Super Micro Computer, Inc. (SMCI) Earnings Call Transcript & Summary

Hi, everybody. Thank you for joining. Tim Long here, Barclays IT hardware and comm equipment analyst. Happy to have Super Micro, Michael with us today. I appreciate the time. I know it's a pretty busy one. You go through the safe harbor, you're good.

Oh, yes. I forgot about that. Please refer to our safe harbor statements on our website for -- regarding forward-looking statements.

[indiscernible] memory. You need..

shortener version [indiscernible].

Okay. Great. All right. We've got some stuff to go through here. Maybe we'll just hit on kind of the hot investor questions that we're getting. Maybe let's just recap. Obviously, last quarter, you had some pushouts and maybe just walk us through kind of what caused it and how you read into that as we look forward into future deployments?

Yes. So from that perspective, I think if you look at what's going on in the end markets, in the industry, there's a lot of dynamicism happening and we had some configuration changes that delayed some shipments. Obviously, we're -- I would say, obviously, we're through that. We talked about the December quarter at the time when $13 billion in orders for GB300, which is a pretty massive number, and we moved the needle on the guidance to from $33 billion to $36 billion. So visibility, demand all look really good. I think we've seen some evidence recently of others in the spectrum talking about increased CapEx. So that environment is cooking right now.

Okay. And then just talk a little bit about these configuration changes. They've happen before, not just the Super Micro to others in the industry. It just seems there's a lot more complexity. So at the end of the day, does this -- do these type of dynamics, obviously, it's a pushout, not great in the near term. But is this dynamic of things getting more complicated, a positive for Super Micro? How do you view it?

So we view it -- there are some folks that talked about standardization and what we're seeing from an end customer perspective is that everyone has some degree of differentiation or complexity. So if you think about what we're trying to do, not every customer is going to run a GB300, not every customer is going to run Vera Rubin in the future. There's multiple verticals and stacks that NVIDIA is trying to support. They're similar with AMD, similar with the ASIC, the ASIC debate, and there's multiple players there. You have Intel's in the mix and you have ARM in the mix. And so what we're trying to do is support an application optimization for our customer base. And it's big and broad, and the market opportunity is large. And so where we're going as we move forward is to support AI applications in mass where our competition might be just building one product for one solution. So we're doing that. And at the same time, we're solution setting around it in DCBBS. And it all goes down to the point of like complexity. And so future racks are going to look different than the traditional rack, right? There might be power side cards, et cetera. There's lots of things. Liquid cooling is a pretty big component. And so we're supplying all the componentry that surround that and the margin profile there has improved. So as we move forward, we'll be a full stack supplier for AI and other enabled application solutions. So it will make Super Micro look a lot different as we go out of several quarters, if not 1 or 2 years. And if you believe the $1 trillion market that some folks talk about or [ 3 or 4 ], and we're just 10% of the market where we kind of are today. It's a $100 billion opportunity that we'll get to serve as we move forward. And that's tremendous. And again, it's scale and then we're all in on building that for our customer base.

Okay. Great. maybe you mentioned the really positive orders in the revenue outlook. Just how do you -- how do we think about visibility? Obviously, this is a business where there's big chunky deals and something can change late. So how do we get comfort in kind of linearity? Or is it -- do we get to a point where the orders in the backlog book is so big that there's a little bit more flexibility to build to it?

Well, that would be great from a standpoint of visibility, I think we have very good visibility and we think that, that will ultimately come to play. So this gave us the confidence to go from $33 billion to $36 billion for fiscal '26, which is a June year-end for us. So I think we're -- our guidance practice here will reflect that as we move forward. So I think we can just articulate that more precisely as we move forward.

Okay. You mentioned the industry worries about standardization and gross margins, and you mentioned some of the things. If you just elaborate a little bit on what you think develops over the next few years that enables the Super Micro to maybe improve that gross margin percentage on these big AI type deals?

