Penguin Solutions, Inc. (PENG) Earnings Call Transcript & Summary

October 6, 2026

NASDAQ US Information Technology Semiconductors and Semiconductor Equipment earnings 61 min

Earnings Call Speaker Segments

Executive

executive
#1

Thank you all for being here. Kash Shaikh, Stephen Comin, and Aaron Johnson are with me on this call. You can find our earnings materials in the Investor Relations area of our site, and I'd urge you to go through them; I'd also ask that you spend a little time looking over the presentation, since it holds extra material that rounds out what we'll be discussing. On this call, except where we note otherwise, every mention of financial measures means non-GAAP financial measures. Non-GAAP measures shouldn't be viewed on their own, as a replacement for, or as better than our GAAP results. You can locate a reconciliation of those measures to the GAAP measures they most closely match in our press release and the slide presentation that goes with it. Anything said on this call about future results and events counts as forward-looking statements resting on present beliefs and assumptions, and such statements aren't promises of future performance. A range of risks and uncertainties, which we cover in our press release, our earnings call presentation, and our SEC filings, could cause actual results and events to vary materially from what's projected. Other than where the law requires it, we take on no duty to update our forward-looking statements; I'll now hand things to Kash.

Kash Shaikh

executive
#2

Good afternoon, everyone, and thanks for being with us on the Penguin Solutions call covering the fourth quarter of fiscal 2026 along with our year-end results. Before we discuss our results, I'd like to begin by highlighting a key addition to our leadership team. We announced earlier today that Stephen Coming is joining Penguin Solutions in the role of Chief Financial Officer. Stephen brings substantial experience as a public company executive, with a proven record in financial leadership, scaling operations, and Investor Relations. We're happy to have Stephen on board at such an exciting moment for Penguin, and investors will get the chance to know him over the coming months; on today's call, Aaron Johnson, our intra-CFO since July, will take you through the financial results and the fiscal 2027 outlook as planned. I'd like to express my thanks to Aaron for the leadership he showed during this transition, and I look forward to continuing to work alongside him as he goes back to serving as Vice President of Finance and Accounting. Now, on to our results. The team produced an outstanding quarter, capping a very strong fiscal year. During Q4, we achieved company records for net sales, gross profit dollars, operating income, operating margin, net income, and adjusted EBITDA. Net sales reached $567 million, which is 68% higher year-over-year; EPS came in at $1, up 133%, which reflects strong operating leverage that allowed EPS to grow at a faster pace than net sales. I'd also like to thank our customers for the confidence they continue to place in Penguin. Our AI-driven businesses kept scaling. In Q4, non-hyperscale AI infrastructure and integrated memory accounted for 78% of company net sales, with growth of 141% year-over-year. Our overall company performance accelerated meaningfully during the second half of fiscal 2026 after we launched our AI factory platform strategy and sharpened our focus on the data center market; we brought the business into closer alignment with robust AI-driven demand, raised our investments in product innovation, and sped up go-to-market execution, with emphasis on neo cloud and enterprise customers. The results speak for themselves. Following net sales that were relatively flat year-over-year in the first half, growth climbed to 48% in Q3 and 68% in Q4, producing second half growth of 58%. Full year net sales came to approximately $1.73 billion, an increase of approximately 26% year-over-year. We also posted record full year operating income. In AI infrastructure and integrated memory, bookings rose faster than net sales during the fourth quarter, and we closed fiscal 2026 with record company backlog and strong momentum as we head into fiscal 2027. Given our second half performance, bookings growth, the continued strength of our memory business, and, more importantly, accelerating demand from Neo Cloud customers for our data center AI infrastructure solutions, we are raising our fiscal 2027 growth outlook above the preliminary view we shared last quarter. We expect record fiscal 2027 net sales, with EPS growth anticipated to outpace net sales growth as we keep scaling and driving operating leverage. I'll go through what we're