The Zhitong Finance App learned that Super Micro Computer (SMCI.US) announced the results for the first quarter of fiscal year 2027 after the market on Tuesday and held a results conference call.
Q1 revenue of $19.05 billion, up 298% year over year, was higher than FactSet's agreed estimate of $17.53 billion; non-GAAP diluted earnings per share of $1.81, expected to be $1.23, and GAAP diluted earnings of $1.75 per share. The non-GAAP gross profit margin of 12.8% was higher than the company's previous estimate of 12.0% — 12.5%, but it fell 4.8 percentage points from 17.6% in the previous quarter, making it the most intensive questioning point for phone calls. After the financial report was released, the stock rose more than 7% after the market, then expanded to around 10%.
Chief Financial Officer Mark Long responded that the month-on-month decline in gross margin was “not evidence that AI platform revenue lacked profit margins”, reflecting which customers picked up goods in which quarter; looking at the fourth quarter and the first quarter combined, revenue for the two quarters was about US$31.1 billion, and non-GAAP gross margin was about 14.5%, an increase of about 5 percentage points over the previous year. The company raised its revenue target for the 2027 fiscal year from US$65-72 billion to US$72-82 billion, but at the same time, the gross margin for the second half of the year is expected to be lower than that of the first half of the year.
The most focused question in this call was the sharp fluctuation in gross margin between quarters. The non-GAAP gross margin for the first quarter was 12.8%, down 4.8 percentage points from 17.6% in the previous quarter, while the share of AI solution revenue rose from 59% to 71% during the same period. As a result, the market is concerned about whether large AI orders are diluting profitability. Chief Financial Officer Mark Long refuted this: “The month-on-month decline is not evidence that AI platform revenue lacks profit margins; it reflects which customers are picking up goods in which quarter.” He suggests looking at a two-quarter consolidated scale: the total revenue for the fourth quarter and the first quarter was about US$31.1 billion, non-GAAP gross margin of about 14.5%, compared to US$10.8 billion and about 9.5% for the same period last year, that is, revenue nearly tripled, and gross margin increased by about 5 percentage points. At the same time, he acknowledged that the gross margin range for the full fiscal year was implied to be lower in the second half of the year than in the first half of the year, because the share of AI revenue will move higher towards the 70% to 80% range. J.P. Morgan Chase indicated in a review after the pre-announcement that this quarter's gross margin dropped significantly by 17.6% compared to the previous quarter, and the sustainability of profit margins is still the focus of attention. The stock had a cumulative decline of 27% in the previous month, and market concerns about AI capital expenditure efficiency were the main background.
In his opening statement, Mark Lang pointed out that the revenue of about $2.3 billion in the first quarter corresponds to deployments that were originally scheduled to be delivered earlier and were postponed due to customer and data center readiness. After excluding this portion, the underlying revenue was approximately $17.6 billion. He stressed that the full-year revenue range does not assume a repetition of this catch-up revenue; it reflects continued month-on-month growth in basic demand and significantly better visibility of customer deployment schedules than 90 days ago.
A large data center customer contributed 44% of revenue this quarter, compared to 28% for the full year of FY2026; 10-Q also showed that the two customers accounted for 49% and 17% of accounts receivable at the end of the quarter, respectively. Mark Lang explained that this concentration is a product of the current scale of AI infrastructure construction. “The scale of a single deployment today may exceed our single quarter revenue a year ago.” It is expected that the concentration will remain high during the construction period and decline as corporate and sovereign businesses expand. He added that deferred revenue increased by about $3.1 billion during the quarter, mostly due to prepaid customer payments. The number of accounts receivable turnover days dropped from 59 days to 36 days, and no significant credit losses were recorded during the quarter.
Operating cash flow remains negative, and cash conversion cycle shortened by 52 days
Net outflow from operating activities for the first quarter was $434 million, capital expenditure was $47 million, free cash flow was negative $481 million; operating cash flow for the full year 2026 was negative $6.8 billion. The company's cash conversion cycle was reduced from 149 days to 97 days, an improvement of 52 days in a single quarter. Based on the revenue scale of the first quarter, this is equivalent to a reduction of approximately $10 billion in working capital. Among them, the number of inventory turnover days improved by 45 to 74 days, the number of receivable turnover days improved by 23 to 36 days, and the number of payable turnover days was reduced from 29 days to 13 days, consuming approximately $3 billion in cash. Cash and equivalents at the end of the quarter were $6.9 billion, bank loans and convertible bonds totaled $8.6 billion, net debt of $1.7 billion, and inventory of $15.05 billion. The company said it met all the terms of the debt contract at the end of the quarter.
