Welcome to the Supermicro Fireside Chat at the Goldman Sachs Communacopia and Technology Conference. I have the privilege of having Mike Staiger, Senior Vice President of corporate development at Supermicro. My name is Kat Murphy, and I cover SMCI and IT hardware here at Goldman Sachs. We have about 35 min for today's presentation, inclusive of Q&A at the end. To start, Mike, thank you very much for being here. Appreciate the opportunity to get to talk to you in front of this audience. Let's start on the raised fiscal 2027 guide. Really impressive to see on the earnings print last quarter, expectations for $65 billion-$72 billion in revenue, which implies another year of substantial growth on top of revenue nearly doubling or more than doubling in the last year. Can you talk about the major building blocks that are supporting that outlook as we look forward to next year?
Yeah. Great. Hey, thanks for having us. It's good to see you and catch up. I think it's not lost on anyone that demand is actually pretty excellent. I think a couple of years ago we were $15 billion, and now we're on our way to doing $15 billion for a year, and now we're on our way to do $15 billion for one quarter and beyond. The building blocks, obviously, the foundation here is that AI adoption is underway. We're leading the charge with innovation. We have a whole host of different various solutions that we're bringing to market with our partners. The order book, obviously, at $60 billion, underpins the guidance for the year. We're pretty focused on a scheduled build-out, so to speak, for our end customers. There's a diversification within that customer base that's really helpful.
We see it broadening out, and it looks to be a multi-year cycle, and customers are looking for solutions as opposed to components, so to speak. It's pretty awesome.
In terms of thinking about this growth going forward too, you underline the healthy demand, the diversification of customers. How should we think about, maybe put more simply, parsing the growth between TAM expansion and then Supermicro gaining share? You're obviously going after this opportunity in enterprise, which is not a place you played historically. How should we think about the balance of those two driving the growth next year?
There's multiple layers of that, which is awesome. But the TAM itself has been segmented out as $2 trillion-$4 trillion, and whether or not the market truly believes that's the TAM. But if we're thinking it's $2 trillion-$4 trillion, and we're 10% of that market, and we've said this before, we're 10% now of the market itself, which implies that we're a significant enterprise player. We'll go back to that. But that gives us a $200 billion look, and I think that there's more opportunity for us because as we move forward in this market, our focus is on solutioning the customer. If we're solutioning the customer, by nature, we should be gaining share. I think the underpinnings of this is that we're transitioning from a traditional compute stack that people have been following for years, server storage, networking, et cetera.
The deployment model could change to next generation innovative solutions, which we're leading the charge. This is why this has been driving the growth, and customers are wanting application-optimized solutions for AI and beyond. We're developing those with a whole host of partners, and we can get into that later. From that perspective, that underpins the growth on the innovation front. We look to capture more customers. We're focused on the customer as opposed to how much of the market share we can gain, so to speak. There's a lot of nuance to that, but I guess we can dive into that with whatever next question we have.
Yeah. Maybe a part of that would be helpful if you could touch on. Supermicro's right to win historically has been your engineering-led model that's been differentiated relative to some of the domestic OEM peers that you have. As AI infrastructure becomes increasingly more complex with every chipset, with every requirement, how does your engineering focus, do you feel, differentiate you versus some of your peers?
Well, I think it's major, right? The time to market, the time to engineering, the time to solutioning, excuse me, is a material differentiator from what we're doing. So when we working with multiple large name partners, many of which were here today, they all have variations of their different chipsets and different platforms, and those platforms all need a different motherboard architecture, a framework, and we're able to manage or understand that element and put it into a product, into a system. So you take any chip, any server into any rack, into any data center as a whole total solution or an AI factory. The engineering piece of this is what leads customers to us and which leads us to gain share. And we've always done this. We've done this in the enterprise day one, back in the days of X86.