seeing in the market before I turn to the outlook. AI is turning into superintelligence, and it keeps shifting from a chatbot that advises into an agentic operator that carries out tasks and workflows with growing autonomy. That transition is pushing inference to scale and producing workloads that consume more and more memory. Inference is lifting demand throughout the whole data center stack. GPUs stay essential, yet CPUs plus standard memory, DRAM included, have to scale as well. We think Penguin is positioned well to lead in this market opportunity where data center, AI infrastructure and memory solutions meet. Enterprise and AI native companies are tackling these AI compute infrastructure requirements in many different ways, mostly by taking AI infrastructure from new cloud providers and, in certain instances, constructing their own on-premises AI factories. For persistent high utilization AI workloads, inference among them, big enterprises are increasingly weighing on-premises AI infrastructure, which can deliver improved token economics, data control and performance. Our full stack AI factory platform brings together 5 core elements alongside our partner ecosystem. First, cluster wear AI operating system software for AI factories; second, memory AI and integrated memory solutions. Third, GPU and CPU systems sold under our compute AI brand. Fourth, Origin AI factory architectures; and fifth, end-to-end design, build, deploy and manage services. What distinguishes our platform is a rare mix of OEM like differentiated product innovations plus end-to-end AI infrastructure systems integration capabilities; that lets us act as the single builder and operator of AI factories across the entire life cycle for our customers. Customers require more than hardware and assembly. This matters especially for neo cloud providers, where we have recently seen a surge of interest that we think signals strong product market fit for our full stack platform. For new clouds, we can act as a single operating partner across the entire life cycle. Helping customers satisfy service level agreements and generate revenue sooner while they concentrate on securing additional offtakers and scaling their businesses. Let me now move to our business results, beginning with advanced computing; in Q4, advanced computing net sales came to $154 million. Accounting for 27% of company net sales and rising 11% year-over-year. Data center AI infrastructure growth inside advanced computing, powered by our AI factory platform, was considerably stronger. Non-Hyperscale AI infrastructure net sales climbed 99% year-over-year and made up 66% of advanced computing net sales, compared with 60% in the prior year quarter; that growth more than offset continued runoff from other traditional businesses and shows the ongoing shift in advanced computing toward data center, AI infrastructure. During the quarter, we brought on 6 new AI infrastructure customers including 4 Neo clouds, a large quantitative trading firm and an enterprise customer. We closed fiscal 2026 with strong momentum and bookings have accelerated further as we enter fiscal 2027. Our land and expand model stays central to our go-to-market strategy. Successful initial deployments open up opportunities to expand across our platform, including cluster wear AI and multiyear managed services engagements that usually run 3 to 5 years. During fiscal 2026, 17 new AI infrastructure clients came on board, and 12 existing clients grew their business with Penguin. Momentum among Neo Cloud customers is strong. One Neo Cloud, backed by a major South Korean technology company, chose us to design, build, deploy, and manage an NVIDIA GB-300 NVL-72-based platform, and that first deployment is expected to underpin a wider GPU as a Service expansion. A neo-cloud provider that is publicly traded and holds over $3 billion in signed multiyear AI infrastructure contracts picked Penguin for a multiyear engagement covering AI infrastructure deployment plus 24/7 operation services, backed by cluster were AI, our AI factory operating system software. Lektra, a neo cloud, chose our full stack AI factory platform to deploy and optimize distributed AI micro data centers powered by existing carbon-free energy. We supply validated reference designs and video-based AI GPUs along with expert services featuring cluster war AI and support. And for a different neo cloud provider holding $10 billion in contracted compute from a leading AI lab, we were chosen to deploy and operate a 36,000 AI factory in Norway; this multiyear engagement shows the scale of our platform and our capacity to bring large-scale AI infrastructure into production. Such engagements show an important evolution in our business.

