Global Technology Conference. My name is Ruplu Bhattacharya, and I cover EMS companies, electronics manufacturing services companies, as part of Bank of America's IT hardware and supply chain equity research team. Today we have the team from Flex, and I'm honored to have CEO Revathi Advaithi here. Revathi, as you know, has joined the company. I think she joined in 2019, and she's done a phenomenal job on two fronts. One is like an amazing job on margins, which, you know, have really, really gone from the low end of the spectrum to the high end of the spectrum. She's also created a lot of value for shareholders. You may remember the Nextracker acquisition and then divestiture. Revathi, thank you so much for coming today.
No problem. Thanks.
I think a key topic today will be the CEC business, and we're honored to have Rob Campbell, who leads that segment here. CEC, for those of you who don't know, is the Communications, Enterprise, and Cloud. Even I have to look at the acronyms. Thanks so much for coming. I think we're going to have a lot of discussion on that. We also have Michelle Simmons, who has just joined the company as Head of IR. Michelle, thanks for coming today. Revathi, we have a lot of things to talk about, but I want to ask you about your vision for the data center business. How are you positioning Flex to compete in this market? What is your strategy in data center, and where do you think, what are some of the competitive advantages that Flex has?
Yeah, maybe I'll start, Ruplu, with kind of just talking about our vision for data centers, where we started with a few years ago and kind of where we are today. We're one of the only suppliers today in the data center space that does both IT integration and power products both. Usually, people will do one or the other, but not both together. What put us in this position was, this was well before power became a thing for data centers and, you know, NVIDIA became a thing and all that, and AI was a top topic. Flex had an embedded power business that was basically designing and building the power that powers the chip. Having come from the energy space myself and companies like Eaton, I'm very intrigued by the fact that, hey, that was not done by any of the electrical folks.
This vision that someday power intensity in data centers is going to become a big thing. The people who are building that power and then finding a way to integrate that most seamlessly will become critical. If you think about data center customers, the folks who design the IT infrastructure are different than the folks who design the power infrastructure, and the two almost never meet, right? The view was, hey, imagine the person who's going to put all this together because as power becomes more intense, cooling becomes more intense, you know, how you build this and scale this is going to become really important. Our business really is uniquely positioned with the sense that we build the IT infrastructure and we design and build the cooling associated with it.
We design all the power that powers the chip, and then we design and build all the power that's coming into the data center itself. Today, hyperscalers are putting that technology more and more together. They were not doing that four years ago, by the way. They were not thinking about this. Now they are, because when you are going from a 30 kW rack to a 1 MW rack, the complexities of it are pretty significant. You have to bring in people who are understanding power, how to cool it, all of that to build this infrastructure out. That is how our thesis started. Today it is playing out because we are in those technology conversations, and then we build that end to end.
It's around a $5 billion business for us, and we have kind of said that before that in our earnings call that it's growing 35% this year. Really well ahead of everything that you're hearing from the space itself. It is very beneficial from a margin standpoint as a creative to Flex overall. That is a little snapshot for what Flex does in power from and data centers itself. Very, very unique positioning today that is becoming more of a reality from a technology interplay.
Yeah, no, that's a great overview. Let's build on that. You talked about some staggering growth rates. I think for the data center business last year, you grew 50% year on year. Like you said, you've guided 35% year on year. Help investors get comfortable with the long-term high growth rate in these segments. Like, what are the secular drivers in compute and power and cloud and power that are actually going to sustain these levels of growth?
I would say the most important thing is look at what's happening with AI, right? I mean, we have barely kind of touched the tip of that in terms of the growth that you're seeing from AI and what the potential could be. You know, I would say we believe in that macro pretty significantly. We believe that what the world's going to see from an AI infrastructure build-out over the next decade is going to be fairly significant, and that is going to be compute hungry and power hungry. I would say nobody's going to debate that macro. I was in a CEO conference last week, and everybody unanimously says how we use AI, we've barely started, right? You see that in our own lives.
We believe in the macro of the AI infrastructure build-out, and we also believe in the fact that, hey, power density is going to be super critical and is going to be a part of this build-out that's going to be pretty significant. I'd say that's why you should get comfortable with it. When we gave our long-term guidance a couple of years ago, last year, sorry, we said 20% growth through the cycle. That's because, in general, we are conservative. Then we did 50% last year, and then we did 30%. We're going to do 35% this year. The view is, hey, long-term, even if this is a 20% growth business, that's fairly fantastic, I would say, in terms of what you're seeing in this space.
