Penguin Solutions (PENG) Fiscal Q4 2026 Earnings Call: AI Growth Drives Record Sales

Key Takeaways

• Penguin Solutions reported fiscal Q4 2026 net sales of $567 million, up 68% year over year and 18% sequentially. Diluted EPS rose 133% year over year to $1.00.
• AI-driven businesses—non-hyperscale AI infrastructure and integrated memory—accounted for 78% of quarterly net sales and grew 141% year over year.
• Integrated memory generated record quarterly net sales of $341 million, up 158% year over year and 24% sequentially, driven by higher volume and pricing.
• Non-hyperscale AI infrastructure sales increased 99% year over year and represented 66% of advanced computing revenue. Penguin added six AI infrastructure customers during the quarter, including four neo-cloud providers.
• Management expects fiscal 2027 net sales growth of approximately 40% at the midpoint to about $2.43 billion. Diluted EPS is projected to increase approximately 55% to $4.45 at the midpoint.
• The company ended fiscal 2026 with record backlog. Management said bookings grew faster than net sales in AI infrastructure and integrated memory, while the memory backlog extends at least four quarters.

Key Financial Data

Financial measures are non-GAAP unless otherwise indicated, consistent with the company’s presentation.

Operating expenses increased 13% year over year to $73 million in Q4. For the full year, expenses rose 4% to $267 million, materially below the 26% increase in net sales.

Penguin ended the quarter with $647 million in cash and cash equivalents after completing a $750 million 0% convertible note offering due in 2031. Capital expenditures were $4 million for the quarter and $12 million for the year.

Business and Operating Performance

Integrated memory

Integrated memory was the largest business, contributing 60% of Q4 company net sales. Revenue reached $341 million, up 158% year over year. Full-year sales increased 99% to $924 million and represented 53% of total revenue, compared with 34% in fiscal 2025.

Management attributed growth to both higher volume and pricing, supported by demand for memory used in AI-driven data center applications. The company added two memory customers in Q4 and 15 during fiscal 2026, while 29 existing customers expanded their business.

Penguin also reported stronger bookings and pipeline for its CXL memory expansion products. Its development roadmap includes CXL-based large-memory expansion cards, a Memory AI KV cache solution, future photonic high-bandwidth memory products and a CXL AIC product designed to repurpose installed DDR4 memory.

Advanced computing and AI infrastructure

Advanced computing produced Q4 net sales of $154 million, up 11% year over year and 12% sequentially. Full-year revenue declined 14% to $559 million, reflecting the wind-down of Penguin Edge and lower hyperscale hardware sales.

Within the segment, non-hyperscale AI infrastructure grew 99% in Q4 and 72% for the year. The company added 17 AI infrastructure customers during fiscal 2026, while 12 customers expanded their engagements.

Neo-cloud demand was a central growth driver. Management described multiple engagements involving the design, deployment and operation of AI factories, including an NVIDIA GB-300 NVL-72-based platform and a 36,000-GPU AI factory in Norway. Several engagements include ClusterWare AI and managed services lasting three to five years.

Penguin said its differentiation comes from combining hardware, ClusterWare AI software, memory products, AI factory architectures and end-to-end design, deployment and management services. The company also holds NVIDIA AI Factory Specialized Partner status.

Optimized LED

Optimized LED generated Q4 net sales of $72 million, up 7% year over year. Full-year sales declined 3% to $249 million. Management said it remains focused on disciplined execution and profitable opportunities in this business.

Management Guidance

The fiscal 2027 revenue outlook exceeds the preliminary view shared in the prior quarter. Management expects operating expenses to grow considerably more slowly than sales, supporting operating margin expansion and EPS growth above the revenue growth rate.

The company said its outlook is supported by record backlog, stronger bookings, continued integrated memory demand and accelerating AI infrastructure activity. Advanced computing growth is expected to be driven by non-hyperscale AI infrastructure, partly offset by the absence of Penguin Edge sales and lower expected hyperscale revenue.

Risks and Areas to Watch

• Customer deployment schedules may shift revenue and earnings between quarters, particularly for large AI infrastructure projects.
• Product availability, component lead times and supply constraints could affect project timing and shipment cadence.
• Memory pricing, supply and costs may change during fiscal 2027 and influence revenue mix and gross margin.
• Gross margin is sensitive to the mix of hardware, software and services. A higher contribution from AI infrastructure hardware reduced full-year gross margin despite stronger integrated memory profitability.
• Growth required substantial working capital investment. Inventory rose to $749 million from $255 million a year earlier, while operating activities used $152 million of cash in fiscal 2026.
• The company’s convertible notes increased the diluted share count, which management said caused Q4 EPS growth to trail operating income growth sequentially.

Analyst Q&A Highlights

Management attributed the higher advanced computing outlook primarily to bookings secured since the previous earnings call and a stronger pipeline, particularly among neo-cloud customers. It said neo-cloud unit economics are broadly similar to enterprise on-premises AI factory deployments, with opportunities to attach ClusterWare AI and multiyear managed services.

On margins, management said integrated memory remains stronger than previously expected. Advanced computing benefits from software and services, although a higher hardware mix can place pressure on margins. ClusterWare AI and managed services generally have high attachment rates on current opportunities.

Management said most of the neo-cloud projects discussed on the call have already started. Depending on the engagement, Penguin is procuring equipment, building data centers or operating deployed infrastructure.