Kash Shaikh

executive
#3

More and more, customers are picking Penguin to serve as an end-to-end partner for designing, deploying, optimizing, and running AI infrastructure. We kept pushing cluster wear AI forward too, adding new Agentic AI capabilities that handle automation of AI factory operations, remediation, and infrastructure monitoring. Through our AI factory operations agent, administrators get a natural language interface that delivers operational insights. To put it plainly, admins can now hold a conversation with their AI factory. We also hold the NVIDIA AI factory specialized partner designation, which is invitation-only and reflects how deep our partnership with NVIDIA runs — a partnership that keeps getting stronger — along with more than 10 years of know-how in building and managing AI factories based on NVIDIA technology. Last, we set up a relationship with an additional AI hardware supplier, which helps widen supply availability and meet rising demand. Shifting now to integrated memory. Net sales for the fourth quarter hit a record $341 million, which works out to 60% of company net sales, rising 158% year-over-year and 24% sequentially. During Q4 fiscal 2026, we brought on 2 new memory customer logos. Results from our land-and-expand strategy keep coming as well. Over the course of fiscal 2026, we brought in 15 new customers, and 29 existing customers grew their business with us. We're witnessing growth in both volume and pricing. Underlying our growth is robust data center demand fueled by Agentic AI, which includes rising memory requirements from AI and inference workloads. Memory is a cyclical market, yes, but we think AI is contributing a structural demand driver, especially for memory solutions within the data center. Penguin's approach differs from that of a traditional commodity memory manufacturer. Data center solutions are our primary focus. Engineering design, firmware development, advanced validation, negotiated pricing, and an asset-light model — all of these are meant to produce more resilient economics through memory cycles; bookings and pipeline for our CXL memory expansion products keep getting stronger, and here are a couple of proof points. In Q4, we landed business with a leading global AI server manufacturer whose platform supports large-scale AI training and inference. That win underscores the strength of our memory portfolio for data center AI infrastructure; we also broadened our relationship with a leading next-generation AI infrastructure inference provider, which is rapidly scaling AI inference deployments using our CXL memory expansion cards. Outside our core memory business, we're creating new solutions meant to tackle the expanding memory bottleneck in data center AI infrastructure. Our CXL-based large memory expansion cards and memory AI KV cash solution are built to boost memory capacity and enhance the economics of Agentic AI and real-time inference. Future memory appliances are part of our memory AI road map as well, among them photonic high-bandwidth memory solutions intended to grow GPU memory capacity and bandwidth. We're also working on a CXL AIC product line meant to repurpose installed DDR4 memory for AI infrastructure, which helps customers extend asset life and reduce the cost of scaling memory-intensive workloads. Lastly, we put in place a new supply arrangement with a leading memory supplier to back the growth of our data center AI infrastructure portfolio and bolster our supply position. Taken together, these developments underscore our long-term focus on data center memory and AI infrastructure. Now, a quick note on optimized LED. That segment also had a solid quarter, posting $72 million in Q4 net sales, which marks 7% growth compared with a year ago. Data center, AI infrastructure, and memory stay at the top of our strategic investment list, and we'll keep running the LED business in a disciplined way. Looking ahead to fiscal 2027, we aren't just trying to grab demand. We're working to create an operating model that's more repeatable and scalable around that demand, all while putting customer success at the heart of what we do. Along those lines, we're rolling out Agentic AI throughout our operations, aimed at business outcomes we can actually measure. Agentic AI coding workflows are being used by our engineering teams to speed up cluster are AI development. Meanwhile, AI-powered tools are assisting us in shortening product cycles and making the supply chain more responsive. On that note, let me give you our revised fiscal 2027 outlook. On our Q3 earnings call, we shared preliminary fiscal 2027 expectations calling for net sales and diluted EPS growth of roughly 30% versus a year earlier, measured from the midpoint of our fiscal 2026 outlook. Given our fourth quarter results and ongoing AI-driven demand, the fiscal 2027 outlook we're offering today surpasses that preliminary view on both net sales and diluted EPS. Because of our strong operating leverage, we anticipate diluted EPS will rise more quickly than net sales. Aaron will give you the specifics shortly. I'd like to wrap up with three points. One, our strategy is delivering and growth is getting broader. AI-driven businesses made up 78% of Q4 net sales, expanding 141% compared with a year ago, with integrated memory still strong and AI infrastructure now picking up markedly. Two, significant operating leverage is fueling profitable growth while letting us keep investing in innovation. EPS is climbing faster than net sales, even as we keep funding differentiated products and capabilities. Three, we see this value creation opportunity as lasting and still very early, since we sit at the crossroads of AI infrastructure and memory, paired with cluster ware AI and services, which gives us a distinctive long-term edge. I'd like to thank our team for their dedication and execution across fiscal 2026 and for the momentum we're bringing into fiscal 2027. With that, I'll hand things to Aaron to go over the financials and our outlook.