Well, X86 is still the platform, but a big platform. We were able to put more compute in a system that with a lower power draw, with better thermodynamics at a lower cost. At the time, no one cared about power and cooling. The DNA of the company is all about packaging these solutions to enable the customer to do more with their gear. So if you take that ethos, so to speak, and you move forward, and you can productize that, and you see the AI platforms actually requiring what we're able and capable of doing, where half of our staff is engineering, our focus is on this. We're sitting in Silicon Valley, all the major partners there, we can flex incredibly fast. And so we're bringing brand new technology to the market in volume and in scale, and that's a great viewpoint for customers and they notice.
Part of the mix story should also help one of the main investor questions around Supermicro, which is around your margin profile. Incredibly strong margins last quarter, beat guidance expectations, I think 17.5% gross margins. And you've talked about returning the margin profile for the company back to stable double digits next year. What are some of these structural changes within the business versus maybe one or two years ago that gives you line of sight into that stable double-digit margin framework? Is it mix? Is it the plan being executed as expected? Anything to share there to give us confidence that Supermicro will get back into that range sustainably?
Yeah. I think the backdrop of the market in early days, there was a few large buyers and a lot of folks who wanted to serve those buyers. There was a lot of price competition in that segment. I think many of the larger buyers realized that when they were getting, I hate to use the word, systems or reference architecture, that they had to do a tremendous amount of work to stand that up. I think one of our partners has mentioned that if you have 8 racks and it takes you 30 days to make them operational, it's like $3 million. Some of these operators are looking at 100 racks to 1,500 racks deployments. You're talking hundreds of millions of dollars if you cannot stand up your infrastructure immediately. I think the end market's looking for time to online, time to revenue.
If you think about what's happening at the neocloud level, and you see some of these major contracts where these neoclouds were, many of them and growing, excuse me, that we're serving, they're getting hyperscaler contracts. In essence, our systems at the neoclouds are serving the hyperscalers. So they need to get those systems up and running immediately, time to market. When we solution those customers, there's a solution value. We're putting more content into those particular packs for the customers in the form of DCBBS. If you go back a year and a half ago, we told the market that we were going to do DCBBS building blocks, so we have power cooling, CDUs, rear door heat exchangers, storage switching, and software to integrate these things.
When we put together a system or a solution for a customer, the integration, the L11, L12 element of this, they can turn it on day one, and more importantly, they get time to revenue. But the more important element of this is the reliability of that integrated system is significantly, we believe it's significantly higher than competitors. If your system availability is there for you and you can charge dollars for it, downtime kills you, right? You're losing revenue. So they're gaining the capability quicker, and they have greater reliability, so the cost savings to the customer is significant. So that's the solution value. That's supportive of the double-digit gross margin element that we pointed to the market. We got into 10, well, let's say the midpoint 10.6 for the quarter.
There is an element of, "Oh, there is a downtick and there's some contracts in there that potentially drug through from the prior quarter that put some pressure on that." But the focus is the solution to customer, and as we do more on the DCBBS and we have more components that are more integrated, that we're doing more ourselves, those margin profile is higher. So that mix and the customer mix, so we'll add the diversification of the customer base helps. We're doing more than just serving neoclouds, enterprises, and sovereigns. The enterprises and sovereigns, and inclusive of neoclouds, do need the solutions because everyone wants to get their systems up and running and serve their customer, because if they're serving their customer, it's a revenue generator or it's a productivity enhancement for the organization, and that's kind of the direction we're going.
That's what we're focused on.
You're talking about solutions architecture. You made reference to DCBBS or the Data Center Building Block Solution. For people who may be less familiar, can you define exactly what that is for Supermicro? To the extent that these are things that Supermicro is doing internally versus partnering with some third parties to bring in capabilities, talk about that mix and how it makes more attractive the offering to enterprise.