The company did not quantify how much of fiscal 2027 advanced computing revenue is covered by backlog. It characterized backlog across memory and AI infrastructure as strong and said the memory backlog is higher than the revenue recognized in Q4.

Management also defended its planned operating expense growth as sufficient to support expansion. It cited engineering operations concentrated in India and Taiwan, manufacturing in Malaysia and the use of AI tools across software development and corporate functions as contributors to operating leverage.

Full Earnings Call Transcript

Complete Earnings Call Transcript

Management Remarks

Operator

Hello, everyone. Thank you for joining us, and welcome to the Penguin Solutions Fourth Quarter 2026 Earnings Call. [Operator Instructions] I will now hand the conference over to Lana Ader, Investor Relations. Please go ahead.

Unknown Executive

Thanks, everyone, for joining us. With me today are Kash Shaikh, [ Stephen Comin ] and Aaron Johnson. Our earnings materials are available on the Investor Relations section of our website and I encourage you to review these materials.

Also, please take some time to review the presentation, which includes additional content to complement our discussion today. During this call, unless otherwise indicated, all references to financial measures refer to non-GAAP financial measures.

Non-GAAP measures should not be considered in isolation from, as a substitute for or superior to our GAAP results. A reconciliation of these measures to their most directly comparable GAAP measures can be found in our press release and accompanying slide presentation.

Statements made during this call that relate to future results and events are forward-looking statements based on current beliefs and assumptions and are not guarantees of future performance.

Actual results and events could differ materially from those projected due to a number of risks and uncertainties, which are discussed in our press release, our earnings call presentation and our SEC filings. Except as required by applicable law, we assume no obligation to update our forward-looking statements.

Now I'll turn it over to Kash.

Kash Shaikh

Good afternoon, everyone, and thank you for joining Penguin Solutions fiscal 2026 4th quarter and year-end earnings call. Before we get into our results, I want to start with an important addition to our leadership team.

Earlier today, we announced that [ Stephen Coming ] has joined Penguin Solutions as our new Chief Financial Officer. Stephen is an experienced public company executive with a strong track record in financial leadership, operational scaling and Investor Relations.

We are pleased to welcome Stephen at an exciting time for Penguin and investors will have the opportunity to get to know him in the coming months. For today's call, Aaron Johnson, who has served as our intra-CFO since July will walk you through our financial results and fiscal 2027 outlook as planned.

I want to thank Aaron for his leadership during the transition and I look forward to continuing to work with him as he returns to his role as Vice President of Finance and Accounting.

Now let's turn to our results. Our team delivered an exceptional quarter capping off a very strong fiscal year. In Q4, we set company records for net sales, gross profit dollars, operating income operating margin, net income and adjusted EBITDA.

Net sales were $567 million, up 68% year-over-year. and EPS was $1, up 133%, reflecting strong operating leverage that enabled EPS to grow faster than net sales. I would also like to thank our customers for their continued confidence in Penguin. Our AI-driven businesses continued to scale.

In Q4, non-hyperscale AI infrastructure and integrated memory represented 78% of company net sales and grew 141% year-over-year. Our overall company performance accelerated meaningfully in the second half of fiscal 2026 following the launch of our AI factory platform strategy and increased focus on the data center market.

We aligned the business more closely with strong AI-driven demand, increased investments in product innovation and accelerated go-to-market execution with a focus on neo cloud and enterprise customers. The proof is in the results.

After relatively flat year-over-year net sales in the first half, growth accelerated to 48% in Q3 and 68% in Q4 driving second half growth of 58%. Full year net sales were approximately $1.73 billion, up approximately 26% year-over-year. We also achieved record full year operating income.

In AI infrastructure and integrated memory, bookings grew faster than net sales in the fourth quarter and we ended fiscal 2026 with record company backlog and strong momentum heading into fiscal 2027. Based on our second half performance, bookings growth, continued strength of our memory business and more importantly, accelerating demand from Neo Cloud customers for our data center AI infrastructure solutions.

We are increasing our fiscal 2027 growth outlook from the preliminary view we shared last quarter. We expect record fiscal 2027 net sales with EPS growth expected to outpace net sales growth as we continue to scale and drive operating leverage.

I'll walk through what we are seeing in the market before turning to the outlook. AI is becoming super intelligence, I as it continues to move from chatbot adviser to agenetic operator, executing tasks and workflows with increasing autonomy. This shift is driving inference at scale and creating increasingly memory-intensive workloads.

Inference is increasing demand across the entire data center stack. GPU's remain critical, but CPUs and conventional memory, including DRAM, must also scale. We believe Penguin is well positioned to lead in this market opportunity at the intersection of data center, AI infrastructure and memory solutions. We are seeing enterprise and AI native companies address these AI compute infrastructure needs in a variety of ways, primarily consuming AI infrastructure from new cloud providers and, in some cases, building their own on-premises AI factories.

For persistent high utilization AI workloads, such as inference, large enterprises are increasingly considering on-premises AI infrastructure, which can offer better token economics, data control and performance. Our full stack AI factory platform combines 5 core elements with 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 under our compute AI brand. Fourth, Origin AI factory architectures; and fifth, end-to-end design, build, deploy and manage services.