By December, more than 8,000 racks were produced per month, and DCBS and software formed a profit gripper. Liang Jianhou said that the company is expanding fully automated production capacity in Silicon Valley and Taiwan, and is speeding up containerized data center production in Asia and the US. The goal is to produce more than 8,000 racks per month, including more than 4,000 liquid-cooled racks by December of this year. He repeatedly emphasized the commercial value of DCBS (Data Center Modular Solution): covering the deployment of a gigawatt-scale AI factory to a single-container POD, combined with the three software packages SuperCloud Composer, Supermicro Datacenter Manager, and Supermicro Operations Manager, and a newly expanded global field service organization to form a bundle of hardware, software, and services. He sees this combination as a path for gross margin back into the double-digit range; in the Q&A, he added that DCBS includes not only hardware modules, but also software and services, and has become a business model for the company.
The company expects second-quarter revenue of $19.5 billion to $21.5 billion, gross profit margin of 12.0% — 12.5%, GAAP diluted earnings of $1.59-1.83 per share, and non-GAAP diluted earnings of $1.66—1.90 per share. The revenue target for the full fiscal year was raised from US$65-72 billion to US$72-82 billion. The median value of US$77 billion was about 12% higher than the median target; the annual gross margin guideline was 11.8% to 12.2%, implying that the second half of the year was lower than that of the first half of the year. Mark Lang made it clear that the constraining range is not demand or production capacity, but rather the speed at which customer data centers receive the equipment already produced and the supply of key components. “Currently, the hard constraints for AI construction in the entire industry are land, electricity, and data center housings.”
In response to questions about AI capital expenditure efficiency, Liang Jianhu said that the company can directly solve customer ROI issues
In response to recent questions from investors and market experts about AI capital expenditure efficiency and return on investment, Liang Jianhou said that the company has been focusing on server and storage manufacturing in the US for more than 30 years, and has a local engineering and manufacturing base that is difficult for competitors to match. By combining DCBBS's hardware, software and services, it can improve system availability and energy efficiency by reducing the delivery cycle (TTD), launch cycle (TTO), and improving system availability and energy efficiency, “so that customers' capital investment can be converted into revenue faster and more cost-effectively than anyone else.” At the same time, he revealed that the inventory and production turnaround time has been reduced from about 100 days to 74 days, with the goal of further reducing it to 50 days in the next few quarters.
New CFO Debuts, Says It Will Focus on Financial Discipline and Transparency
This was Mark Long's first conference call since becoming Chief Financial Officer. He said that he has been working in finance in the field of technology infrastructure for a long time and joined Ultraman because “the hardest part has already been built” — engineering advantages, leading liquid cooling and rack-level integration, and a card position in the largest infrastructure cycle in history; what the company currently needs is financial capacity that matches engineering and operational capabilities. He focused his work on financial discipline, transparency, more in-depth shareholder communication, working capital management and capital structure optimization, and said he will explain the dilution, maturity structure, and available quota every quarter. In response to a question from Bank of America, he said that the company currently has no equity financing plan, and that the debt market depends on available conditions and agreed leverage structures.
Full minutes of the meeting
Opening remarks and management statements
Moderator: Welcome to Ultramicrocomputer's first quarter results conference call for the 2027 fiscal year. Attending the conference today were Founder, Chairman and CEO Liang Jianhou, Chief Financial Officer Mark Lang, former Chief Financial Officer David Wiegand, and Michael Steiger, Senior Vice President of Corporate Development.
Michael Steiger (Senior Vice President of Corporate Development): Thank you, good afternoon, and welcome to the Ultramicrocomputer FY2027 first quarter conference call. I was accompanied by founder, Chairman and CEO Liang Jianhou, Chief Financial Officer Mark Long, and former Chief Financial Officer David Wiegand.
At this point, you should have received the press release issued by the company after the closing of the regular trading period. The press release can also be viewed on the company's website. The call is being broadcast live on the company's investor relations website, and the replay will be held until the second quarter of FY2027 results call. The content of today's conference call is the property of Ultramicrocomputer, and may not be reproduced or transcribed without the company's prior written consent.
In this call, we may make forward-looking statements based on current expectations, including business plans and performance, ongoing market trends, and future financial results. These statements are subject to a number of significant risks and uncertainties, and actual results may differ materially. For factors that may affect future financial results and business, please refer to documents such as today's results announcement, our recent 10-K and 10-Q filings, and our possible 8-K filings with the U.S. Securities and Exchange Commission.