We were initially engaged with the market in the initial build-out, reference architecture, and making that racks for customers and delivering racks. I mentioned before that just delivering a rack technically isn't enough. We saw customers struggle with the cooling, with the power, with power shelves, et cetera, with implementing the cabling and the back plane of the cable management, the management software for the cluster itself. What we've done is develop product, productize those elements so that we can stand up those systems quicker. The more important element is if we control all those elements, then we can get to revenue to the customer quicker, and it's a lot easier to match order to time frame so that they can get online faster. So it's a product suite. It's inclusive of storage and switches. Downstream, we'll probably have more switches.
But at the same time, the software tightly connects this and tightly integrates with this. Ultimately, we will get to a point where we can go across, let's say, factories that were building solutions and do different architectures or different families and cross planes so to manage the footprint, because we think customers are going to ultimately have other GenAI on site or agentic CPUs or traditional CPUs. All those things will be very difficult to manage. So we are focused on downstream, making those things work collectively for the customer. So it is a total solution strategy for the customer. We have not given out what part would be the pack of each individual item. That would be pretty complicated.
But it is clearly working, and we have shown the muscle strength of going from a lower margin, or we had a headwind in core that scared people, to be honest. Now we are on a trajectory that the solution is looking solid from a margin perspective. So it is a great thing for us, and we are looking forward to doing more with our partners as more product comes on stream.
Gotcha. I want to touch on the software piece, but first could you just confirm on the ways in which you are selling these DCBBS products or components within that solution? Is this all Supermicro branded
Yes
cooling and power, or are there other parties that you are working with?
We design almost everything, and we do have partners that help solution these pieces. It will be a Supermicro design, and if ultimately it is something that we think that we can do in-house, we will do it. There are many other pieces that we do, and again, it is part of the solution. They are all integrated together, and they work better. Again, if a customer wanted to DIY, and we have seen this, there are all kinds of mismatches and issues and problems. We have been able to essentially productize this so time to online for the customer is significant. It is all about innovation and staying at the forefront of the curve, and we have seen customers go order various different flavors or types of GPUs from our partners that match workloads.
We are able to flex, and within whether it is NVIDIA or AMD or Intel or Arm-based, be able to flex to those products. The breadth of product is amazing, and the use cases are expanding or broadening out, and we think it is a multiyear cycle, and we have all kinds of solutions for those customers. That is why they are coming to us, and inclusive of expanding our go-to-market presence in the enterprise, our service capability in the enterprise, et cetera.
On the software front, you have talked in the past about early traction in the suite of management software that Supermicro sells. There is the SuperCloud Composer, there is the Data Center Manager, there is the Orchestration Manager. Why is Supermicro the right person to sell this to your end customers, and what are the other solutions that they may have? It is a very small share of revenue today, but it is a place where you are investing.
Yeah.
Anything to share on that front?
Well, so software and services, I think, in fiscal 2026, we just finished $538 million and growing, so it can be a substantial part of the business. It's obviously I won't say obviously. It's higher margin. Software margins are great. Everyone knows that. From a standpoint of the integration and all the control plane for the racks and the cooling, the storage, et cetera, it's internalized in that system, so it makes it seamless for the customer. We think it's a pretty essential toolkit. We've integrated that with VCF, and so that's VMware AI Factory. That's another proof point from a standpoint of partnering and an expansion in the enterprise, where the factories and the management plane from a VMware perspective, we can reach more customers in the enterprise.
Is this something that you sell or ship directly, or is it part of
It's part of
how you're winning the bid?
It's part of the solution, yeah.
Okay.
It'll go with our systems. Ultimately, I believe its potential to cross different brand clusters. We'll see. Right now, it's a total part of the total solution package.
Going back to this notion of DCBBS opening up the aperture of the types of customers that you can serve. Is this an opportunity to win new customers, or are you seeing your existing, maybe larger, potentially more cost-sensitive customers looking now at the broader scope of Supermicro's offering and maybe engaging it in different product lines and features?