Aaron Johnson

executive
#4

Thank you, Kash. I'll start by covering what we delivered in the fourth quarter and across all of fiscal 2026, and after that I'll share how we see fiscal 2027 shaping up. During the fourth quarter, both net sales and earnings per share came in above the guidance we had given, which shows our AI infrastructure solutions are being adopted more widely and that AI-related demand for data center memory keeps expanding. Solid growth in net sales, careful control of expenses, and better profitability combined to produce significant operating leverage. We achieved all-time company highs for net sales, gross profit dollars, operating income, operating margin, net income, and adjusted EBITDA. Across the entire year, we likewise posted records for gross profit dollars and operating income. Turning to the fourth quarter first, net sales reached $567 million, which is 68% higher than a year earlier and 18% above the prior quarter. Gross margin came in at 28.8%, operating margin at 15.8%, and diluted earnings per share at $1, an increase of 133% compared with the same period last year. Over the full year, net sales totaled $1.73 billion, a rise of 26% against fiscal 2025, while diluted earnings per share climbed 51% to $2.87, which is close to double the pace of net sales expansion and illustrates the operating leverage built into our model. Product net sales in the fourth quarter amounted to $510 million, equal to 90% of the company's total net sales, and were up 86% from a year ago. Services net sales came to $57 million, or 10% of total net sales, a decline of 11% year-over-year that mirrors reduced hyperscale services volumes, though growth in multiyear AI factory managed services partly made up for that drop. Net sales broken out by business segment came in as follows: advanced computing posted fourth quarter net sales of $154 million, an 11% increase from a year earlier and 12% higher than the previous quarter. For the full year, advanced computing net sales were $559 million, a decrease of 14% that reflects the wind down of Penguin Edge along with weaker hyperscale hardware sales. Our Non-Hyperscale AI infrastructure business expanded 99% during the fourth quarter and accounted for 66% of segment net sales, and it grew 72% over the year, strengthening its role as the main engine of growth for that segment. As for integrated memory, fourth quarter net sales hit a record $341 million, making up 60% of total company net sales, which is 158% above the prior-year period and 24% above the preceding quarter. For the full year, integrated memory net sales reached $924 million, up 99%, and made up 53% of total company net sales, compared with 34% in fiscal 2025. That expansion mirrors robust demand for memory that supports AI-driven data center applications, in line with our strategic emphasis on this market. For optimized LED, fourth quarter net sales were $72 million, accounting for 13% of total company net sales, up 7% from a year earlier. For the full year, net sales came to $249 million, down 3%. This business improved its profitability as we keep executing with discipline and pursuing profitable opportunities. Fourth quarter gross margin stood at 28.8%, 70 basis points higher than the prior quarter, helped by better profitability in integrated memory that was partly offset by weaker advanced computing margins. Combined with robust net sales growth, that produced record gross profit of $163 million, up 21% from the previous quarter and 57% from a year earlier. For the full year, gross profit hit a record $508 million. Our gross margin came in at 29.3%, which is a decline of 1.7 percentage points, largely driven by a bigger mix contribution from AI infrastructure hardware, even though integrated memory delivered stronger margins. The increase in gross profit dollars shows the leverage our larger scale is creating. Operating expenses for the fourth quarter totaled $73 million, rising 4% from the prior quarter and 13% compared with a year earlier. That sequential step-up mostly reflects greater incentive compensation tied to our strong financial performance. Across the entire year, operating expenses rose 4% to $267 million, far under the 26% increase in net sales. Consequently, operating expenses as a share of net sales fell 3.4 percentage points, which speaks to a structural gain in how scalable our operating model has become; combining those higher gross profit dollars with spending discipline produced a record fourth quarter operating income of $90 million, climbing 39% quarter-over-quarter and 129% year-over-year. Because net sales growth far exceeded the growth in operating expenses, operating margin widened 2.4 percentage points sequentially and 4.2 percentage points versus a year ago, reaching 15.8%; for the full year, operating income hit a record $241 million, an increase of 44%, while operating margin widened 1.7 percentage points to 13.9%. What these results show is that scale paired with disciplined spending is converting growth into greater profitability. Diluted earnings per share in the fourth quarter came to $1, which is 19% higher sequentially and 133% higher year-over-year. That sequential EPS growth rate sits below our 39% operating income growth, a reflection of a larger share count after our convertible notes offering. For the entire fiscal year, diluted earnings per share amounted to $2.87. That figure is up 51% and $0.22 above the top end of our July outlook. Adjusted EBITDA set a record at $93 million in the fourth quarter, up 115% from a year ago and 38% from the prior quarter. For the full year, adjusted EBITDA was $256 million, an increase of 37%. Before I go through working capital, let me remind you that days sales outstanding, days payables outstanding and inventory days are all calculated by us on a gross sales and gross cost of goods sold basis. Gross sales totaled $1.57 billion, while gross cost of goods sold came to $1.42 billion in the fourth quarter. The gap between gross and net sales is mainly tied to our memory businesses logistics services, which are accounted for on an agent basis. Now to working capital. Net accounts receivable grew to $796 million, up from $308 million a year ago, which reflects expansion in our memory business and the gross billings that come with it. Even with that larger balance, days sales outstanding improved to 46 days, compared with 53 days last quarter and 51 days a year ago. We are managing these investments in line with customer deployment schedules and supply availability as we support growth. Inventory rose to $749 million, up from $255 million a year ago, reflecting purchases made to support record backlog and expected customer deployments across our memory and AI infrastructure businesses, along with higher memory costs. Inventory days stood at 48%, up from 42 last quarter as we positioned inventory against a record backlog, and down from 51 a year ago. Our accounts payable rose to $817 million, versus $267 million during the same period last year. That increase mirrors purchasing activity tied to expanding both our memory and AI infrastructure operations, along with elevated memory expenses. Days payable outstanding came in at 52%, a decline from 62% in the prior quarter and from 54 a year earlier, mostly driven by when purchases landed inside the quarter. The cash conversion cycle stood at 42 days, against 33 days in the previous quarter and 49 days 12 months back. Efficiently financing growth while keeping working capital management tight continues to be a key priority. Turning to our capital structure more broadly. In the quarter, we closed a heavily oversubscribed offering of $750 million in 0% convertible notes maturing in 2031, which we see as a sign of solid investor belief in our company and our long-range plan. These notes carry an initial conversion price near $116.70 per share, a premium of roughly 50% over where our stock traded at pricing. In addition, we put in place capped call transactions meant to limit possible dilution up to an initial cap price of about $175.05 per share. At the same time, we swapped roughly $296 million in principal amount of our 2029 and 2030 convertible notes for a mix of cash and common stock shares, then directed part of those proceeds to pay off in full the $100 million still outstanding under our credit agreement. This proactive, disciplined move lengthened our maturity schedule and bolstered our capital structure, giving us more room to invest behind our AI factory platform strategy and the growth prospects that lie ahead.