What sets our platform apart is its rare combination of OEM like differentiated product innovations and end-to-end AI infrastructure systems integration capabilities. This allows us to serve as the single builder and operator of AI factories across the entire life cycle for our customers.

Customers need more than hardware and assembly. This is particularly important for neo cloud providers, where we have seen a recent surge in interest that we believe reflects strong product market fit for our full stack platform.

For new clouds, we can serve as a single operating partner across the entire life cycle. Helping customers meet service level agreements and generate revenue faster while they focus on securing additional offtakers and scaling their businesses.

Let me now turn to our business results, starting with advanced computing. In Q4, advanced computing net sales totaled $154 million. Representing 27% of company net sales and growing 11% year-over-year. Data center AI infrastructure growth within advanced computing driven by our AI factory platform was significantly stronger.

Non-Hyperscale AI infrastructure net sales grew 99% year-over-year and represented 66% of advanced computing net sales, up from 60% in the prior year quarter. This growth more than offset continued runoff from other traditional businesses and reflects the ongoing shift in advanced computing towards data center, AI infrastructure.

During the quarter, we added 6 new AI infrastructure customers including 4 Neo clouds, a large quantitative trading firm and an enterprise customer. We exited fiscal 2026 with strong momentum and bookings have accelerated further entering fiscal 2027. Our land and expand model remains central to our go-to-market strategy.

Successful initial deployments create opportunities to expand across our platform, including cluster wear AI and multiyear managed services engagements that typically span 3 to 5 years. Across fiscal 2026, we added 17 new AI infrastructure customers, while 12 customers expanded their business with Penguin.

We are seeing strong momentum with Neo Cloud customers, including a Neo Cloud backed by a leading South Korean technology company selected us to design, build, deploy and manage an NVIDIA GB-300 NVL-72-based platform with the initial deployment expected to support a broader GPU as a Service expansion.

A publicly traded neo-cloud provider that has more than $3 billion in signed multiyear AI infrastructure contracts selected Penguin for a multiyear engagement to provide AI infrastructure deployment and 24/7 operation services, supported by cluster were AI our AI factory operating system software.

Lektra, a neo cloud selected our full stack AI factory platform to deploy and optimize distributed AI micro data centers powered by existing carbon-free energy. We provide validated reference designs and video-based AI GPUs and expert services with cluster war AI and support.

And for another neo cloud provider with $10 billion in contracted compute from a leading AI lab, we were selected to deploy and operate a 36,000 AI factory in Norway. This multiyear engagement demonstrates the scale of our platform and our ability to bring large-scale AI infrastructure into production.

These engagements demonstrate an important evolution in our business. Customers are increasingly selecting Penguin as an end-to-end partner to design, deploy, optimize and operate AI infrastructure. We also continued to advance cluster wear AI with new Agentic AI capabilities that automate AI factory operations, remediation and infrastructure monitoring.

Our AI factory operations agent provides administrators with a natural language interface for operational insights. Put simply, administrators can now talk to their AI factory. We are also an NVIDIA AI factory specialized partner an invitation-only designation that reflects the depth of our partnership with NVIDIA, which continues to strengthen and more than a decade of experience building and managing NVIDIA-based AI factories.

Finally, we established a relationship with another AI hardware supplier to help expand supply availability and support growing demand. Now turning to integrated memory. Fourth quarter net sales reached a record $341 million representing 60% of company net sales, up 158% year-over-year and 24% sequentially.

In Q4 fiscal 2026, we added 2 new memory customer logos. Our land-and-expand strategy also continues to produce results. Across fiscal 2026, we added 15 new customers while 29 existing customers expanded their business with us. We are seeing an increase in both volume and pricing.

Our growth reflects strong data center demand driven by a Agentic AI, including increasing memory requirements from AI and inference workloads. While memory is a cyclical market, we believe AI is adding a structural demand driver particularly for memory solutions in the data center. Penguin operates differently from a traditional commodity memory manufacturer.

Our focus is primarily on data center solutions. Our engineering design, firmware development, advanced validation, negotiated pricing and asset-light model are designed to deliver more resilient economics across memory cycles. We continue to see strengthening bookings and pipeline for our CXL memory expansion products, a couple of proof points.

In Q4, we secured business with a leading global AI server manufacturer whose platform support large-scale AI training and inference. This win reinforces the strength of our memory portfolio for data center AI infrastructure.

We also expanded our relationship with a leading next-generation AI infrastructure inference provider, rapidly scaling AI inference deployments with our CXL memory expansion cards. Beyond our core memory business, we are developing new solutions designed to address the growing memory bottleneck in data center AI infrastructure.

Our CXL-based large memory expansion cards and memory AI KV cash solution are designed to increase memory capacity and improve the economics of Agentic AI and real-time inference. Our memory AI road map also extends to future memory appliances, including photonic high-bandwidth memory solutions designed to expand GPU memory capacity and bandwidth.

We are also developing a CXL AIC product line designed to repurpose installed DDR4 memory for AI infrastructure, helping customers extend asset life and lower the cost of scaling memory-intensive workloads. Finally, we established a new supply arrangement with a leading memory supplier to support the growth of our data center AI infrastructure portfolio and strengthen our supply position.

Together, these developments reinforce our long-term focus on data center memory and AI infrastructure. Turning briefly to optimized LED, our LED business also performed well in the quarter, with Q4 net sales totaling $72 million, up 7% year-over-year.