Let's talk about the relative growth rates between power and compute. I think this year you guided compute to be a little bit, or cloud to be a little bit lower than 35% and power to be higher. When you think on these two end markets long-term, you think power outgrows cloud or vice versa?
I don't think it'll be a one versus the other. I think the power growth is going to be related to capacity additions. And power could have grown significantly last year also, but we were flat out of capacity. This year, some of those capacity additions are coming on board, which enables us to drive the growth. It'll be all related to how much capacity at what pace can you add in to deliver that power. It's not one versus the other. Rob's business is bigger, so it's also the larger numbers, right, in terms of growth rate.
Okay. We are going to go as deep as we can into the power business and definitely into the cloud business. I want to talk on one higher level item, which is everybody's talking about tariffs now. Can you give us your thoughts on what you're seeing in terms of pullings or end market demand impact from tariffs and how much of your manufacturing is in the U.S. versus in Mexico versus in China? How is Flex being impacted because of tariffs?
Yeah, I would say that if you think about tariffs, in the last cycle of tariffs that came up five, six years ago and everything that happened with COVID and supply chain resiliency, the world started migrating from a supply chain manufacturing perspective. There has been a tremendous amount of transition over the last five, six years. You can see that in Flex's numbers, whether it's in terms of our square footage and footprint, right? We have moved more towards North America. We have reduced our Asia footprint, whether it's from our revenue growth. You've seen that. What is happening in this cycle is that continuation, but some acceleration in customers who have been slower in that transition.
We are getting a lot of customer calls, I would say, who want to go faster than they were going or who were not thinking about it or are now thinking. Our challenge is we have added a lot of North American footprint. We have to be picky in terms of where we want to grow and why we want to grow. We are looking for complex products, long-term stickiness, better financial results for Flex overall. Those are the kinds of customers we like to grow with. Tariffs and unfortunately, maybe a bad thing for a lot of people, but the supply chain resiliency conversation it creates turns out to be a good thing for folks like us. In general, I would say our view on this is continue with the footprint build-out that we have been doing in North America and help our customers accelerate that transition.
Maybe just a quick follow-up on that. I mean, do you think everyone's focused on manufacturing in the U.S.? How realistic is that? I mean, are you seeing customers trying to move into the U.S.? What needs to happen to really enable that shift in manufacturing?
I would say definitely in complex products that are difficult to make and hard to transport, you do see that migration happening in the U.S. that may not be that labor intensive, right? Consumer products, all of those things that make it in the news cycle, I would say less so, right? That's really harder to do because labor is constrained in the U.S. Rob's business, yes, lots of conversations happening around continuing to build out the U.S. footprint. Power business, the same thing. My belief is on consumer businesses and all, it's more the hype cycle than reality.
Okay. That's a great segment. Let's talk about the CEC business. Rob, for those of us in the audience who don't know this business, talk about like what exactly does this segment do? What do you manufacture? Who do you manufacture for? If you can talk about like, you know, different segments within this, the cloud business versus networking versus enterprise.
Sure. I think it's a great place to start. As you mentioned, CEC stands for Communications, Enterprise, and Cloud. I think Revathi mentioned it's the largest business unit in Flex. We'll take each one of them. The communications piece of ours, and by the way, we've also said that's the largest of these three business lines, is the communications piece. That is really comprised of two major segments. One, I would say, is the, call it the networking piece. By networking, I mean optical networking, IP networking, fixed networking, and those things. Think switches, think routers, think optical switches and things like that. That's one piece of the business that's there. The other major piece of the communications piece is the wireless piece. Think telecommunications, think RAN, think, you know, Open RAN, radios, and so forth.
That's the other piece. The final piece in that is kind of an up-and-coming segment, and I'd even call it wireless, but satellite communication. Satcom is becoming quite important to us. We have a number of customers in that satcom space. That all falls under the communication piece, those pieces of the business. The enterprise piece for us, it's a combination of, it's the main server storage. Think your basic servers, think basic storage. That's probably the bread and butter in the enterprise space. It also includes the silicon providers out there. As they have to go to more and more board-level products than just chips, right? We service them out of the enterprise segment. Also, if they're doing NIC cards, usually the low-end NIC cards, most of it's the silicon providers that do that.