Aaron Johnson

executive
#5

After those steps, we closed out Q4 with $647 million in cash and cash equivalents. That's a $207 million increase versus Q3, and $193 million higher than the same time 12 months earlier. Managing liquidity, along with the working capital needed to fuel our growth, stays a key focus for us. Operating activities absorbed $163 million of cash during Q4, and $152 million across the entire year, largely due to working capital investments backing expansion in our memory and AI infrastructure businesses. For comparison, the prior year quarter saw $70 million of cash used, while fiscal 2025 generated $113 million of cash. We think our capital structure and liquidity give us room to finance this growth, and boosting working capital efficiency continues to be a major priority for fiscal 2027. Capital expenditures came to $4 million for the quarter and $12 million for the year, comfortably below 1% of net sales, while depreciation expense was $5 million and $20 million, respectively. Because of our asset-light model, we can scale up without heavy capital outlays, and we anticipate that continuing in fiscal 2027. Now let's move to our outlook. We're heading into fiscal 2027 with solid momentum, backed by our second half results, a record backlog, growing bookings, sustained integrated memory strength, and even faster acceleration in demand for our AI infrastructure business. Our expectation is that robust net sales growth will lift gross profit dollars while operating expenses rise at a much slower pace; that mix should widen operating margin and produce diluted EPS growth exceeding net sales growth. For fiscal year 2027, our expectation is net sales growth of roughly 40% at the midpoint, plus or minus 10 percentage points, which works out to an increase of about $700 million, reaching approximately $2.43 billion at the midpoint. During last quarter's call, we offered a preliminary view of about 30% growth measured from the midpoint of our then current fiscal 2026 outlook. Taking that same fiscal 2026 outlook midpoint as our starting point, our present fiscal 2027 midpoint works out to roughly 45% growth. This range captures differing growth expectations among our businesses, along with variability in customer deployment schedules and in the availability, lead times and cost of certain products and components. Our record backlog, combined with the demand we're seeing throughout the business, gives us a solid foundation for our fiscal 2027 growth expectations. That said, customer deployment schedules, product availability and shifting memory market conditions might influence the mix and cadence of net sales over the course of the year. We feel the range strikes the right balance between the opportunities ahead and these execution and market variables. Turning to the net sales outlook by segment: for advanced computing, our expectation is full year net sales growth of approximately 40% at the midpoint, plus or minus 10 percentage points, fueled by ongoing momentum in non-hyperscale AI infrastructure, partly offset by the absence of edge sales and lower expected hyperscale sales. When it comes to integrated memory, our expectation for the full year is net sales growth of roughly 50% at the midpoint, give or take 10 percentage points, fueled by ongoing robust demand for memory that supports AI-driven data center applications. Turning to optimized LED, we anticipate net sales for the full year will be relatively flat compared with the prior year as we keep managing that business with discipline. We're expecting a full year gross margin of about 28%, plus or minus 2 percentage points. This outlook captures the anticipated mix across our businesses, including how hardware, software and services are combined, along with our present assumptions about memory pricing and costs. Although those factors shape the gross margin rate, we anticipate that net sales growth and scale will produce higher gross profit dollars. For the full year, we expect operating expenses of roughly $275 million, plus or minus $10 million. Inside that range, we're shifting resources toward our AI infrastructure and memory programs with the highest returns while keeping total spending growth far below net sales growth. This disciplined approach to investment, combined with higher gross profit dollars, should drive additional operating margin expansion. Across the full year, we expect diluted earnings per share of about $4.45, plus or minus $0.70, which represents growth of roughly 55% at the midpoint and surpasses our anticipated rate of net sales growth. The EPS outlook reflects higher gross profit dollars, disciplined operating expense growth and additional operating margin expansion. This range also accommodates variability in when large customer deployments occur, since schedules for individual projects can move between quarters. Our EPS outlook is built on a diluted share count of roughly 63 million shares and an effective tax rate of about 20%. We expect to apply this normalized rate through fiscal 2027, although it could shift along with our geographic earnings mix and changes in the global and U.S. tax environment. Our fiscal 2027 outlook rests on the current demand environment, backlog, customer deployment schedules, anticipated product mix and our present assumptions about supply availability and costs. The timing of customer deployments and product availability could create variability between quarters, and some components still have extended lead times that may impact project and shipment timing. Our outlook also includes current assumptions regarding memory supply and costs, which could evolve over the year, including the anticipated effect of new memory and AI infrastructure suppliers that reinforce our overall supply position and support growth. These factors might influence quarterly cadence and mix, but they're accounted for in our full year planning assumptions. On the whole, our outlook reflects the momentum in our AI-driven businesses and the advantages of a more scalable operating model. We intend to invest selectively behind our highest return opportunities while preserving expense discipline. Our objective is straightforward: turn strong net sales growth into higher gross profit dollars, operating margin expansion and earnings growth that outpaces net sales growth. For additional detail, please see the non-GAAP financial information section and the reconciliation of GAAP to non-GAAP measures in our earnings release and the investor materials posted on our website. Operator, at this point we're set to move into the question-and-answer portion.