Our strategic investment priorities remain focused on data center, AI infrastructure and memory, while we continue to manage our LED business with discipline. As we enter fiscal 2027, our focus is not simply on capturing demand. It is on building a more repeatable and scalable operating model around that demand while keeping customer success at the center of everything we do.

As part of that, we are deploying Agentic AI across our operations with a focus on measurable business outcomes. Our engineering teams are using agentic AI coding workflows to accelerate cluster are AI development.

While AI-driven tools are helping compress product cycles and improve supply chain responsiveness. With that, let me share our updated fiscal 2027 outlook. During our Q3 earnings call, we provided preliminary fiscal 2027 expectations for net sales and diluted EPS growth of approximately 30% year-over-year from the midpoint of our fiscal 2026 outlook.

Based on our fourth quarter results and continued AI-driven demand, we are providing a fiscal 2027 outlook that exceeds that preliminary view for both net sales and diluted EPS. We expect our diluted EPS to grow faster than net sales due to our strong operating leverage.

Aaron will provide the details in a moment. Let me close with 3 observations. First, our strategy is working and growth is broadening. Our AI-driven businesses represented 78% of Q4 net sales and grew 141% year-over-year with integrated memory remaining strong and AI infrastructure now accelerating significantly.

Second, strong operating leverage is driving profitable growth while enabling continued investment in innovation. EPS is growing faster than net sales, while we continue investing in differentiated products and capabilities.

Third, we believe the value creation opportunity is durable and still in the early innings with our position at the intersection of AI infrastructure and memory combined with cluster ware AI and services, creating a differentiated long-term advantage.

I want to thank our team for their focus and execution throughout fiscal 2026 and for the strong momentum we carry into fiscal 2027. With that, I'll turn it over to Aaron to walk through the financials and our outlook.

Aaron Johnson

Thank you, Kash. I'll begin with our fourth quarter and full year fiscal 2026 results and then turn to our outlook for fiscal 2027. In the fourth quarter, net sales and earnings per share exceeded our outlook, reflecting growing adoption of our AI infrastructure solutions and continued AI-driven demand for data center memory.

Strong net sales growth, disciplined expense management and improved profitability drove meaningful operating leverage. We set company records for net sales, gross profit dollars, operating income, operating margin, net income and adjusted EBITDA.

For the full year, we also set records for gross profit dollars and operating income. Starting with the fourth quarter, net sales were $567 million, up 68% year-over-year and up 18% sequentially. Gross margin was 28.8%, operating margin was 15.8% and diluted earnings per share was $1, up 133% year-over-year.

For the full year, net sales were $1.73 billion, up 26% versus fiscal 2025 while diluted earnings per share increased 51% to $2.87, roughly twice the rate of net sales growth, demonstrating the operating leverage in our model. Fourth quarter product net sales were $510 million or 90% of total company net sales, up 86% year-over-year.

Services net sales were $57 million or 10% of total net sales down 11% year-over-year, reflecting lower hyperscale services volumes, partially offset by growth in multiyear AI factory managed services. Net sales by business segment were as follows: for advanced computing, fourth quarter net sales were $154 million, up 11% year-over-year and 12% sequentially.

For the full year, advanced computing net sales were $559 million, down 14%, reflecting the wind down of Penguin Edge and lower hyperscale hardware sales. Non-Hyperscale AI infrastructure business grew 99% in the fourth quarter and represented 66% of segment net sales, while growing 72% for the year and reinforcing its position as the segment's primary growth driver.

For integrated memory, fourth quarter record net sales were $341 million, representing 60% of total company net sales, up 158% year-over-year and 24% sequentially. For the full year, integrated memory net sales were $924 million, up 99% and represented 53% of total company net sales, up from 34% in fiscal 2025.

The growth reflects strong demand for memory supporting AI-driven data center applications, consistent with our strategic focus on this market. For optimized LED, fourth quarter net sales were $72 million, representing 13% of total company net sales, up 7% year-over-year. For the full year, net sales were $249 million, down 3%.

The business delivered improved profitability as we continue to execute with discipline and focus on profitable opportunities. Fourth quarter gross margin was 28.8%, up 70 basis points sequentially, supported by improved profitability in integrated memory partially offset by lower advanced computing margins.

Together with strong net sales growth, this drove record gross profit of $163 million, up 21% sequentially and 57% year-over-year. For the full year, gross profit reached a record $508 million. Gross margin was 29.3%, down 1.7 percentage points, primarily due to a greater contribution from AI infrastructure hardware even with stronger margins in integrated memory.

The growth in gross profit dollars demonstrates the leverage generated by our increased scale. Fourth quarter operating expenses were $73 million, up 4% sequentially and 13% year-over-year. The sequential increase primarily reflects higher incentive compensation associated with our strong financial results.

For the full year, operating expenses increased 4% to $267 million, well below our 26% growth in net sales. As a result, operating expenses declined by 3.4 percentage points as a percentage of net sales, reflecting a structural improvement in the scalability of our operating model.

That combination of higher gross profit dollars and expense discipline drove record fourth quarter operating income of $90 million, up 39% sequentially and 129% year-over-year. With net sales growth significantly outpacing operating expense growth, operating margin expanded 2.4 percentage points sequentially and 4.2 percentage points year-over-year to 15.8%.

