Thank you, everyone, for joining, and welcome back to our 28th annual Needham Growth Conference. It's a great pleasure today to be joined by Penguin Solutions CEO, Mark Adams, and CFO, Nate Olmstead. Thank you, guys, for being here.
Thanks for having us.
Maybe just to kick off here, oh, sorry, one more thing, just housekeeping item. There is a question bar underneath the presentation here. I'll be monitoring any inbound, so please feel free to log in, or you can email them to me directly. But maybe just to get going here, for investors that may be new or revisiting the story here, can you guys give us a quick overview of the company history and value proposition?
Sure. Happy to do that. Thanks, Matt. Penguin Solutions was formerly known as SMART Global Holdings, or SGH, which at the time, about five and a half years ago when I joined, we were primarily a memory module company under the SMART Modular brand. That business has been around for decades, since the late 1980s, and there had been a shift over time to acquire some additional assets beyond memory that were not super relevant at the time that we joined. Over the last, I'd say, five and a half years, the company has morphed away from memory, diversifying into more of a compute solutions business. We did diversify our way out of memory with the sale of a Brazilian business, which was a consumer memory business that didn't fit the enterprise solutions model.
We also heavily invested in an acquisition of Penguin Computing at the time, which was a high-performance compute business. We've spent the better part of the last five and a half years shifting our strategy away from manufacturing-based SMT products to AI infrastructure solutions. You might ask, well, why did you do that? Why was it relevant? It turned out that HPC has been a great precursor for AI. If you think about what goes into high-performance computing, it's large data center clusters supported by environmental solutions such as cooling and power networking. It's all kind of integrated into what's commonly referred to as an AI factory environment. We build out, we design these components, we build them out, we implement them, and in a lot of cases for our customers, we manage them.
The growth of our business over time is really to develop software and services that we integrate with our hardware platforms in many cases to build out these AI factories. So the shift in our business that was once over 80% memory has been where advanced computing has actually been a bigger part of our business as we've increased our deployments away from hardware-based products to hardware and software and services. That's allowed us to dramatically increase our gross margins in the business from below 20% five years ago to, on average, 30% or higher in the last few years. It's really afforded us to build a much better, more resilient business, great financials, solid balance sheet. Right now, we have zero net debt and continue to grow the business.
Had a coming off of a fiscal year that ended the last Friday of August in 2025, where we grew total revenue for the company 17%, and Advanced Computing was higher than that. It's basically this transformation that we've gone through that we're now continuing to invest in to build out a roadmap to increase and expand our funnel of customers, both in enterprise and in sovereign AI, that we think is a great opportunity to grow and build this company going forward.
That's great. Thank you for the background there. And sort of with that, the backdrop of that evolution, how do you think about Penguin's growth strategy over the next two to three years?
I think the growth is going to be driven by a shift away from our legacy customers, where we had a large concentration in a hyperscaler-type environment. And so the growth we're seeing right now is in the market. There's a shift away from just large language model only, hyperscale, customer-concentrated business to more enterprise deployment growth. And that's driven by custom language, large language models, as well as inference application, where we believe that that growth will play a lot more to Penguin's strength. If you look back at the early consumption of GPUs, it was very concentrated in the likes of Microsoft, Meta, Google, Amazon, and the likes. There's a lot of infrastructure built. Can you hear me okay?
Yeah.
I can hear you okay. Can you hear me now?
Yeah, I'm fine.
All right. Sorry about that.
I saw you leaning over. I didn't know if you couldn't hear me.
That was not my end.
So effectively speaking, we think the market's shifting in our favor as companies that don't have these large infrastructure teams rely on companies like Penguin. And that's where the proxy for us is, or the metric for us is, how is our pipeline and bookings looking over time? And I think we commented most recently on our earnings call last week. Our pipeline continues to expand. We're winning new logos. It's all very effectively a good growth story for us. And we believe that we obviously need to convert this pipeline into future bookings and revenue. But we think the quality and the type of engagements we're talking about are exciting. We announced two customer wins that we got bookings for in Q1. We also announced two awards, which are basically bookings waiting to happen.
But we've got through two master agreements with large projects at two Fortune 100 companies, one in oil and gas. It's a new logo for us, and then one in the financial services, which is also a new logo. But again, both significantly large enterprise-type environments. And so those data points are confirming the value proposition of being a trusted advisor in these large-scale deployments of systems, software, and services. And we've got a good track record. We've got some of the largest U.S.-based AI deployments, our Penguin Solutions. And it's an exciting time for us as we look to expand and grow. But it's really the enterprise growth. And then we referenced on the call last week also that we've seen some really nice larger-scale opportunities in what's called sovereign AI, which is normally well-capitalized AI infrastructure on an international, mostly a country or region basis.