Katherine Murphy

analyst
#6

Seeing your updated advanced computing segment guidance for fiscal 2027 — 40%, plus or minus 10% — is genuinely helpful. A quarter ago, my recollection is that you put forward mid-teens growth as the preliminary baseline for that outlook. So, to start: could you explain what has changed during the past 90 days that allows you to lift your advanced computing forecast? Then, secondly, as Penguin shifts more heavily toward the Neo Cloud customer type, in what ways does that alter the unit economics relative to the enterprise customers and quant wins from the early part of 2026? Does that involve a greater amount of hardware? What should we expect in terms of the effect on margin? If there's anything you could offer that would be useful to us on the modeling side, that would be terrific.

Kash Shaikh

executive
#7

Thank you, Kath, for those questions. Regarding your first one, about the preliminary guidance we gave last quarter of mid-teens, and how it shifts once we get to advanced computing, the primary factor behind that is AI infrastructure. So for the most part, this reflects the expansion we've witnessed since our previous earnings announcement. On the bookings side, demand has come in above our forecast, and the pipeline remains robust. As mentioned, a number of the deals I cited illustrate why we believe momentum in our memory business is holding up while our AI infrastructure is speeding up even more. That is what underpins the higher outlook for AI infrastructure, which lifts advanced computing guidance and, in turn, the guidance for the company overall, with AI infrastructure as the main force behind it. Regarding what you asked about neo clouds and unit economics — the unit economics look a lot like what we see with AI factory deployments for enterprises on-premise. That said, something we've noticed among the new cloud customers is that these neo cloud customers represent a really distinctive product market fit for us. So what I mean is, we're able to handle everything for them from start to finish — we'll build their factories and run them on their behalf. That covers buying the hardware, doing the complete hardware design, standing up the data center, and then handling data center management afterward, from day two onward, for as long as three to five years, as I mentioned in a few of the examples. On top of that, there's our cluster wear, which is genuinely what sets us apart when it comes to delivering these services and, even more crucially, hitting the SLAs — and those matter a great deal for these new cloud customers, since they need to offer SLAs to their own offtakers. Taking everything together, though, the unit economics look similar, there's room for more expansion, and we've got a fairly distinctive product market fit with these new cloud customers.

Sajal Dogra

analyst
#8

You've basically jumped a whole year ahead of my bull case, so congratulations. For my question, I suppose: with your earlier guidance, revenue growth and earnings growth were about even, but now you're at last demonstrating operating leverage for the coming year. So I simply want to grasp what has shifted compared with 90 days back that has enabled you to generate that earnings expansion. And I was curious — is that entirely — I mean, how much of it comes from memory — or are we at last witnessing a bit of margin expansion in advanced computing too?

Kash Shaikh

executive
#9

Certainly; so Sajal, to begin with, the guidance we gave 90 days ago amounted to more of a tint and initial steer rather than the real outlook, since it came earlier than we normally provide one, correct? Ordinarily, we give it during Q4 earnings, which is the period when we deliver the update. So that one was far more directional compared with what we're giving now, which examines every aspect, and we possess greater visibility and greater clarity, plus the confidence of running under this new model where, if you consider Q3, our EPS rose quicker than net sales at the company level. And then the same holds in Q4, net sales climbed 68% against EPS growth of 13%; and the visibility we possess regarding bookings, conversion ratios, and so forth and so on. We have far greater clarity, and this is far more a formal outlook than the high-level tint we offered directionally. So since we're nearer, we have greater visibility, and we have greater confidence in sustaining the model as we scale the business; and regarding margins at the segment level, we won't reveal margins at the segment level. But I can say at a high level, memory stays strong, stronger than we anticipated. Considering the value we deliver and the emphasis on the data center. Advanced computing in general has the benefit of services and software; yet, as we increasingly secure wins combining hardware and software, AI infrastructure may face some downward pressure, but simultaneously, the benefit of cluster wear, software, and services carries a very high attach rate typically with these opportunities I mentioned during the earnings call before and the remaining pipeline we observe. So that's the high-level picture, to give you a sense of how we view segment margins.

Ananda Baruah

analyst
#10

Yeah, thanks so much for that. In line with what Kash, Aaron, and Stephen were saying — welcome, by the way — I'd love to dig into the dynamics that are driving not only the bump to the fiscal year '27 guide, but also the faster growth rate we're actually seeing across the main businesses. What mechanics are you observing out in the market that are causing this greater neo cloud capture? What can you tell from what you're able to make out on the commercial enterprise side, and I'd appreciate a bit of background, if that's worth giving, on how much the NVIDIA and Dell relationship is really feeding into these dynamics at a quicker pace.