So in the short term, we have a couple of different levers, right? One, customer mix has had a pretty big influence. As you can imagine, there's been a few operators that have been building out and scale, give me as much as you can -- as soon as you can. So customer mix there and some of these customers are large and have buying power. You have product mix and then you have manufacturing efficiencies. And so as the customer base broadens out, and we talked about this in our last call, we said that for fiscal '26, we would anticipate adding 2 to 4 more scale customers on top of the 4 that we already have. we would grow to 4, and then we would add more scale customers. In Q1, we added a new scale customer and grew with significantly one of our larger customers. So we're tracking from that perspective. But as more and more customers come into the flow, there's an expansion in neocloud space, there's an expansion with respect to sovereign opportunities. As a customer base expands out, we believe that, one, they will be able to mix up with product. Two, they'll have less -- potentially less purchasing power. And three, as we lean in on the manufacturing efficiencies, we can deliver a better. We'll iron out some of the inefficiencies. And if you think about this quarter, December, at least [ 10.5% ] as a brand-new product in mass coming through to the factory. So there was a higher cost, as you well know, following hardware, any launch has overhead. And so this has just happened to be a lot of overhead because of the size and the quantum of this and which lends to the opportunity as we move forward, ironing that out, supporting a massive amount of differentiated product sets and adding on the DCBBS, that's building a whole new scale company as we move forward that will support. Again, we go back to the 10% of the $1 trillion, that's a big opportunity for Super Micro and there's margin opportunity and the operating structure is fairly lean as we do this. So we're moving fast to address a large opportunity, and we think we're advantaging ourselves across the board relative to our competition.

Okay. Yes, it does seem some of the -- your large competitors have seen stability in overall margins, and it seems much more predictable now than it was maybe a year ago. Would you agree with that or in general?

From our perspective [indiscernible] I would imagine that we've been -- the dynamics have been so intense that we've been guiding margins for a quarter at a time, right? Again, we've just basically have built a brand-new system at scale in one quarter that we didn't do the prior few quarters, right? So as we move forward and as things change, it's hard to predict where that will be. But the goal is to move the needle on that to the point where we're double-digit gross margins as we move forward. And that will come as a factor of, again, customer mix, product mix and manufacturing efficiencies. So all those things we're working hard on to move those needle on that as we build a bigger, better company.

Okay. Sounds good. You mentioned in their DCBBS, data center building block solutions. So I just got to make sure I got...

A lot of letters. It's a lot of letters.

Maybe just walk us through kind of what that is, how it's differentiated and what do you think that's going to mean to the company?

So complexity has been -- has grown in the data center, we've been moving forward with that, right? So if you look at the past, what we have always done is more performance in the rack, better power delivery, so to speak, or power savings, more performance, et cetera. So more packaging, more complexity. So this is moving beyond just the rack. And so we see our customers potentially needing more of the -- whether it's cabling services, whether it's the power element, whether it's liquid cooling element, all these things wrapped around. So we have -- I believe we're up to multiple dozens of SKUs to help enable a data center build for each unique customer opportunity. So we're trying to encapsulate as much of that for the customer as we move forward. There's battery backup. There's power shells. As you look at some of the newer rack and development, power side [ card ], their liquid the air cooling elements. So we're putting all those together, depending on what the customer is, what the customer need is. So building a business off of enabling data sets, including the service element to them. And storage and ultimately, switching will be a part of the complex for them in various different segments.

Okay. I mean it's still pretty early. Do you have any sense of customer feedback and what at your bigger customers that might be looking at this or liking about it?

so we've scoped out a few from that perspective, and we are already delivering to some of our customers, some of those cooling elements to help them lower their -- or increase their PUE, right? So there's some early signs of that moving forward. I think the next milestone markers will be when will enable a few of the sovereigns, and we have equipment POCs in many different locations around the world. As those expand because we think that the sovereigns are likely not to run in hyperscale environments, right? There'll be some segregation of workloads, those will have more requirement for our full stack, and we'll grow with them and that should be reflected in results in future quarters.