I think the first engagement process is I need an AI solution or I need AI Factory, I need a system. It's kind of a tailwind to all of that. It's inclusive of it as opposed to a point product. I will say that we have gone to customers and upgraded or enhanced their footprint, whether they needed some cooling solutions or some power solutions. We can do that, and we can probably do that across multi-platforms. We've done it. It really starts with a customer putting in a new deployment or upgrading the existing deployment that will, inclusive of a total solution for those customers.
I want to go back to talking about your strength in Neocloud. Some third-party data estimates 40% market share in the Neocloud AI- server business for Supermicro. There is an investor concern that as some of these most mature companies continue to grow, the most mature Neoclouds, little.
Yeah
weird to say.
It is what it is.
That they will try to engage more with a Taiwanese ODM or a lower cost operating model. You talked about the importance of time to first token, but how does that change over time when the scale of the deal may be less favorable from a unit economics perspective for Supermicro?
That concern was around at $7 billion, at $15 billion, at $22, at $39, and it is going to exist at $68. There is a certain function that they may serve. Again, shipping racks to customers. But we feel that there is a significant amount of customers that need that solution. We will get them up to online, time to online quicker and faster with the L11, L12 validation. As the complexity continues to increase, it becomes more difficult for customers to keep pace. We are keeping at pace.
The second phase of this is that the notion that there is going to be one platform that everyone lands on is clearly not the case when you see hyperscalers trying to do their. They are doing their own silicon, and a number of platforms that are available to put into a rack or to put into a solution or a factory is significantly expanding. We are expanding with all those types of solution sets, and that fragmentation or expansion is something that would be difficult to theoretically mass produce. We are just going to serve the customer with the solutions, because we will be there and ready day one, and if it is part of the pack and we are providing value, we will be fine.
And there is business for them to do on the ODM front, and there is plenty of business for us to do from a solutions front. The distinction between ODM and OEM, we are just trying to serve the customer, right? We are trying to enable our partners, and I think we are doing a fantastic job of that.
On the point of chip diversity and even the custom silicon that's coming out of some of the hyperscalers, what's Supermicro's opportunity to go after, I think Charles has in the past called it the design for manufacturing opportunity set. Is that still something that's in the playbook, or is that less a focus now that the enterprise opportunity is really starting to get some legs?
The interesting thing is there's the engineering desire to solve problems and work with the innovation. The engineers are all in on that. There's things I'm sure that we're doing that are not necessarily in the public domain, so there's nothing that we're not capable of. It isn't we're going to move away from this or that, but right now we're serving quite a bit of. We have many partners that we're serving, many customers that we're serving, so our hands are pretty full in attacking that. But at the same time, customers will continue to come to us and say, "Can you do this for us?" If we can come to terms and agreement, then I think there's plenty of opportunities.
When you think about the $2 trillion - $4 trillion number that's being parsed about, I think the sky's the limit from where we can be in a couple of years.
In terms of some of the constraints that are impacting Supermicro's ability to breach all of this demand. From all of the various different vectors of growth you are going after, how do you think about the engineering intensity, the potential investment that needs to be made in go-to-market manufacturing capacity? Where are investments that need to be made in the business in order to better capture the demand drivers that you are seeing?
From an operating expense perspective, during these phases, we have been very focused and have built in a lot of those process steps. We have discussed our capacity at 6,000 racks per month, 3,000 liquid cooled, 3,000 air cooled. What we have not disclosed necessarily is the new square footage that we have been adding in Johor, and the new buildings that are coming on stream, and a couple of the clean room facilities, the manufacturing capabilities that we have that if you think forward, what we are going to do with those, and we will update the market when they are more solidified. We are expanding pretty aggressively under the covers. I think Charles has always been a foot forward on what he sees, what we need to do to get there. Again, we have been expanding our go to market sales and focusing those.
We had some new leadership changes there that were contemplated for a while, and they are working quite well. I think we are staying pace with what the market is, and 75% growth at these numbers is pretty significant on a year-over-year basis. There is more to be done, and we are working on trying to get to that place because it is built into the model.