Kash Shaikh

executive
#11

Absolutely, Ananda. So to begin with the business acceleration, like I said before, one piece of it is the greater visibility we're getting, plus the growth acceleration; we think one competitive edge we hold is that we aren't a product company that merely ships hardware and walks away at the customer's doorstep — we actually engage with them well before in the cycle, join in their design work according to the requirements they bring, whether that's a big enterprise customer standing up an on-premise factory or a cloud provider, and then collaborate with them < > to grasp their requirement — if it's a new cloud provider, what their SLAs are, who their customers are, and how we can shape the design so it fits their needs. So that consultative side, and particularly the fact that we've done this for over 2 decades — that's an edge. Our team has been working alongside NVIDIA on designing these clusters for over a decade. So we hold that edge in grasping the architecture. The conversations happen at the architecture level; we have the software, the cluster that pulls all of these elements together and makes building and managing these factories far simpler. And then staying with them end-to-end and assisting them in managing as part of our managed services so they achieve those outcomes when they're putting so much capital into these AI factories or data centers is genuinely a distinctive edge relative to the alternatives available to them — whether they turn to product-only companies or services companies, there's no unique combination like the one we deliver through our platform, between the products we have and the services we offer. So that's one edge we're seeing pay off for us. The NVIDIA partnership is definitely helping us, since we were recently nominated by NVIDIA as the AI factory specialized partner. We happen to be one of just a small number of partners that they work with, and that really speaks to what it's like collaborating with them, plus everything our team can do across design, build, deploy, and manage — it gives us that credibility. But once more, when we're brought into these designs and we're running white-boarding sessions alongside the customers to satisfy their requirement, that's a truly distinctive competitive edge we bring, together with our products; so that's mostly for, clearly, AI infrastructure, which is speeding up; on the Memory side, one thing we've accomplished since we rolled out our AI factory platform strategy halfway through fiscal 2026 is this. We've shifted our emphasis more toward data centers for 2 reasons. First, even though memory demand is elevated across every other segment, the consumer segments included, we think that demand could simply be a matter of demand and supply dynamics and possibly cyclical — yet data center memory demand is fueled by AI, particularly agent AI workloads, which automate the workflow and generate real productivity for enterprises, and that will last far longer. And should we say agent TKIs adoption among enterprises sit at 10%, since they're still working out the work flows and the tool is being defined, then for each 10% gain, that demand could double. And we keep watching that demand climb for our memory products in the data center, so that data center positioning is one more edge, both regarding greater demand and regarding how durable that demand is for us; so pair that with these 2 tailwinds on our side, Memory business placement in the data center, fixing memory problems, delivering solutions to OEMs, and now more and more delivering forward-looking data center solutions like CXL memory expansion cards. Those memory expansion cards serve mainly inference workloads, and BC this CXL business is picking up very strongly. So the pipeline keeps growing, bookings keep growing. So those newer solutions, the data center focus, the inference focus, is yet another edge for us as our differentiation and in tackling the inference challenges inside the data center. So those are a few of the things. On top of that, one further thing we're noticing is, broadly, returning to AI infrastructure. Among enterprises, they're mainly relying on New Cloud as a Service for their AI infrastructure. Yet in certain cases, they're weighing on-prem, particularly large enterprises, because at scale, when they run inference applications, cost turns into a challenge. Especially with inference workloads. At that moment, although they're gaining the productivity, they aren't seeing the ROI on the dollars they're putting in, and they consider on-prem factories, and then we get brought in, and they're seeking to solve their cost challenge. Additionally, open weight models are something they're looking at, since that gives them another route to improved economics as they shift away from a cloud-based service toward on-premise. So those are a few of the trends where we're uniquely positioned, given both our memory business and our AI infrastructure business, and that's what's fueling expansion in these 2 businesses.

Rustam Kanga

analyst
#12

Great, and well done on a year that ended in memorable fashion, with the business carrying steady momentum thanks to AI infrastructure. On cash — you mentioned that memory's expansion was split between pricing and units. Last quarter, I recall you sharing a rather interesting point: had you factored in the orders that couldn't go out because of supply limits, that breakdown would land nearer to fifty-fifty, or maybe lean further toward units. So I'm curious — what portion of those unshipped orders has started to come through by now? And when you look at the coming year, how do you weigh pricing against units in the memory segment, especially given your push toward new clients, chiefly around data center solution use cases.

Kash Shaikh

executive
#13

To start, I appreciate you picking up on what I said. That's correct — it matched what we saw in the prior quarter, when memory posted notably robust, record-breaking revenue despite the backlog equaling our revenue. Luckily, that same pattern is holding up. So while memory just delivered yet another record quarter, whatever backlog we hold now sits well above the revenue we booked this quarter. That backlog currently reaches out a minimum of 4 quarters and keeps getting stronger, and that's exactly why we feel confident raising the business outlook. As I noted when answering the earlier question, this really comes down to demand centered on data centers for our main memory operation, plus the CXL products we're building, which have seen a really robust pipeline and demand driven by inference growth in data centers.

Matt Calitri

analyst
#14

Matt Calitri Lariat from Needham; it's really good to see how strong things were as we wrapped up the year, especially with a few of those fresh cloud wins you mentioned. Could you share a bit more about when those projects are going to kick off and start showing up in the model, and what kind of impact they might have? Also, is there something particular we can point to for why you were able to close so many deals this quarter? Or, more generally, why is right now the moment for this neo-cloud opportunity?