We do not really do a lot of that, but as they are coming out with the newer, right, AI-enabled NIC cards that have the bigger processors on there, taking the load off the CPUs, that falls in there. The final piece that is in there is the cybersecurity. We provide the hardware to just about all the top cybersecurity companies out there. That is enterprise. That takes us to cloud. Cloud for us consists of the hyperscalers. That is where we put most of our focus. With that, it is the U.S. or Western hyperscalers. We also deal with what we call tier twos, right? Maybe just a level below the hyperscalers as well. That falls into our cloud business. The other thing I would say is we, about six months ago, acquired a liquid cooling company called JetCool.
JetCool, their liquid cooling, direct-to-chip liquid cooling and CDUs, that falls into CEC. And it actually applies, you know, you'd think for sure, right? The silicon guys use it, but the hyperscaler guys, and now even, right, those enterprise, the high-end switches, routers, and so forth, they're all going liquid cooling as well.
Okay. Let's focus on the cloud business a little. I think there's some confusion maybe with investors. I know most investors know that you have one large hyperscale, but how broad is your business in this hyperscale community? Can you talk about exactly what are you guys building in that cloud business? Are you building individual servers? Are you building boards? Are you building full racks? Give us a sense for the type of business you're doing.
You know, it's interesting. We had a number of meetings today and so forth, and there's this misconception that we've got one large. We don't. I think that's pretty true in the EMS space. Most of my EMS competitors have one large. That's not true with us, right? We have, of the four U.S. hyperscalers, we have engagements with all four of them. We have very, very deep engagements with two of them. We've got probably a unique position in the way that we've engaged with them. That's why I say it's number one. What do we provide for them? It's evolving. By evolving, I mean as we provide different vertical integration opportunities for them, the hyperscalers are basically integrating more and more of our vertical options for them. We may just have started off building a server board for them.
We may end up just doing or start off just doing rack integration for another, which, by the way, for I think a lot of folks in the EMS space, that's what they're doing, one or the other. We will take that server board, for instance, and it actually goes into an enclosure or a tray, and we'll actually build that enclosure tray, bare metal, right? That may go into a rack. We actually build the racks that go in there. You know, there's various components in there, mechanical components that go into the rack and the board. We provide those through a company that we have in ours called Coreworks. All the way, you know, if you continue up the scale, right, we'll take a rack and do a payload configured rack, which is configured with all the power that's required to do it.
We will do, in some cases, a full L11 rack build integration and so forth. I mentioned before cooling, right? We could provide the cooling, not only the cooling, the direct-to-chip cooling in there, not only the CDUs of that, but the actual racks, right? We will manufacture these racks and so forth to handle liquid cooling. There are a number of services that we will provide with these things as well. We might be building a power distribution cabinet, and then you have to do the commission of that cabinet and so forth. I am just kind of going through some of the various things that we offer the hyperscalers.
What I can tell you is every quarter, every year that goes by, the relationship we have, they tend to engage in more and more of our products and our services in addition to the EMS work that we do for them.
A couple of months back, there were some media articles that hyperscalers are reassessing their data center plans or maybe holding off on new leases. How should we think about Flex CEC segment revenues? Are you directly related to CapEx spend? How correlated is your revenue to that? The second part of the question is GPUs versus custom ASICs. A lot of hyperscalers are now focused on their own ASIC, their own silicon. You know, how is your business trending? I mean, do you see opportunities to do both standard GPU servers as well as custom ASIC servers? How do you see that evolving?
Sure. I'll take both of those. The first one, you mentioned how does CapEx, right, affect us? I can tell you there's definitely a correlation, right? If the hyperscaler is going to spend more in CapEx, there's probably a pretty good chance that they're going to spend more with us. You know, that said, we certainly don't get a vote in what their CapEx is going to be. As I mentioned, the type of products and services that we provide, most of what they're going to do from a CapEx perspective is going to somehow entail Flex. We definitely see a correlation there.