All of our statements are based on today, November 3, 2026, and are based on the information currently available to us. We are under no obligation to update such statements except where required by law. In this conference call, we will be discussing non-GAAP financial metrics, and related reconciliation statements can be found in a presentation posted on the company's website.
Our quiet period for the second quarter of fiscal year 2027 begins when business closes on Friday, December 11, 2026. Next, I'll hand over the phone call to Liang Jianhao.
Liang Jianhou (Founder, Chairman and CEO): Thank you Michael, and thank you all for attending the conference today.
After FY2026's annual revenue nearly doubled, FY2027 kicked off with a record. First-quarter revenue reached $19.9 billion, exceeding our pre-announced ceiling and well above our initial forecast of $15 billion. This is a remarkable achievement that shows that more and more customers prefer our overall AI and IT solutions. Most importantly, our Data Center Modular Solution (DCBS) is at the core of this growth, bringing deep customer trust and preference, improving our profitability, and creating sustainable long-term shareholder value.
For more than 30 years, ultra-microcomputers have focused on providing customers with optimized overall solution value. We will continue to accelerate customer deployments with engineering technology, global resources, and high-yield manufacturing capabilities. Furthermore, I am happy to see that not only large cloud vendors and new cloud vendors, but also companies and institutions in the vertical fields of enterprise, storage, IoT, and national and enterprise “sovereignty” rely on and adopt ultra-small overall solutions every day.
In the first quarter, we delivered on our promise to increase bottom line profitability. Non-GAAP gross margin increased 330 basis points year-over-year to 12.8%. This improvement comes from the optimization of enterprise AI and IT customer structures, a diversified product line based on system modules, and an increase in the share of revenue contributed by the more profitable DCBBS moat, which includes DCBBS hardware, software, and unique services. This is directly in line with our strategic roadmap: building a sustainable double-digit gross margin model by providing better technology innovation, faster time to market (TTM) and time to market (TTO), and overall solution value.
After more than two years of actual customer deployment verification, DCBS brings four core values that are unrivaled by traditional server OEMs and overseas foundries:
First, optimized delivery time (TTD): factory-tested, plug-and-play rack-scale systems and DCBBS modules (including complete data center management software) are delivered directly to the customer site.
Second, optimize time to launch (TTO): Shorten the time for computing power to be ready to go live from a few quarters to a few months, and further shortened to weeks.
Third, DCBS uses modular design to eliminate heat dissipation, power supply, software and service bottlenecks to maximize system availability.
Fourth, reduce total cost of ownership (TCO): Dramatically reduce energy consumption and facility downtime, and accelerate and improve data center profitability.
Whether the customer is building a gigawatt-scale AI factory or deploying a single-container POD solution with rapid delivery, ultra-micro DCBS can help customers build AI infrastructure faster, better, greener, and at a lower cost.
Since optimized hardware alone cannot operate a high-performing AI data center, managing thousands of liquid-cooled GPU racks requires intelligent real-time software orchestration. SuperCloud Composer (SCM) provides seamless rack-level management and full-stack hardware-facility integration; Supermicro Datacenter Manager (SDM) provides complete power and thermal management from a single node to the entire building infrastructure to achieve end-to-end hardware lifecycle management; our Supermicro Operations Manager (SOM) suite includes Director and Developer The Console module makes it easy to implement bare metal, models, and tokens as a service for large language models and agent-based AI workloads.
Attaching SCM, SDM, and SOM directly to the DCBS product portfolio eliminates software fragmentation for customers while increasing the bundling rate of high-value software. Additionally, SCM's enhanced automated fault reporting service now has AI-driven root cause diagnosis capabilities, significantly improving data center availability and giving customers complete operational confidence.
Equally critical is our newly expanded global data center services organization. Our on-site deployment team is supported by R&D engineers and well-trained post-delivery service personnel, and directly collaborates with customer teams through tiered services to efficiently maximize cluster availability. Using automated telemetry, our service team continuously monitors cooling, power supply and node health in real time to resolve potential issues before they cause downtime; dedicated field engineers are responsible for deployment verification and immediate hardware maintenance. This proactive, high-reach service model builds deep customer trust, brings in high-value recurring service revenue, and fulfills our highest promise: to bring customers a completely worry-free experience with high-quality overall infrastructure solutions.
To efficiently achieve this unparalleled breadth of solutions, our global manufacturing layout and scale are rapidly climbing.