Kash Shaikh

executive
#15

Yes. We aren't providing that granular a level of detail about which project and when, but at a high level I can say that the majority of these projects I mentioned are already underway — meaning we're handling their data centers or constructing them; and for a few, we're in the middle of procurement; so these aren't things in the future. They're already taking place. Actually, we've been engaged on these projects for multiple months, even if a portion of it has sped up into the bookings lately. We were engaged on these fresh cloud opportunities, together with other enterprise opportunities, which is the reason I referred to Neo Cloud as a growth opportunity for us in the prior earnings announcement even if we weren't able to share the details, and now that we've begun working with them, and we're managing these factories in certain cases, and we're seeing bookings rise. That's the reason we're sharing additional details, and we'll keep sharing additional details regarding these new cloud deployments; but as I said, one of the reasons is the unique product market fit that exists between what we offer and our expertise in designing these AI factories or data centers, expertise spanning decades, to help them reach their goals — it's a fairly unique competitive advantage we hold in the market with these opportunities.

Brian Chin

analyst
#16

Nice work on these results, and I appreciate you taking our question. We might sort of have two things bundled here. Since you mentioned that record backlog heading into the fiscal year, plus the growth guidance for computing, can you roughly tell us what portion of the fiscal '27 advanced computing revenue outlook is already covered by backlog? And the second piece is clearly the really impressive flow-through on the model and the EPS — the non-GAAP EPS guidance for next year. But with that quite small year-over-year OpEx increase, are you putting enough into head count and retention to perhaps boost bandwidth and utilization and chase more projects? Since your revenue is climbing so much, it seems like you've got ample room to put even more money to work.

Kash Shaikh

executive
#17

Correct. I'll address your initial question first. The portion of our revenue that's covered by backlog — those specifics aren't something we disclose. Still, speaking broadly, I'd describe it as very strong — a very strong backlog spanning both memory and AI infrastructure. That's the reason we're driving toward a record backlog. Turning to your question about operating expenses: the significant operating leverage we mention stems from our efficient operating model. To clarify what efficient operating model means specifically, consider R&D — research and development — along with our engineering. More than 95% of our engineering is based mainly in India, which is a low-cost location, and that gives us an edge in scaling far more economically; our memory engineering is also in Taiwan, another inexpensive location. Therefore, even while we grow, our costs remain highly efficient relative to other businesses that might lack such a cost-efficient model; our manufacturing, meanwhile, is in Malaysia — every one of these locations is low-cost, and as the business expands, that represents a fixed cost for us when we bring on added capacity, and we do have the capacity; so even when capacity is added, our costs or expenses do not rise in a linear fashion alongside that revenue. We are, in addition, applying AI quite broadly throughout the company. Over the past roughly eight to nine months, we've seen substantial efficiency improvements. Take our software engineering staff: every single one of them, regardless of location, now relies on AI for core generation, and that's letting us grow beyond the limits of a conventional approach. AI is deployed across nearly all of our departments, finance and legal included, and it's enabling us to expand far more effectively. So when you combine this model with our cost-efficient locations and the scaling we're doing there, plus our use of AI, it adds up to a highly efficient setup. That said, to address what you asked, we are absolutely putting money into innovation, and that's exactly why we keep winning more. When it comes to the items I brought up, AI differentiation shifts by cluster, and securing additional XL for our memory offerings gives us yet another edge — we'll keep putting more money behind that. So even if the dollar figures don't appear to rise while productivity improves alongside a growing scale, our efficient model generates operating leverage, and that's what lets us keep delivering stronger profitability. Take a look at operating profit: $90 million, a record, and the prior quarter was also a record at almost $64 million. We still expect that as we pursue profitable growth grounded in the scalable model — the official model — that we've built, we'll be able to invest more and more in innovation, and productivity for the company will keep climbing even so.

Operator

operator
#18

We're wrapping up the call here, as there's nothing else on the agenda. Thanks for joining us, and you're welcome to hang up at this point.

Read the full transcript via the API

You're viewing the first half of this call. Get the complete Penguin Solutions, Inc. transcript — plus 132,000+ transcripts from 7,000+ companies, speaker segments, AI summaries and full-text search — through the EarningsCalls.dev API.

Get the API View API docs →

This call discussed

For developers and AI pipelines

Programmatic access to Penguin Solutions, Inc. earnings transcripts and 132,000+ others is available through the EarningsCalls.dev REST API. Plans from $39.99/month — full transcripts, speaker segments, full-text search, and the recently-added /api/v1/transcripts/recent polling endpoint for ETL pipelines.

More tools for stock research

OptionsBell: an unusual options activity scanner that tracks options flow across all US stocks and emails you when large trades appear in Penguin Solutions, Inc. or any ticker on your watchlist.

StockMarketScan: a stock screener with alerts: 24 pre-built screens refreshed every 5 minutes, watchlists and email notifications. Both offer API and MCP access and fair monthly pricing.