What I can tell you is just from what we're seeing, you referred to some of the stories and so forth, and I think those stories were followed up by, if you're trying to give some clarification, is we're not seeing any slowdown, right, in demand, in forecast, and so forth. We're not seeing that from any of these folks. As far as, you know, why some of those things came out, I'll let them address that. As far as the second question on, you know, custom silicon, you know, versus, you know, OEM GPUs and so forth, we're heavily involved with the hyperscalers on both. We're actually quite agnostic as to, you know, do we do or do we prefer doing custom silicon or off-the-shelf silicon. We get engaged with both.
We're involved with the silicon providers and providing the embedded power that is needed to power those from there. We're engaged with them as far as the cooling that's required from there. If it's custom silicon coming and being developed from one of the hyperscalers themselves, exact same thing. We're, you know, we'll build the board for them, but we'll also talk to them about the embedded power almost. I can tell you that, you know, we have from the power side, very heavy engagement with all four hyperscalers, right, on power products that we're providing for them. The net-net of all that is we are very engaged and very happy to work with them on both custom silicon as well as off-the-shelf silicon.
Okay. Let's transition to talking on margins. One of the things that I was surprised and pleasantly surprised was that you guys are guiding the 6% operating margin a year ahead of time. You know, it's strange, Revathi, I always thought of reliability as the end markets that have higher margins, but agility has been producing higher margins than reliability. How should we think about, you know, the long-term margin potential for these two segments? Rob, if you can talk about within agility, how should we think about the relative margins for this cloud business versus networking and versus enterprise?
I'd say just in terms of the margin story, you know, across Flex and across each of the six business units, we have really focused on kind of, you know, fixing the mix of the portfolio within each of the business units, which definitely drives a margin improvement. Then there's relentless focus on disciplined execution and operational excellence that drives margin improvement, right? How I think about margin long-term is that those two factors won't change because we'll continue to shift mix. You see that how we have done that in data centers. Within health, we continue to shift mix. We want more med devices, less med equipment. In almost every business segment, we're trying to shift mix in terms of which customers we go after and why. Overall, the margin story should continue to, I'd say, evolve the way it's evolved for Flex overall.
The reliability versus agility story, I'd say, is, you know, the way you should think about it is that, yes, you know, reliability should have better margins longer term. I think it's had its challenges, right? I mean, when you think everything's going well, the automotive segment kind of starts to move around, which is higher fixed costs, so it drives kind of margin challenges. I would say our health story has been fantastic from a margin perspective. Our renewables business, which is also in reliability, has seen a pretty significant impact over the last few years. There are some fundamental drivers that are compressing margin for reliability that, as you overcome those challenges, it should have tailwind for reliability margins. Agility margins, I would say, driven by two significant things.
One is Rob's data center business and how that's growing, not just in terms of accretive margins, but the things he's doing for integration and, you know, vertical integration and bringing things like racks and cooling in. Even in areas like consumer products and lifestyle, we try to only address customers who will give us services because that drives end-to-end margins. We are really driving this idea of how do you service a customer and who gives you value in all our business units. I say that our margin story has just started, at least in my view, and it has a long ways to go. You can talk about data center margins for CEC and kind of how that's going.
Sure. I think it's important to note, and it's very basic when you say it, but the more complex a product is, right, the more value Flex is going to bring to the customer, and more value typically equals more margin. You can just go from that to say in the data center space that that is a product that is typically pretty complex, pretty leading edge, you know, and so forth. We're providing a lot of value. Therefore, the margins that we get in the data center, as Revathi mentioned, are accretive, right, to our margins as a whole.
Part of what helps drive that, in addition to the complexity of the products, is some of the things that Revathi just mentioned in our consumer and lifestyle products, things like that, is, you know, when you can take those complex products and now start adding services to it. In services, we might be doing direct fulfillment, we might be doing configured order, we might be building a product for them and doing commissioning, we might be doing decommissioning, and repair and warranty. These services all are, you know, again, significantly accretive. As we do more and more, which all of those, right, are opportunities for us, which we're capitalizing on in the data center space. You can also start layering in the actual product portion where it's, right, branded product with IP from Flex.
Whether that's in the liquid cooling space, whether that's in the embedded power space, whether that's in the critical power space, all that helps add, right, accretive margin.