For the full year, operating income reached a record $241 million, up 44%, with operating margin expanding 1.7 percentage points to 13.9%. These results demonstrate how scale and disciplined spending are translating growth into higher profitability.

Fourth quarter diluted earnings per share was $1, up 19% sequentially and 133% year-over-year. The sequential EPS growth rate trails our 39% operating income growth, reflecting a higher share count following our convertible notes offering.

For the full fiscal year, diluted earnings per share were $2.87. The up 51% and $0.22 above the high end of our July outlook. Adjusted EBITDA was a record $93 million in the fourth quarter, up 115% year-over-year and 38% sequentially.

Full year adjusted EBITDA was $256 million, up 37%.

Before I walk through working capital, a reminder that we calculate days sales outstanding, days payables outstanding and inventory days on a gross sales and gross cost of goods sold basis. Gross sales were $1.57 billion, and gross cost of goods sold was $1.42 billion in the fourth quarter.

The difference between gross and net sales relates primarily to our memory businesses logistics services which are accounted for on an agent basis. Turning to working capital. Net accounts receivable increased to $796 million from $308 million a year ago, reflecting the growth in our memory business and the associated gross billings.

Despite the higher balance, days sales outstanding improved to 46 days from 53 days last quarter and 51 days a year ago. We are managing these investments against customer deployment schedules and supply availability as we support growth.

Inventory increased to $749 million from $255 million a year ago, reflecting purchases to support record backlog and expected customer deployments across our memory and AI infrastructure businesses as well as higher memory costs. Inventory days were 48%, up from 42 last quarter as we positioned inventory against a record backlog and down from 51 a year ago.

Accounts payable increased to $817 million from $267 million a year ago, reflecting purchasing activity to support the growth of our memory and AI infrastructure businesses as well as higher memory costs. Days payable outstanding was 52%, down from 62% last quarter and 54 a year ago, primarily reflecting the timing of purchases within the quarter.

Our cash conversion cycle was 42 days, compared with 33 days last quarter and 49 days a year ago. Funding growth efficiently and maintaining disciplined working capital management remain important priorities. Moving to our broader capital structure. During the quarter, we completed a significantly oversubscribed $750 million 0% convertible note offering due in 2031 which we believe reflects strong investor confidence in our business and long-term strategy.

The notes have an initial conversion price of approximately $116.70 per share, representing an approximately 50% premium to our share price at pricing. We also entered into capped call transactions designed to reduce potential dilution up to an initial cap price of approximately $175.05 per share.

Concurrently, we exchanged approximately $296 million principal amount of our 2029 and 2030 convertible notes for a combination of cash and shares of common stock and used a portion of the proceeds to fully repay the $100 million outstanding under our credit agreement.

This disciplined and proactive approach extended our maturity profile and strengthened our capital structure, providing additional flexibility to invest behind our AI factory platform strategy and the growth opportunities ahead. Following these actions, we ended the fourth quarter with cash and cash equivalents of $647 million. That is up $207 million from the third quarter and $193 million from a year ago.

We remain focused on managing liquidity and the working capital required to support our growth. Cash used in operating activities was $163 million in the fourth quarter and $152 million for the full year, primarily reflecting working capital investments to support growth in our memory and AI infrastructure businesses.

This compares with cash used of $70 million in the prior year quarter and cash provided of $113 million in fiscal 2025. We believe our capital structure and liquidity provide flexibility to fund this growth while improving working capital efficiency remains an important priority for fiscal 2027. Capital expenditures were $4 million in the quarter and $12 million for the year, well under 1% of net sales, with depreciation expense of $5 million and $20 million, respectively.

Our asset-light model allows us to scale without significant capital spending, and we expect that to continue in fiscal 2027. Now turning to our outlook. We enter fiscal 2027 with strong momentum, supported by our second half performance, record backlog, bookings growth, continued integrated memory strength and further acceleration in demand for our AI infrastructure business.

We expect strong net sales growth to increase gross profit dollars while operating expenses grow at a considerably slower rate. That combination is expected to expand operating margin and drive diluted EPS growth that outpaces net sales growth.

For fiscal year 2027, we expect net sales to grow approximately 40% at the midpoint, plus or minus 10 percentage points or an increase of approximately $700 million to approximately $2.43 billion at the midpoint. Last quarter, we shared a preliminary view of approximately 30% growth from the midpoint of our then current fiscal 2026 outlook.

Using that same fiscal 2026 outlook midpoint as the starting point. Our current fiscal 2027 midpoint represents growth of approximately 45%. The range reflects different growth expectations across our businesses as well as variability in customer deployment schedules and the availability, lead times and cost of certain products and components.

Our record backlog, together with the demand we are seeing across the business provides a strong foundation for our fiscal 2027 growth expectations. At the same time, customer deployment schedules, product availability and evolving memory market conditions could affect the mix and cadence of net sales throughout the year.

We believe the range appropriately balances the opportunities ahead with these execution and market variables. Looking at the net sales outlook by segment. For advanced computing, we expect full year net sales to grow approximately 40% at the midpoint, plus or minus 10 percentage points, driven by continued momentum in non-hyperscale AI infrastructure partially offset by the absence of edge sales and lower expected hyperscale sales.

For integrated memory, we expect full year net sales to grow approximately 50% at the midpoint, plus or minus 10 percentage points driven by continued strong demand for memory supporting AI-driven data center applications. And for optimized LED, we expect full year net sales to be relatively flat year-over-year as we continue to manage the business with discipline.