We are actively expanding physical production capacity, and major projects in Silicon Valley and Taiwan are underway, adding millions of square feet of fully automated production space and active service capabilities. Additionally, we are speeding up containerized data center production in Asia and the US. In total, our global manufacturing and service network enables us to produce more than 8,000 of the world's most powerful racks per month, including more than 4,000 liquid-cooled (DLC) racks by December of this year.
At the technical level, Ultra Micro continues to lead the market on major chip platforms:
We were one of the first manufacturers to ship Nvidia Vera Rubin NVL-72 racks equipped with DCBBS and DLC-2 liquid cooling stacks. We've also recently launched Nvidia Rubin and Vera CPU racks, and an optimized DSX-POD solution is coming soon. On the AMD side, demand for the MI350 and MI355X platforms is strong, and we are moving towards the overall Helios and MI-450 solutions. Additionally, we are advancing the Intel Xeon 6+ platform and a new ARM AGI-based CPU-based solution optimized for the proxy AI market. At the same time, we continue to expand our comprehensive edge and IoT product portfolio and continue to expand our storage product line to bring real-time, complete inference capabilities directly to the intelligent edge.
These new products have accelerated our strategic focus on enterprise customers — as the enterprise expands to large-scale agent models, enterprise data governance, and high-throughput inference workloads. We recently announced a number of major strategic partnerships to combine ultra-micro DCBS with leading global technology companies, including Cisco, NetApp, Nutanix, etc. More first-tier partners will be announced soon, and we are deeply embedding the ultra-micro DCBS layout into the world's most trusted companies.
I know some investors and market experts have recently raised questions about the efficiency and return on investment of AI capital expenditure.
Let me be clear: Ultrafine has a unique advantage in this environment. For more than 30 years, we have been focusing on the server and storage industry in the US. We have spent more time than almost all of our competitors, and have established a mature and complete local manufacturing and engineering foundation, making it difficult for our competitors to match.
This deep American roots, combined with our unique DCBS overall hardware, software, and services solution, directly solves customers' ROI challenges by optimizing delivery time (TTD), time to launch (TTO), system availability, and energy efficiency. Whether customers are building large-scale first-line AI factories or deploying enterprise agent-based AI workloads, Ultramicros can turn their capital investment into revenue faster and more cost-effectively than anyone else we know.
We promote strict internal discipline through continuous operational optimization, factory automation, and skill improvement through technology and process improvements and training. Over the past two quarters, we've successfully reduced our inventory turnaround time from around 100 days to 74 days. Looking ahead, we aim to further reduce inventory and production cycles to 50 days over the next few quarters. This ongoing efficiency optimization will continue to reinforce ultra-small cash flow and working capital cycles, and drive higher bottom-line profitability. No matter how competitive the market is, more customers will prefer ultra-micro DCBBS — including data center end-to-end hardware, management software, and end-to-end data center services.
Here's the outlook for fiscal year 2027:
While overall demand and growth are strong on all fronts, we must also recognize that shortages of critical components are still impacting delivery times across the IT industry, even GPUs. New customer projects and strategic ecological partnerships are climbing strongly, but we want to be cautious about the availability and fluctuating prices of some key components. Based on what we have seen so far, we expect revenue targets for FY2027 to $72-$82 billion, which is about 12% higher than the target set a quarter ago.
Before handing over the phone call, I'd like to highlight a few important personnel changes:
First, we're excited to welcome senior industry technology leader Shesha Krishnapura to the board, whose deep experience in data center operations will support our scale-up. Second, I would like to sincerely thank David Wiegand for his leadership as Chief Financial Officer for the past six years, a period of rapid growth. We look forward to his continued support and contributions in optimizing operations to a larger scale. Finally, I am pleased to introduce Mark Long as our new Chief Financial Officer. His decades of leadership experience in technology and finance will be a strong support for our next phase of growth.
Next, I'll hand over the phone to Mark.
MARK LONG (CFO): Thank you Charles and good afternoon everyone.
This is my first time attending a conference call, so allow me to speak for two minutes before going into numbers.
I've spent most of my career in finance positions in technology infrastructure, including being the chief financial officer of a publicly traded company in the industry. What drew me to the ultra-small part: the hardest part was already built — deep engineering advantages, real leadership in liquid cooling and rack-level integration, and a critical card position in the largest infrastructure construction we've seen in our careers.
What the company needs now is financial capacity that matches its engineering and operational capabilities. I'm excited to join Ultra Micro at this critical point and look forward to helping Charles and his team continue to deliver an expanding portfolio of innovative solutions to an expanding global customer base, including some of the largest and most demanding customers who are shaping the future of the AI era. To do this, ongoing financial discipline, transparency, deeper communication with stakeholders, working capital management, and capital structure optimization will be the focus of my work.