We've made it 22 minutes, and I haven't asked about CPU servers yet.
Yeah.
Can you talk about how this is. It's certainly always been in the core competency of Supermicro.
Yeah
A renewed focus by the market on how neoclouds and enterprises are thinking about the role of general compute. How did that factor into results in this quarter, and then the outlook going forward? How does that mix ultimately impact the margin profile of Supermicro as well?
The mix factor was beneficial for us in the prior quarter. I think everyone knows that. The enterprise customers and the CPU-based systems are important, a key underpinning of what we have done, and I think I mentioned earlier with Intel, I hate to use the word back in the day, putting more two processors on a board to four processors on a board, like, wow, that was really cutting edge at the time. Lowering the power draw was cutting edge. Making blades with 20% lower power draw was cutting edge. We cut a lot of teeth in that arena and have a lot of enterprise customers and a lot of great partnerships on the OEM side that are CPU based. As we move forward, we see this really solid opportunity, one, to continue with that, because there's an aged fleet of general compute or enterprise servers.
And we have said this before in the past, that fleet will need to be upgraded, condensed, solidified, whatever you want to call it, to make room for new capacity. If you listen to some of our partners and they talk about AI CPUs, they will be running the agents, they are going to be going into the models, and there is model verticalization. There is an expansion of applications in the space to just kind of create more of an opportunity for agents. So the agentic element of AI is there, and some are using traditional CPUs, and we see the opportunity for many to use Arm-based CPUs. So that opportunity set at the enterprise for a solution next to the core enterprise is significant.
So we are pretty excited about it, and we have all kinds of products on the come in this zone, and we will wrap it around DCBBS, and it will be an enhancer. And the enterprises need the help. They need the solution. They are not going to build it themselves. And even our larger neocloud customers will probably verticalize within their businesses, not just a model. They will have all kinds of different services, and we will look to supply them. So it is a really big opportunity. We are pretty excited about it.
A big part of enterprise, the difference relative to neocloud, is the need for help and support in both the integration, the deployment, the ongoing maintenance of these systems. How should we think about the investment that Supermicro is making to be more front-footed in that some of your legacy peers have very established services practices? But not necessarily the same model for Supermicro?
So this will be kind of interesting because as we move forward, could the services model change? Obviously, we are expanding go-to-market and services. But if you're on the agentic side, what will the service model look like? Will there be self-healing? Will there be other elements that we could do to augment that? Sure, we will have service personnel in the field, and if we need to scale up in particular areas, we will do that. So, we're conscious of expanding the service model as well as maybe some innovative solutions from a service perspective for some of these systems. So there's a lot to be done there, and there's a lot of systems. Again, if we are 200, 300 in size, there'll be clear evidence of us supporting those customers.
And then one last one touching on the OpEx side. In the expanding customer set, there's likely going to need to be a change in the go-to-market. You talked about some leadership changes on that side, but thinking about Supermicro's legacy engagements were driven by the strength of your reputation and the products that you put forward. But as you service more customers, you may need to engage with the channel, go through some other kind of vendors or partners. How are you thinking about that over time as you reach more into that $2 trillion - $4 trillion TAM?
I think the interesting thing is if you look at the IDC numbers, our share gain, our position, I think we are number two. We clearly have sights on number one. I think it is embedded in the organization, in the scale, and the growth rate that we have had. It just may not be as obvious for some reason, because the customers that we generally have are very technology-focused. They are using technology to lever their business, and there is a lot more of them that are doing that. By nature of serving some really sophisticated customer base with some really interesting product, and high quality and amazing reliability is driving more customers to us, which is forcing the expansion of the organization to serve even more. The success of our products are leading the way, and I think being product-focused first is very helpful to gain more customers.
We are expanding a pretty controlled way. Can we expand faster if we spent more money? We will see. We are doing a lot, and I think it is a positive for the organization.