We expect full year gross margin of approximately 28%, plus or minus 2 percentage points. Our outlook reflects the expected mix across our businesses, including the mix of hardware, software and services as well as our current assumptions for memory pricing and costs.

While these factors influence the gross margin rate, we expect net sales growth and scale to drive higher gross profit dollars. We expect full year operating expenses of approximately $275 million, plus or minus $10 million. Within that range, we are reallocating resources toward our highest return AI infrastructure and memory programs while holding total spending growth well below net sales growth.

This disciplined investment approach, together with higher gross profit dollars is expected to drive further operating margin expansion. For the full year, we expect diluted earnings per share of approximately $4.45 plus or minus $0.70, representing growth of approximately 55% at the midpoint and exceeding our expected rate of net sales growth.

The EPS outlook reflects higher gross profit dollars, disciplined operating expense growth and further operating margin expansion. The range also allows for variability in the timing of large customer deployments as individual project schedules can shift between quarters.

Our EPS outlook assumes a diluted share count of approximately 63 million shares and an effective tax rate of approximately 20%. We anticipate using this normalized rate throughout fiscal 2027, though it may change with our geographic earnings mix and developments in the global and U.S. tax environment.

Our fiscal 2027 outlook is based on the current demand environment, backlog, customer deployment schedules, expected product mix and our present assumptions regarding supply availability and costs. The timing of customer deployments and product availability may create variability between quarters, and certain components continue to have extended lead times that can affect project and shipment timing.

Our outlook also incorporates current assumptions for memory supply and costs, which may evolve during the year, including the expected impact of new memory and AI infrastructure suppliers that strengthen our overall supply position and support growth.

These factors may affect quarterly cadence and mix but are reflected in our full year planning assumptions. Overall, our outlook reflects the momentum in our AI-driven businesses and the benefits of a more scalable operating model. We plan to invest selectively behind our highest return opportunities while maintaining expense discipline. Our objective is clear: convert strong net sales growth into higher gross profit dollars operating margin expansion and earnings growth that outpaces net sales growth.

Please refer to the non-GAAP financial information section and the reconciliation of GAAP to non-GAAP measures in our earnings release and the investor materials available on our website for additional detail.

With that, operator, we are ready for Q&A.

Operator

[Operator Instructions] Your first question comes from the line of Katherine Murphy with Goldman Sachs.

Question-and-Answer Session

Katherine Murphy

Very helpful to see the improved advanced computing segment guidance for 40% plus or minus 10% for fiscal 2027. I think last quarter, you had talked about mid-teens growth as being the preliminary starting point for your outlook there. My first question would be, can you help us think about what has improved in the last 90 days that allows you to take up your outlook for advanced computing?

And as a follow-up, as Penguin expands more into the Neo Cloud customer type, how does that change the unit economics relative to some of the enterprise customers and quant wins that you had in the first part of 2026? Is there more hardware? How do we think about the margin impact? Anything that you could share to help us from a modeling perspective would be helpful.

Kash Shaikh

Thanks, Kath for the questions. To your first question about our initial preliminary guidance last quarter in the mid-teens over to for advanced computing, mainly driven by AI infrastructure. So this is primarily the growth that we have seen since the last earnings announcement.

The booking strength, we have booked more than we expected, and we have a very strong pipeline and as you heard, some of these bookings that I shared as examples, give us the confidence that our memory business strength continues and our AI infrastructure is further accelerating, which is really resulting into the increased guidance for AI infrastructure, driving advanced computing guidance and also increasing the guidance at the company level, primarily driven by AI infrastructure.

And on your question about neo clouds, and unit economics. Unit economics are quite similar to our enterprise on-premise AI factory deployments. However, one of the things we have observed with the new cloud customers, we have a pretty unique product market fit for these neo cloud customers.

As in, we can provide them end-to-end build and operate their factories on their behalf, including procurement of the hardware end-to-end designing the hardware, building the data center and then taking on the management of the data centers, day 2 and beyond for up to 3 to 5 years, as I shared in some of the examples, along with our cluster wear, which is really our differentiation in terms of how we deliver the services and, more importantly, meet the SLAs, which are really important for these new cloud customers to be able to provide SLAs to their offtakers.

But all in all, similar unit economics, further expansion and pretty unique product market fit with new cloud customers.

Operator

Your next question comes from the line of Sajal Dogra with Rosenblatt Securities.

Sajal Dogra

You're essentially front-running my bull case by a full year, so congrats. I guess for my question, on your previous outlook, your revenue growth and earnings growth was roughly in line but now you're finally showing operating leverage for next year. So I just want to understand what's changed versus 90 days ago that's allowed you to drive that earnings expansion. And I was wondering, like, is that all -- like is it -- how much of that is memory -- or are we finally seeing some margin expansion in advanced computing as well?

Kash Shaikh

Sure. So Sajal, first of all, the outlook we provided last quarter was more of a color and preliminary outlook than the actual outlook because it was sooner than we provide, right? As typically, we provide in Q4 earnings, the time frame where we are providing the update.

So that was much more directional versus what we are providing is looking at all the aspects, and we have more visibility and more clarity and also the confidence of operating in this new model where if you look at Q3, our EPS grew faster than net sales company level. And then same in Q4, net sales grew 68% versus the EPS growth of 13%.