Those two are exciting because a lot of times these are countries that don't have the resource or the capabilities or just the skill set and assets, and they have to go out and find it somewhere. We've gotten invited to the dance a few times. We just came off our first international deployment in August and early Q1 of this year in Korea. We've got a number of other opportunities that we're looking at that could also represent growth. Between enterprise growth and adoption of AI, moving from proof of concept to full production and larger sovereign deals, we think that plays well to our growth opportunity over the next 12-24 months.
We kind of just hit on this, but like you said, the first quarter was solid from an execution standpoint. Anything else you want to highlight of what went well during the quarter and where your focus is as you look to the rest of the year?
I think one thing that's interesting as we keep on going down this transformation curve, we sold the remaining interest in a Brazil consumer memory business, which just didn't fit the story. The enterprise memory business is doing quite well, both in terms of current operating performance, 40% growth year over year, largely due to favorable memory demand in both our legacy customers and in AI customers. And then we're continuing to invest in AI enablement with memory in products like Compute Express Link, which is also referred to as CXL, or our investment and innovation around Optical Memory Transport. And that is exciting as well, a little bit more out into the future. But memory continues to be a very key part of our AI infrastructure solutions portfolio. And we think that will be a key asset for us going forward.
In the past, memory kind of had a consumer halo and a negative headwind in terms from an investor standpoint. We got rid of our consumer business, and all of our business today is in enterprise.
Yeah. So to stick on memory for a second, it was a strong quarter in Integrated Memory. And we understand that Penguin's role there is as a value-added reseller, targeting early adopters of technology like CXL and OMT, along with the longer lifecycle projects and custom builds. Is there any way to bifurcate out where you're seeing strength right now, or is it pretty broad-based?
Still in our core compute networking telecom business is kind of where most of the revenue is today. CXL is a market that's just coming online right now in terms of revenue. We have some early-stage revenue. We just received our biggest order ever, which was a $10 million CXL order for an AI silicon systems company. They're a combined vertically integrated AI infrastructure systems company. And so it makes sense the market's kind of evolving over time. I think CXL is a good example of how great technology needs to be designed in and ramped and what have you. And we're in that early stage of the design and ramp process. We are qualified at two large OEMs. There's other orders from an enterprise systems company. So it's early, but we're well-positioned.
But by and large, the majority of our revenue today is still in our core compute networking telecom business.
I saw yesterday too, you guys announced you're CXL compliant for one of your memory modules. So congratulations on the announcement there. And can you help us understand what that means for the business and how that, I guess, broadens the opportunity here?
I mean, I think for us, any of the products we design that will be compatible and enabled or compliant with the CXL solution set, it does broaden the aperture where we can sell our products. The market adoption of CXL right now is, as I said, in its early innings. We're going to be enabling as much as we can because we want to foster that adoption. We're definitely on the early stage of innovation and a leader in CXL solutions today. That has not been the history of the company. We've been traditionally a fast follower just based on the business model we're in. I would say this CXL opportunity is exciting for us because, yeah, people are now coming to us as a leader in CXL.
And so we want to make sure that the enablement and compliance of all of our products where it's applicable expand the pie, so to speak, to help our customers implement.
And what are you seeing in the memory market broadly? We hear about rising DRAM pricing and stuff like that. I guess, what are the trends there and how's that impacting the business?
Needless to say, I mean, it's been an unprecedented time in memory in terms of demand and the supply favorability. What I would say is it looks like for the next 12 to 18 months, and I'm just telling you what I see in the market, it looks like it's going to be pretty tight supply conditions. We are a buyer and we are a seller of memory, right? When you think about our business model, we do see constraints. So far, we feel pretty good in terms of the supply and ability to meet our demand. As I said, we grew 40% year over year, Q1 to Q1, 2025 versus 2024. Sorry, 2026 versus 2025. So we're watching this very carefully.
One thing I would say is that in the memory business, the manufacturers make the least amount of gross margin on high-volume consumer-based applications, and none of which we're selling anymore, and so the products that still remain fairly in good balance are the higher margin products, and we have a value add on top of that, and so our ability to gain supply is definitely more favorable than someone who's in the mobile phone or PC business, per se, because that's the least attractive from a memory portfolio for the semiconductor companies, and so we never really have run into severe memory downturn, and we're watching it, obviously, but our ability to grow is largely based on our relationships with our supply chain partners, and I would say that we're confident we can continue to grow.