Okay. Great. Maybe if we could touch on capacity a little bit. It's an industry-wide topic for everyone in value chain. So I know you guys you're building and increasing your rack output. So maybe walk us through the time line there and what that's going to do for revenue opportunities?

So we have multiple locations, Silicon Valley, Taiwan, Malaysia and the Netherlands. And we're looking for more sites in Americas. And we've been expanding capacity. Currently, we're planning to exit fiscal '26 at 6,000 total racks, 3,000 of them will be liquid cooled. And from that perspective, we will be unlikely to be satisfied with those results, right? We've continually added capacity as we move forward. But if you take a look at what the liquid cool racks, ASP commands, it's a pretty big price tag. I'd say, it's a $3 million per rack kind of price tag. It gets you to a pretty ample revenue opportunity if you do the math. So we're cognizant of that. But we're also cognizant of making sure that these facilities are producing the most reliable and efficient systems for end customers. So our capacity plans are in line with what we think is the opportunity easily to be able to support the figures we talked about earlier.

Okay. And you mentioned the $1 trillion or $3 trillion or $4 trillion. So this is not the end. I'm assuming that it's [indiscernible].

Well, It's not the end, but at the same time, if you -- like the dynamic -- if the opportunity set continues to expand and again, it's tied to the application. We're all in San Francisco right now, and you see the Waymo's buzzing around this city, but if that becomes a nationwide thing, there's to be multiple vendors supplying autonomous driving vehicles, whether it's the transportation sector for trucking, et cetera, it's a huge opportunity, right? So naturally unit volume should expand, data center capacity should expand, right? So -- and our partners, whether it's NVIDIA or AMD or [ beyond ], are talking about very large market TAMs, and we're in support of all of those. And so those are the numbers that we're kind of benchmarking with you off of. And again, we're roughly 10% of the market right now. We have a much bigger share, I think, of AI equipment than that number. But if it continues in that path, we'll certainly need capacity to serve it. And then there'll be capacity requirements for the DCBBS element, right. That's where this non-rack that it's not liquid cooled. So we'll focus on both of those things. And the unit volumes, et cetera, we'll be able to support the end markets.

Okay. I wanted to take some time to talk about customers and concentration. And you mentioned typically having 2 to 4 scale customers and adding another 1 last quarter. How diverse and it sounds like you're going to add some more, how diversified do you see the business be coming if we look out a few years?

Well, this is an interesting question. You've been covering the space for quite some time. And if you think of the traditional data center enterprise business, right, which I think a lot of analysts are looking at it from that lens, if we're looking at it from the lens of application, application optimization and you think about the context of the hyperscalers and what they're doing, they have multiple different applications that they're delivering to their end customers that are running on different various forms of equipment that need a certain amount of support. They're starting to take workloads and put it into the neoclouds. And that's an amazing kind of element there. So we have a large footprint with the neoclouds and they're expanding. They're offering services. At the same time, they're getting services from the hyperscalers. Enterprises are testing out AI workloads in the neoclouds, whether or not they bring them in-house or not, remains to be seen. We do expect them to do that to control their own data. There will be enterprises that are data-centric that all want to support that as well. So you think of the famous application companies that could turn whatever their enterprise application is into an AI source. So we'll be able to support all of those things. So the customers base is, from our perspective, is anyone that cares about application optimization. So we're not so focused on, is it an enterprise customer? Is it -- what segment it is. But most customers now are aligned to application optimization. So it opens up a big world for us. So we have neocloud, we have enterprises and sovereigns is a grouping. Customer concentration right now is a little bit skewed because the land grab is on for certain players. So they've been moving fast, and we expect the rest of the market to catch-up and backfill, right? So we'll have plenty of opportunity with a broad product set. That all said, by [ virtue of ] success with some of the known customers that we have, large technology companies, I think most folks know that we have something to do with the [ Oman ] complex, right? So very forward-thinking application development there. Operators understand that we're applying and supporting these customers, and they're coming to us. So there's really no large-scale customer that's not at least engaged with us in some respect, like what can you do for us and -- from either, a technology perspective or even a capacity perspective. So we expect the customer base to broaden out over time. Again, we have to if we're going to support trillion plus on the market opportunity in a fairly short order of time.