And the line of sight we have in terms of the bookings and then the conversion ratios and so on and so forth. We have much more clarity and it is much more of a formal outlook than the high-level color we provided directionally. So as we are closer, we have more visibility, and we have more confidence in continuing the model as we scale the business.

And in terms of the margins at the segment level, we won't disclose the margins at the segment well. But I can tell you at the high level, memory continues to be strong, stronger than we expected. Given the value we provide and the focus on the data center.

Advanced computing overall has the advantage of the services and the software. However, as we increasingly have the wins where we have both hardware and the software, AI infrastructure may have some downward pressure but at the same time, the advantage of cluster wear, software as well as the services has a very high attach rate in general with these opportunities that I shared in the earnings call earlier and the rest of the pipeline we see. So that's at the high level, to give you an idea of how we are looking at the segment margins.

Operator

Your next question comes from the line of Ananda Baruah with Loop Capital.

Ananda Baruah

Yes, really appreciate it. What are some of just sort of in keeping with Kash, Aaron and Stephen welcome. The dynamics that are sort of underpinning sort of the -- not just the increase in the fiscal year '27 guide, but the acceleration of the growth rate actually across the key businesses, what are some of the mechanics that you're seeing out there in the marketplace that are leading to sort of increase neo cloud capture.

What are you seeing to the degree that you're able to discern from commercial enterprise and would love to just get a little context, if it's worth providing on the degree that we've the NVIDIA and Dell relationship are actually contributing to some of these dynamics at an increased pace.

Kash Shaikh

Sure, Ananda. So starting with the acceleration of the business, as I mentioned earlier, part of it is the increased visibility that we are seeing as well as the growth acceleration.

We believe one of the competitive advantage we have is we are not a product company that is just providing the hardware and leaving at the doorstep of the customers, we actually worked with them much earlier in the cycle, get involved in their design based on the requirements they have, whether they are a large enterprise customer deploying an on-premise factory or a cloud provider and then work with them [indiscernible] understanding their requirement to -- if it is a new cloud provider, what is their SLAs, who are their customers and how can we align the design to meet their requirements.

So that consultative aspect and especially the fact that we've been doing it for more than 2 decades. That's an advantage. So our team was working with NVIDIA designing these clusters for more than a decade. So we have this advantage of understanding the architecture. The discussions are at the architecture level. We have the software, the cluster were that brings together all of these elements and makes it much more easier to build and manage these factories.

And then end-to-end staying with them and helping them manage as a part of our managed services for them to get those outcomes when they are investing so much capital in these AI factories or data center is really a unique advantage as compared to the alternatives they have whether they go to product-only companies or services companies, there isn't a unique combination that we provide with our platform between the products we have and the services that we provide.

So that's one of the advantage we are realizing is helping us. NVIDIA partnership is certainly helping us because we are we were recently nominated by NVIDIA as the AI factory specialized partner. We are the -- one of the few handful of partners they have and this is really the experience working with them and all of the things our team is capable of between design, build, deploy and manage helps us have that credibility.

But again, when we get involved in these designs and we are having the white-boarding sessions with the customers to meet their requirement, that's a very unique competitive advantage we have along with our products.

So that's mainly for obviously, AI infrastructure, which is accelerating. In the Memory business, one of the things we have done since we launched our AI factory platform strategy midway through the fiscal 2026.

We've been focusing more on the data centers for 2 reasons. One is the fact that while memory demand is high, everywhere in other segments, including the consumer segments, we believe that demand may still be just a demand and supply dynamics and maybe cyclical however, the demand in the data center for memory is driven by AI, especially with the agent AI workloads, which are automating the workflow and creating really the productivity for the enterprises, that is going to sustain much longer.

And if anything, let's say, agent TKIs adopted in the enterprise is 10% because they are still getting the work flows defined and the tool is defined for every 10% increase, that demand can double. And we continue to see that demand increasing for our memory products in the data center so that positioning in the data center is another advantage, both in terms of increased demand as well as the durability of the demand for us.

So combine that with these 2 tailwinds we have, Memory business position in the data center solving memory issues providing the solutions for OEMs and now increasingly providing forward-looking data center solutions such as CXL memory expansion cards.

Those memory expansion cards are primarily for inference workloads and BC this CXL business taking up very strongly. So the pipeline is increasing, bookings are increasing. So those newer solutions, data center focus, inference focus is another advantage for us as our differentiation and solving the inference challenges in the data center.

So those are some of the things. In addition to that, one more thing we are seeing is in general, going back to AI infrastructure. Within the enterprises, they are primarily using New Cloud as a Service for the AI infrastructure.

However, in some cases, they are considering on-prem, especially large enterprises because of the fact that at scale when they are using inference applications, cost becomes a challenge. Especially with inference workloads. At that point, while they are getting the productivity, they are not seeing the ROI for the dollars they are investing and they consider on-prem factories and then we get involved, and they are looking to address their cost challenge.

And they are also considering the open weight models because that's another way for them to make the economics better as they are moving from a cloud-based service to on-premise. So these are some of the trends where we have a very unique fit between our memory business as well as AI infrastructure business that is driving the growth for these 2 businesses.

Operator

Your next question comes from the line of Rustam Kanga with Citizens.