That's a good transition. Let's talk about the power business. Like, when you say you have a product portfolio, typically EMS companies have shied away from having their own products because they don't want to compete against OEMs. But this is different. I mean, so what exactly, can you talk to us about what are the products that you're providing in terms of this power business? What is the target customer set? And how do you plan to grow this power business?
You want to take it?
Sure, I'll take it. I would say, you know, the customer set is pretty broad, but we really have focused a lot into the data center space. Whether that's the hyperscalers, the colos, the tier twos, that really is the space that's there. Your comment earlier about, you know, trying not to compete and so forth, your customers, I mean, that's probably a reason, you know, why don't we build and brand our own switches, right? Let me tell you, I think I probably build the majority of the switches in the world today for just about every switch. I would absolutely be competing against my biggest and best customers out there. In the power space, you know, while some of the major players are customers to Flex, they're also suppliers to Flex. It's not, you know, that deep relationship in there.
The competition is not as head-on, if it makes sense, you know, to say that. In that space, one of the things that we've seen is, and as we've really gone up, and we've used this little phrase from grid to chip. We really are. When you're coming off the grid and using mid-voltage switch gear and then low-voltage switch gear and, right, internal of the data center with busways and then power distribution cabinets and metering and, you know, all the way until you get to the rack and using power distribution cabinets, power shelves, and then the embedded power, just that entire flow from the grid to chip, it really puts Flex in a unique position.
When you then couple it, as Revathi started off, right, talking to you, Ruplu, with, oh, by the way, we also do all the IT portion as well, right, which is what all falls under me. As power density is going up, right, and therefore heat is going up and everything else, all of a sudden now, right, the electrical portion is having to talk to the IT portion, and it is really all flowing together so that the customers, whether it be the hyperscalers or the tier twos, they are now having, from the time they start deciding what they are going to be doing two years from now, they are having to get their power folks together with their IT folks. Oh, by the way, there is really only one company out there that you can call on the phone to talk to both pieces.
It's been a big advantage for us.
We have about two minutes left, and I want to cover two topics. One is, Rob, talk to us about this consignment business and, you know, how does that impact revenues. I want to end by, Revathi, you talking to us about what is your vision for Flex and why should investors invest in Flex now? Is there anything that the market is missing? How do you plan to create value, whether it's in terms of share buybacks or M&A? What is your view on shareholder returns, capital allocation, how you plan to do that?
Just very quickly on consignment, what I'll say is, look, we meet customers where they want to be. We have customers that we do full turnkey and all we do is turnkey. We have other customers where we do consignment, a lot of consignment and so forth. Either model works for us, right? We structured it so that it works for us. You know, what we have stated is that the percentage of our revenue that is being affected by consignment is going up. We've talked about that. It's gone from 11% two years ago to, I think, 14% or 20% now. It is definitely growing. When we talk about the growth numbers that we have, 50% last year in data center, by the way, data center is driving the bulk of that.
35% this year, those are the numbers after we take out all that consignment revenue, right? We are able to still have that type of growth even playing the fact that that consignment revenue is not counted in that.
I'd say my answer would be that if you think about Flex in the manufacturing services space, which is most of contract manufacturing, our goal is to do complex things, charge a premium for it, attach services to it, and continue to grow margins. Our productivity journey has just started. In the data center space, the technology migration is unique from a power standpoint, from a cooling standpoint, and an IT integration standpoint. We're a $5 billion business today that's growing at 35%. Our goal is to be the technology player in that that brings those unique products together to solve these kind of heat and power problems that we are seeing today. We expect that in both areas, we want to excel. In many ways, like I said before, I feel like our margin journey has just started and we have long ways to go.
In terms of capital allocation, M&A versus
I would say it would not change. We have done a really good job of balancing, you know, M&A investments with a pretty good share buyback strategy. We look for value in M&A, and we have done a good job of balancing both, and we will continue to do that, including a very nice kind of organic CapEx investment that we have done. I would say no change in that strategy. It has delivered great shareholder returns in the last kind of six years, and we expect to follow that same playbook.
All right, great. I mean, look forward to great things from Flex. Revathi, thank you so much for joining us today. Rob, thanks for all the details. Michelle, thanks for joining. Thank you for being here.
[crosstalk] Thank you.