Yes, I get the focus on the big applications, but there's just so much focus on when the sovereigns are going to start hitting and everyone seems to be focused on sovereign and enterprise because there's the perception that they'll need more technology help and therefore, stack is better for Super Micro. So maybe just touch on where you are with those verticals?

So my understanding on that -- from that perspective, on the sovereign side, I know we have engagements across the globe. We have proof of concepts. We have agreements. I think the major hurdle -- not a major hurdle, but a large hurdle would be pre-Thanksgiving when there was quite a bit of a [indiscernible] in the White House and licenses being granted or at least approved for, say, Middle East -- Middle Eastern regions. So I think there's a lot of progress, and we'll see more signs of deployments as we get into '26. We're close to it right now. and the proof points will become obvious. And whether they deploy immediately or through a partner, yet remains to be seen, but there's a lot of activity, and we're involved in, like I said, almost a large amount of those. Okay.

You did mention the $13 billion last quarter in GB300 orders. Could you just talk a little bit about the complexion of that? Like how many big customers? Is it new -- some of it, new footprint, some of it is expansions?

On average, I would say it's -- there's a couple of customers in that cohort, but one was fairly large. So -- and I think there was a new -- there was some new and there was some old. So it was mix footprint.

Okay. I just want to touch on like Edge AI solutions. I think that's something you guys have talked a little bit more about. Maybe give us a sense how you see Super Micro playing there as we get more inferencing in Edge?

So we focused on solutioning for those type of opportunities with some of our partners, key partners. And so it's still early and still all to come, but if you would consider the RTX solution from NVIDIA. There's other elements and other ways to get to those markets. So we have lots of activity with -- on the IoT side or the Edge side. So we anticipate that to be a pretty big volume market opportunity for us downstream. Right now, I think the focus has been on the data center, data center building blocks, and that's an extension of that strategy, and it's on the come.

Okay. And then you mentioned before, not just NVIDIA, but AMD and custom ASICs. So what are you guys seeing as far as diversification from just the NVIDIA GPU engines?

I think from that perspective, we have deployments with AMD and scale with some of the large customers that are NVIDIA customers. And I think that they're taking those workloads that that are capable of running at scale and supporting the customer. So the AMD supporting the customer and the customers' workloads. And so I think there's ample opportunity for us. We're excited about the forward map that they have as well. And we see customers requiring them for different applications across the stack. So for us, we're in a very good position to be able to support either one. And like you said, there's customers who are running both of those solutions. I don't know if it's a dual-source strategy, if you want to call it that. But the ecosystem is expanding and ultimately, potentially that gets into the ASIC market and the multiple players there.

Does that present a different set of engineering challenges or...

I would imagine there's a lot of ability to leverage on the engineering side. They'll all be different, and we have half of this -- the force is engineering, and that's a key strength. And it's becoming more and more important, and we have teams working on different platforms across the stack. So that's a really big focus, and it's -- I think investors should understand that if you believe that this broadens out and there's different use cases, in different verticals or if there's a different model for specific industry verticals, we'll support that. And if you get a better cost per token because you're using Super Micro's engineered solution, we have a customer and we tend to retain customers quite for quite a long time because we work with their application, and we continue to make the evolution of technology to meet with they're trying to deliver customer. Tesla is a great example of that.

Okay. I do want to go back to maybe like kind of competition industry. It seems like you guys were pretty early in like rack scale type deployments, but now you have the 3 big semi companies doing racks themselves, you have most EMS and ODM focusing on instead of the piece parts, the full rack. So help me understand like that industry development and what positions you guys -- I think you've been doing it longer, but what positions you better there?