That's great. And then to jump over to the advanced computing side of the business, one of my main takeaways from the earnings report was the ability to grow that business and beat Street forecasts despite not recording any hyperscaler hardware revenue. Where do you credit the success in being able to diversify that customer base to? And how sustainable do you see that motion?
Yeah. So look, I think that's a great point that we try to emphasize a little more because I'll give you a case study. Two years ago, we had a very similar situation, and we didn't have the pipeline or customers. In the, it was kind of where we were in the evolution of the business. We couldn't react. Well, it wasn't even react. We didn't have the customer depth to be able to offset basically what's a 20% headwind going into 2026 revenue-wise. And I think Q1 shows us that we are adding customers and it's starting to make a difference. Maybe just around this question, maybe you can kind of talk about the tracking of non-hyperscale revenue and how it's evolved over the last year or so.
Yeah, sure. So within the advanced computing business, we've got a few different products. But the main thing, about 75% of it is the HPC, high-performance computing and AI business. Of that, if we look at the non-hyperscale business, in FY25, that grew 75%. The non-hyperscale AI business grew 75%. This first quarter, it grew more than 250%. So it's starting to get bigger and more material. It's off of a low base, but it reached about nearly 25% of total company revenue. This last quarter was non-hyperscale HPC AI. So now we're getting to the point where it's a much more material business for us, jumped from five to six% in the first quarter a year ago. So we made a lot of progress throughout FY25. But we think that we can do even more of that in the back half of this year. And yeah.
When you think about the back half of the year you guys spoke with you, what exactly are you seeing in the pipeline that gives you confidence in that second half?
I think as we stated on the earnings call, there's two things. The number of opportunities that we're engaged in are increasing, and then the quality of these wins that we're having are high quality. As I mentioned, we were awarded two Fortune 100 transactions with new companies, one in the financial industry, one in the oil and gas. We had two logos that we did get, so in the quarter, we had Q1, two new logos that we shipped in the quarter, so what I would say is, whether it be bookings or the pipeline that hopefully leads to bookings, the quality of what we're seeing in our pipeline is dramatically better than it was a year ago, and that's largely due to the dynamic of the growth of enterprise from proof of concept to production rollouts, which is early, early innings for sure.
That plus some new sovereign cloud opportunities that we're engaged with, it's a combination that the level of activity is just more. And we see the potential to drive a better, more diversified customer engagements.
When you think about that quality of the pipeline, and we kind of talked to how the volatility and diversifying that customer base led to the wide revenue range you guys issued and reaffirmed on the quarter, what steps are you guys taking on your end to increase pipeline visibility aside from just, like you said, as the opportunity grows?
Well, I mean, look. I think as we validated our strategies working, we've hired better leadership and sales. Most recently, we hired our first-ever company CRO because when we were part of a holding company, each of the business lines had their own sales go-to-market. We've shifted that and hired our own corporate CRO in charge of the engagements. His name is Thierry Pellegrino , formerly handled major accounts at NetApp. So he was an SVP of major accounts at NetApp. And we brought in just some better talent who's more suited for the solution selling model. And also on the marketing side, our involvement with key partners in terms of building out those relationships, whether it's NVIDIA or CDW, we're doing more tactical stuff and strategic stuff around identifying new customers.
I would also say that our success at Meta, at Shell, at other places, at Georgia Tech, in a world that didn't have a lot of enterprise success because it was all hyperscalers upfront, the enterprise adoption that we were fostering has put us in a really good realm of people who know about us and are expanding us. One of our largest pipeline opportunities today with a large sovereign AI transaction opportunity came because someone left Meta, went over there, and they said, "You have to use Penguin.
They're the best at what they do," and so whether it's word of mouth based on our own performance or it's channel-based or it's hiring better people or going to more unique trade show or marketing activity, we have built out a program of going to market that's got a lot more investment behind it now as we've seen the market grow, and I think it's starting to benefit.
I'd imagine too, there's somewhat of a snowball effect in this business too, right, of once you can point to, "Here's what we did for this company at scale. Here are the results, how quickly they got into that." It makes it a little bit easier to win that next incremental customer, right?
Agreed. Yeah. No doubt.
And so as this momentum builds and things progress through the pipeline, where should we be looking for signs that bookings are coming in and progressing, I guess?
I think just as we report every quarter, we talk about kind of new customer logos. I think that's one because we're trying to cure a business that had relatively high customer concentration. That is improving. But I think if you think about our business focus, it's this diversification away from just a small customer set. New awards, new customer company logos is really critical. I also think how we report the business. You mentioned earlier this Q1 data point of, "Geez, you lost 20% of your revenue, and yet you delivered a pretty solid quarter up year over year slightly." We think of that in those terms, so our ability to continue to deliver against that and meet our guidance, which is coming off of a 17% corporate growth year.
