairman and CEO, Fabrinet: The point about NPO and CPO is they are more in the packaging realm than in the traditional contract manufacturing realm. That business just by its very nature, because it has a lot less material content, a lot more value add activity, the margin tends to be a lot higher, the margin percentage, but then the ASP is lower. You kind of trade one for the other. I think that work can really help the margin, but it is like everything else, you have to have a mix of that work. You cannot have all, like you do not want all steak in your diet, but you do not want all fiber either, if you follow me. It is a bit like that. You are going to have a mix of business, a mix of higher volume, maybe lower margin business, but then this business would be much higher margin but lower ASP. But we think very sticky business, because these things are difficult to make. Moderator: Okay. Great. I am going to ask one more question from the audience that is here, and then just a final question to you from me. But the audience question, which again, I am just going to read, how can we think about the three other customers in Datacom, so non-NVIDIA? Are these 1.6T? How do you think about share, and how much visibility do you have to the ramps? Seamus Grady, Chairman and CEO, Fabrinet: The non-NVIDIA customers at Datacom, let us say hyperscale direct and then merchant, they are mostly initially 800 gig short reach transceivers, and then moving to 1.6 in the future. But our initial foray will be with 800 gig. Moderator: Do you spend any time thinking about share? You are not going to have a majority share at any of these customers, I do not imagine, unless they are a small merchant. I should let you talk, not me talk. Seamus Grady, Chairman and CEO, Fabrinet: We do not overly analyze share. Obviously, we want to have a decent share for us to be able to provide good service for the customer and for us to be an important supplier for them. We want to have a decent share. It is not always easy for us to calculate the share. We tend to end up guesstimating the share, and they do not always tell us what share we have. Certainly, we focus more on can we grow the business with the customer to a significant level whereby we are able to provide them good service? If you are only doing $5 million or $10 million a year, it is very hard to provide good service for the customer if your revenue is down at that level. Whereas, when you are up in the multi-$100 million level, it is much easier to have the right resources, the right team, and infrastructure in place. We are more focused on growing each of these opportunities to become a significant revenue opportunity and significant piece of business for us rather than obsessing about share. Moderator: Yep, great. My final question, you just reported last night, obviously not the best stock reaction today, but this is just one day. My question to you is just what message do you want to leave investors with as we start on this fiscal 2027 journey? Seamus Grady, Chairman and CEO, Fabrinet: Well, I think we have a track record of excellent execution over many, many years. We've gone from several years of double digits compound CAGR, and then in FY 2025, we had, I think, 19% growth, FY 2026, 36% growth. While executing, we believe better than anyone in our industry, in the contract manufacturing industry. We have a track record of excellent execution and really deep customer relationships that are actually expanding. We're growing both the number of customer relationships and deepening the relationships with those customers. We have really excellent growth in front of us and a really excellent customer base and really good diversification. That's the message really. We believe we're a very good company as we run the company very well. We're very focused on doing an excellent job for our customers and hopefully providing excellent returns for our shareholders, and we plan to continue to do that. Moderator: Well, great. I really enjoyed speaking with you today, gentlemen. Keep up the great work. I look forward to following it more. Again, thanks for being here. We all appreciate it. Seamus Grady, Chairman and CEO, Fabrinet: Thank you, Mike. We appreciate it. Thank you. Thanks for everyone for participating. Thank you. Bye. This article was generated with the support of AI and reviewed by an editor. For more information see our T&C.