So what positions us really, really well is what our focus is, is to make that customer happy, right? And at each step of the way, I mean, we started out as a motherboard company. Every step of the way, it's like there's just more technology, more to do. And even the new platforms that everyone's talking about and like, well, this is going to be a standard. It's like there doesn't appear to be a standard anywhere at this point in time. We're enabling that technology. And if we can find new ways to either lower the power [ draw ] or improve the cooling element or enabling bandwidth or enabling connectivity that's in a better, broader way for the customer, I mean, that's all gold for us, right? And so we're doing that with all the different partners. And you can imagine the engineers are like kids in a candy store, trying to figure out what to do, at the same time, there's -- need a focus on the optimal cost solution. So the complexity increases, it favors what we're doing because we're all in on complexity. So that's one element. The second element is it's very understated is the reliability, the quality of what we're doing when we put these things together. These systems multitude, millions of parts just in a rack, but if we integrate all these things and make them work together collectively, and we can lower or improve -- lower the power by 2% or improve the performance by 1% through some architecture change of some sort or improvement. And you do that across scale for a scale operator, that's amazing for them, right? That's value. But at the same time, if the system is more reliable than someone who's basically assembling, you have increased reliability and the operators that's saves them bottom line costs. So better reliable equipment is a value proposition. So the combination of these things as we move forward, are going to be magnified. They're going to be magnified by multitudes of platforms. And if you look at it, who will be the leading supplier in 2027 for AI optimized applications, I would argue that we stand to be in a really good position to be that dominant supplier. We've shown increases in market share for the past couple of years relative to some of the brand names that you know well. And we continue to focus on the customer. And that's where, where the real value comes in, keeping the customer satisfied, and that's what we're optimizing for right now.

Okay. Great. We covered a lot on gross margin. You get that all the time. Curious about as you branch out into other technologies and get the building blocks together. It seems like there's a lot more engineering or investment intensity. So how do we manage really kind of R&D and OpEx profile in the wake of probably a more limited gross margin dynamic over time.

Well, I think we've always been a fairly lean and frugal organization while scaling up. And so that cadence and that focus hasn't changed, there could be periods where it's a little more intense, but the goal is to keep a lid on that and make sure that we're within our boundary, so to speak. So the efficiency of the organization is pretty excellent, which kind of goes back to supporting all the platforms and all of a sudden, we are the de facto standard, and there's no competing platform to offer changes the pricing dynamics. And with the sensitivity of the model and the change in the dynamics, they should translate to better returns for shareholders. And so that's the path that we're on, and it's a long-term view, but the company is all in on doing that and at scale, at a fast pace with the intensity of competition. And we seem to be winning more and more, witness $13 billion in orders for just this one platform in December alone.

Okay. Yes, maybe just one other back to kind of the customer side. You talked about -- just curious of your sense on how sticky the Super Micro platform is when you're in at a scale customer. There was a period where it was like someone wins a big deal and someone else wins a big deal. Everyone is obviously going to be dual sourcing or triple sourcing because it's getting so big. But talk a little bit about that stickiness and what is it that, that will keep you in a strong position at some of the bigger players?

So from that perspective, it's really making the customer happy with the technology and the reliability and the service element of that, this engineering support of that side of the house. And we've had some scaled customers that were not so scaled, not too long ago. And we continue to be not just in the hunt, but a dominant supplier, and we've grown with those companies. And so we think that, that will continue and the retention rate of our customer base is fairly high. Sure there could be sometimes where there's some puts and takes, but it's very high, and we look forward to satisfying those customers and keeping those customers. Now if we didn't keep those customers, at the same time, you'd be asking me a different question, right? So we're all about the customer and we keep them, we retain them, we grow them. And we're looking for ways to help them, and that kind of really led us to what's the next enablement, right? The next enablement is like the data center building blocks enablement as opposed to just providing them with one component. So that encompasses a lot of different platform opportunities for those customers. So keeping them happy, keeps them in the seat, and we've done a really good job by witness of the revenue growth over the past couple of years.

Okay. Great. I think we're bumping up against time here. So thank you, everyone, for joining. Mike, thank you.

Thank you. Really appreciate it. Thanks a lot.