Maybe I will ask one more, and then I will see if there is any in the room. In terms of the pace of expansion, something that to be mindful of is needing to go back to the capital markets.
Yeah
to raise financing. Can you talk about the balance there and thinking about certain levers like working capital intensity? Some of the factors that are in the broader equation when determining how aggressively to go after some of these opportunities?
I think that the interesting part of that is that capital intensity has been high. We have seen a pretty nice diversification of the customer base. More enterprises, there is more credit in the end markets, and we have done a pretty good job as we move forward with respect to terms with customers and timing of delivery to reduce the intensity on the working capital side. We are making conscious efforts to improve that as we move forward so that we can become self-funded downstream and get back to a point where we can return capital to shareholders. It is probably going to be a little while because the growth right now is pretty extraordinary. There is a lot to juggle, customers and capital conversions, and innovation.
But I think we are doing a really effective job at this point in time, and I think we are in good shape for FY 2027.
Great. Are there any questions in the room? We have a mic runner.
Just a quick question. To what extent are you starting to sell to, say, sovereigns as well as enterprises that want on-prem air-gapped AI systems? Is it minimal or is it growing?
No, it's great that you get to highlight that again. Enterprises are coming on stream. We have more than several sovereign deployments as we speak globally. They're clearly spending. I think the early motion was enterprise spending was into the neoclouds. Sovereign spending was into the neoclouds. We were capturing it early on, but we're seeing clear evidence of them moving forward and putting factories on the floor and expanding there.
Wow.
That opportunity set looks phenomenal. I think someone had asked, how big could sovereigns be? If you go $2 trillion- $4 trillion and you do the math, it could be a pretty big 10%-25% of the total market. They're fully funded, so they're wonderful in that respect. Enterprises, we're used to engaging with them, and so the newer customers that had swamped the market a year or two ago are now more mature. I think there's a maturation and expansion of this market that it's still early days that look really good from our ability to help push forward.
Any other questions from the audience? Another one for you. You talked about all the ways in which you are selling AI solutions to neocloud, to sovereigns, to enterprises, but Supermicro is an enterprise itself. How are you using applications of AI internally in a way that helps you inform your customer behavior and your customer relationships, and anything to talk about, especially as you're making some of these OpEx investments to offset some of those expenses?
Katherine, that's a really interesting question. I know there's broad-based use internally for various elements. But from a standpoint of a rollout of a specific program, on the engineering side, TBD, I don't know what they're doing. I'm sure they're using it in some form or fashion. But I know that we're very focused on the factory side from an automation perspective to improve several of the processes there. Those should be helpful, but I just don't have the exact what. Do we have a killer app here? But we're definitely using it.
In our last couple of minutes here, just in closing, what are some aspects of the Supermicro strategy that you want to highlight that we've talked to or we didn't maybe get to discuss today that you think are important to focus on for investors in the next, we'll call it, three to five year timeframe?
Well, I think the number one thing to think about is the innovation and the ability to bring any of the platforms to market for customers in a total solution package, and that the next generation of AI infrastructure has a different kind of cast to it. We're not impeded by any legacy technology that we have to continue to support on an ongoing basis for, let's say, for enterprises. It's all net new for us. That gives us a lot of degrees of freedom to innovate and bring other elements to customers that they may need. So we've been able to flex pretty quickly on understanding what the market does need. And again, the pace of change has been so phenomenal that the opportunities set for us to continue innovating is there, and I think that keeps the customer conversation very strong.
It's well beyond the days where you're just buying something off the shelf and you run it for and have someone service it, and that's it. This is well beyond that. So we're very focused on delivering that to the end customer. So I think that's a pretty misunderstood element of what we're doing, and it's important. It's a really important factor.
Great. Any other final thoughts for us?
No, I think we covered a lot.
Okay.
Thank you.
Thank you, everyone, very much.
All right.