Rustam Kanga

Great. And congrats on a memorable close to the year with continued momentum in the AI infrastructure-driven business. cash, you stated that the memory growth was a function of both price and volume. I believe last quarter, you made an interesting point that if you were to count the backlog that you couldn't ship because of supply constraints. That, that split would be closer to even or even lean more towards volume.

So I'm just curious how much of that backlog has started to convert here? And then as you look ahead to next year, how are you thinking about that price versus volume mix in the memory business, especially as you focus on new customers, primarily with data center solution use cases.

Kash Shaikh

First of all, I want to give you the credit for metering my comment. And you're right, that was what we observed last quarter, where the revenue was pretty strong record revenue for memory even though the backlog was as much as our revenue.

Fortunately, we are seeing the same trend continue. So as much as we had another record quarter for memory, our backlog, if any, is much higher than the revenue that we recognized in this quarter. and the backlog now extend to at least 4 quarters, and it continues to strengthen, which is why we have the confidence to increase the outlook for the business.

And it is really, as I mentioned in the previous question that it's a function of data center-focused demand for our core memory business as well as the new CXL-based products that we are developing had a very strong pipeline and demand because of inference taking off in the data centers.

Operator

Your next question comes from the line of Matthew Calitri with Needham & Company. Matthew, your line is now open.

Matthew Calitri

Matt Calitri Lariat, Needham here. And great to see the strength to close the year here, including with some of those new cloud wins you talked about. I'm wondering if you could give any more detail there on like when will these projects start and begin to like be recognized in the model and what the impact could be?

And also, like, is there anything specific to point to as to why you had so much success closing deals in the quarter? Or just like more broadly, like why now is the moment for this neo-cloud opportunity?

Kash Shaikh

Yes. So we are not sharing that level of details in terms of which project when, but I can tell you at the high level, most of these projects that I shared have already started as in we are managing their data centers over building their data centers.

And some of them, we are in the process of procurement. So they are not future. They are already happening. In fact, we were working on these projects for several months, even though some of it has accelerated into the bookings recently. We were working on these new cloud opportunities, along with other enterprise opportunities which is why I was referring to Neo Cloud as a growth opportunity for us in the previous earnings announcement even if we were not in a position to share the details now that we have started working with them, and we are managing these factories, in some cases, and we are seeing increased bookings.

That's why we are sharing more details, and we will continue to share more details around these new cloud deployments. But as I mentioned, one of the reason being the unique product market fit we have between the products we provide as well as our expertise in designing these AI factories or data centers, decade-long expertise to be able to help them achieve their goals is it's a pretty unique competitive advantage we have in the market with these opportunities.

Operator

Our next question comes from the line of Brian Chin with Stifel.

Brian Chin

Congratulations on the results, and thanks for letting us ask a question. Maybe kind of 2 partners here. Given that record backlog commentary coming into the fiscal year and the growth guide for computing roughly how much of your fiscal '27 advanced computing revenue outlook is already in backlog.

And then the other part of this question is obviously, really strong flow-through in terms of the model and the EPS -- non-GAAP EPS guide for next year. But on that really kind of low OpEx increase year-to-year, are you investing enough in terms of head count, retention areas to maybe kind of improve bandwidth and utilization and go after more projects.

With the revenue growth, it seems like you have a lot of coverage maybe to invest even more.

Kash Shaikh

Right. So let me address your first question. We are not sharing the details of how much of our backlog is covering the revenue. However, I can tell you at the high level, it's pretty strong, pretty strong backlog across both memory as well as AI infrastructure.

That's why we'd effort to a record backlog. And in terms of your question about operating expenses. So when we say the strong operating leverage, it is really a function of the efficient operating model we have. And when I say efficient operating model, to be specific, for example, for R&D, research and development and our engineering.

Predominantly 95-plus percent of our engineering is in India, low-cost location which gives us an advantage to scale much more cost effectively. We also have our memory engineering in Taiwan, another cost-effective location. So even as we scale our cost is very efficient as compared to other companies who may not have as much of a cost-efficient model.

And then our manufacturing is in Malaysia, all low-cost location even as the business scales, that's a fixed cost for us as we add that capacity and we have the capacity. So even if we add the capacity, the cost or the expenses don't increase linearly with that revenue. We are also using AI pretty extensively across the company.

So for almost 8 or 9 months, we've had a lot of productivity gains. As an example, 100% of our software engineers wherever they are, are using core generation for -- so AI for core generation, which is helping us scale beyond what we can do in a traditional model.

We are using AI in pretty much every functions, including finance, legal, and that is helping us scale much more effectively. So between building this model and locations that are in cost-effective location and scaling in those areas as well as using AI -- it is a pretty efficient model.

But to answer your questions, for sure, we are investing in innovation, which is why we are winning more. The products I mentioned, cluster varies AI the differentiation, helping us win more XL for the memory products is another advantage, and we will continue to invest more.

So while the dollars may not seem to increase the productivity as well as the scale is increasing, based on our efficient model driving the operating leverage, which allows us to continue to deliver more profitability. And if you look at our operating profit is a record of $90 million, was a record last quarter, almost $64 million.

And we continue to expect that while we will drive profitable growth but based on a scalable model, official model we have created, we can increasingly invest in innovation, and we will even though the productivity will continue to increase for the company.

Operator

There will be no further questions at this time. This concludes today's call. Thank you for attending. You may now disconnect.