I think our guide was right around 6% for 2026 with the 20% headwind factored in. As long as we continue to acquire these customers and convert pipeline into bookings into revenue, I think those are the data points I'd be looking at.
In addition to that, our deferred revenue will grow as we get services renewals and services bookings. Then on the inventory side, we don't break that out by business. Some of the inventory increase that you'll see likely comes from memory with the higher prices. The inventory gets valued higher from building a strong backlog there. From a company perspective, one can look at inventory, though, as one indicator of bookings and backlog building. That's very helpful. Thank you. To zoom out again to the total revenue guidance, while that was maintained, you shifted around some of the underlying components there. What exactly are you seeing on the optimized LED side that made you a little more cautious?
The LED piece for us, it's not exactly core to the AI infrastructure business we're pivoting to. And the macro environment in LED is just not very favorable right now. There's massive oversupply from a manufacturing footprint perspective. Demand for LED is not as strong as it has been in the past. It doesn't mean it won't recover. It's just that's where we are right now. We had some tariff headwinds based on some of the activity out of DC and the guidelines there. The macro environment is just not great. I mean, Samsung exited the business. That's public. Lumileds is in bankruptcy. That's public. It's just not a great world. Even the Chinese manufacturers that were subsidized by the government are struggling. And so we're trying to operate this business for cash flow and operating margin preservation and not trying to just get profitless revenue.
And so we're doing that effectively. Our gross margin of the business are great, probably if not the best, one of the top gross margin businesses in LED in the market. We don't own our own factory, so we're not sitting there saddled with high CapEx in the business. We generated a profit. But the demand for products out of China, which is tougher, but also even some of our tier one OEMs has been challenging. And we're going to look to continue to put that business in a better place, whether it's through a partnership of some sort, similar to what we did in Brazil, or run the business as we have been with profitability and cash flow in mind, not trying to grow the top line at all expense. Nate, any other additions?
Yeah, I think that's right. I mean, I'm pleased with the profit improvement that we've seen in LED. And I think the business is operating well in a pretty tough demand environment. And we'll continue to be focused on being lean and protecting the IP while we fight through these macro challenges. I wanted to hit on profitability too. So that's a good segue for you guys. So you reduced gross margin guidance, which is a factor. You got things like Penguin Edge being phased out of the business and then growth in memory and AI hardware, which is a little lower margin. But at the same time, you lowered the OpEx forecast. Where are you finding success driving leverage? And how are you thinking about that moving forward?
Yeah, I think, Nate, I'm going to let you get into more detail. But we've had these different businesses. I mean, we probably didn't touch on that enough, but we came from this holding company mentality. And there was a lot of what I would say stranded costs and inefficiencies in the business. We would be doing that regardless because there's just more efficiency in the business. We don't need two HR, or we don't need an HR executive in every line of business that we've had in the past when we acquired something. We had all these IT systems. We're now over to one SAP system. So when you look at the business model itself and the shift from a holding company structure to a more traditional corporate structure, there's just natural efficiencies there.
And then we're shifting some of our resource spend outside of higher cost areas to lower cost areas. Those would be the two things that I would say as we think about how we're optimizing OpEx. But Nate, you probably have more you want to add?
I think you hit on some of the good detail points. I'd say completing integrations of past acquisitions has kind of been one area, is one way to think about it. As a holding company, that maybe wasn't the emphasis. You wanted to run these things as sort of standalone businesses. But like Mark said, consolidating the ERPs, consolidating some of the back office functions and things like that has provided some benefit. And then I think anytime that you're a portfolio company, you're always evaluating, are you investing in the areas with the best long-term growth potential? So that's an exercise we're continuously doing. So we'll find some efficiencies in some areas and then reinvest that in others. In some cases, we're making investments, by the way, Matt, up into cost of sales and to service delivery, right?
Because that's an element of growth for the business with good returns. So sometimes we're taking those efficiencies that we find in OpEx and just reinvesting them. Maybe it's sales coverage, another big focus for us, software development, another big focus for us, those types of things.
Got it. Yeah. Yeah. And I think it's a great point when you guys talk about the shift out of the holding company model into this more unified approach. So we talked about the CRO hire and steps you're doing there. What else needs to be done from an operational standpoint as you guys think about attacking the opportunity here? I think the biggest thing for us is to continue to build differentiation into our offering in software and services. We have that today. If you look at our gross margin when you compare it to a Superm icro, I think they just guided to 6.5% gross margin. And our business is dramatically higher than that.
In order to continue to add value to our customer relationships, we want to be building out capabilities for future needs, orchestration, data analytics, security, and then provide additional services that really enhance our value add as a trusted advisor in AI implementations, and so I think that said, we don't see any large capital spends per se. It's mostly in capabilities around the further future demands of our customers' needs and matching that with what we're investing in.
Understood. I want to hit on some of the demand drivers too. One of the things you guys talked about is how you're starting to see signs of customers moving towards inference use cases and sort of out of this training environment we've been in for the last couple of years. What exactly are you seeing there?
Really, when you think about the use and AI in general, for AI to be effective, these large language models have to be of high efficiency and quality and accuracy and what have you. The production environments are really based on inferencing. They're using AI as a tool to drive better decision-making, better productivity, better information, and so in the financial sector and the oil and gas sector, companies are investing in these types of AI implementations that are really commercially viable as opposed to just traditional ChatGPT type engagement, and so this shift away from large language model training per se to real production by virtue of the applications that people, the workload that people are developing, it is more of an inference solution that is allowing people to use this underlying technology to drive their business behavior.
As those conversations shift too, where is the understanding of where they're going to pull budget for these things and do they have established? Or is that at this point pretty locked in as they've kind of been thinking through this? Or is it still kind of trying to find a source for this stuff?
I think it's a good question because I don't think there's one answer. I will tell you that I'd say over the last 12 to 24 months during the shift of, "Yeah, let's put some money into exploring it," to now, "Let's build out, let's roll this out." I think there's more confidence in the return on investment in AI than there was two years ago. I think two years ago, it was kind of cute and let's go explore it. And people, quite frankly, didn't understand the return. I think in many cases, people going on to the next phase of production environments, they've been able to see or at least have confidence in a significant return on investment in their AI deployments.
I think there's also this kind of interesting phenomenon, which is like there's not many generational AI opportunity in general, the level of scrutiny in the boardroom and at the C-level of these companies because it's a competitiveness that if your competitor has something and you don't, it could be a material impact on your market position and your business overall, and so there's a lot of energy to figure this out. I don't know that I have a good answer on where the money is coming from in every company because it's going to be different by company, but I'll tell you, it is a good example. In frequency trading, we're finding that seconds are everything, and better information requires, allows for better information to make these decisions, and so we're seeing a large commitment to invest in AI in that vertical.
I think that's also true in oil and gas. It's also true in healthcare, which I think for us is a much newer vertical. We haven't had much success there. We've got some early pipeline opportunities we're involved in. But I think it comes down to when the quality of information and the timeliness of information is critical and your competitor might be driving a better solution to provide more timely information. I think it's kind of, AI is going to be top of mind for people and how to integrate that as part of the value proposition. And we've seen that mature dramatically over two years. Two years ago, it was a lot of tire kicking. And I think today, we're seeing a lot more momentum in the enterprise to shift away from that to true production environments.
Yeah. No, I agree. That's consistent with what we're hearing across our coverage where it's funny how it's sort of not that it was cyclical by nature, but people got all excited about ChatGPT and then you almost get fatigued hearing about all this AI stuff. And now it's like, wow, we're really starting to see use cases and.
I think it truly is starting. I mean, it's not like we're in the fifth inning of a nine-inning game. This is early stuff, which is exciting for us because we can see the amount of inbound activity and just companies from different walks of life finding us and wanting to work with us, and it's an exciting time.
Absolutely. I want to make sure we hit on S cyld ClusterWare before we run out of time here too. So version 13 release came out in November. What were the highlights of that release? And what are you guys seeing on the adoption side there?
It's early stage for that release. It was just 60 days ago. We're trying to build in more capabilities for companies to use virtualization in software. And what does that mean? Most people think of virtualization inside of a cloud environment where a sovereign cloud or a neocloud customer can splice out the resources and have multiple customers using that. Even in large companies today, in large enterprise, they want this level of virtualization to be able to share resources across different departments who might be using AI. So one of our big focuses in life is to make our software more shared resource-friendly so that when people build out these AI factories, they can offer it across the enterprise and scale up an implementation that way.
And when we think about Scyld ClusterWare, we're trying to build in more of that features over time as well as in future releases target things like orchestration and security. And so if you combine all that together, we're building a toolset that when we deploy, we can use some of these capabilities that match our customers' needs along those feature sets.
That's great. Unfortunately with that, we're out of time. Mark, Nate, thank you guys so much for joining us here. It's exciting times for Penguin. We're really excited to keep following the story here.
Great. Thanks for the support.
Thank you.