Now let's talk about this quarter.
Total revenue was $19.9 billion, up 298% year over year and 79% month on month. This is higher than our guideline range of 14.5 billion to 15.5 billion US dollars, and higher than the estimated range of 18.5 billion to 19.5 billion US dollars given on October 20. As Charles said, approximately $2.3 billion of this corresponds to deployments that were scheduled to be delivered earlier and were delayed due to progress in customer and data center readiness. Excluding this portion, the underlying revenue was approximately US$17.6 billion.
AI solutions contributed approximately 71% of revenue, compared to 59% in the previous quarter. Based on current order reserves, we expect AI-related solutions to account for approximately 70% — 80% of revenue in FY2027.
Enterprise and channel revenue was $4.9 billion, accounting for 25% of revenue, up 229% year over year and down 12% month over month. OEM equipment and large data center revenue was $14.7 billion, accounting for 74%, up 321% year over year and 167% month-on-month, driven by a steep rise in mass production of multiple large-scale AI data center design orders. Although 5G, telecom, and edge businesses are still relatively small, they have grown significantly year over year and month over month.
By region, the US accounts for 80% of revenue, Europe 11%, Asia 6%, and the rest of the world 3%. The US grew 771% year over year and 103% month over month. Although the US remains the largest market for large-scale AI data centers, we are also seeing momentum in other regions — large customers are expanding globally, and governments and companies are planning and investing in large-scale AI data centers in Europe, Asia, and the Middle East.
A large data center customer accounted for 44% of revenue this quarter, compared to 28% for the full year of FY2026. As you'll also see in 10-Q, the two customers accounted for 49% and 17% of accounts receivable at the end of the quarter, respectively. I'd like to provide more context on this.
This level of concentration is an inevitable result of the current scale of AI infrastructure construction. The size of a single deployment today could exceed our single quarter revenue a year ago. We expect the concentration level to remain high during the construction of these projects and decline as the scale of corporate business and sovereign business expands.
Just as important as concentration is the quality of risk exposure, and I'll give that. Deferred revenue increased by approximately $3.1 billion this quarter, the vast majority of which was paid by customers before delivery. The number of accounts receivable turnaround days improved from 59 days to 36 days. We have recorded no significant credit losses this quarter. In large deployments, we often require deposit payments and are usually invoiced according to delivery milestones; we assess counterparty credit before committing to manufacture capacity and critical components; and we also have $1.8 billion in accounts receivable purchases, which provides flexibility in terms of both cash timing and credit exposure.
In this market, we have rejected businesses where terms make us uncomfortable; in the current environment, I expect we will continue to make this choice.
The non-GAAP gross margin was 12.8%, 200 to 240 basis points higher than our 10.4% — 10.8% guidance range, and 330 basis points higher than the same period last year. The GAAP gross margin was 12.7%.
It's also 480 basis points lower than our Q4 report of 17.6%, and I'd like to break it down because a single factor explanation would be misleading.
The key factor affecting gross margin was the increase in AI revenue share in the first quarter from 59% in the fourth quarter to 71%. The customer and product structure of our business in the first quarter was biased towards large-scale AI deployments, and the gross margin of this type of business was lower than that of more diversified enterprise and channel businesses. The benefits brought about by tariff relief and lower inventory provisions this quarter were largely offset by some of the low gross margin deferred revenue confirmed during the quarter.
I want to be precise about one thing because I know how structured data might be interpreted. The month-on-month decline is not evidence that AI platform revenue lacks profit margins; it reflects which customers are picking up goods in which quarter. The clearest way to observe underlying trends is to look at the two most recent quarters together, which can eliminate the time-point impact of delays. Fourth-quarter and first-quarter combined, revenue was approximately $31.1 billion, and non-GAAP gross margin was approximately 14.5%. In the two quarters of the same period last year, revenue was approximately US$10.8 billion and gross margin was approximately 9.5%. According to this normalization approach, revenue nearly tripled, and gross margin increased by about 500 basis points.
We continue to demonstrate significant operating leverage. Revenue increased 298% year over year rather than GAAP operating expenses up 89%.
GAAP operating expenses were $477 million, up 67% year over year and 5% month over month. Non-GAAP operating expenses were $384 million, up 89% year over year and 8% month over month. The month-on-month growth reflects rising personnel costs and our investment in R&D of multiple new platforms for AI data centers and enterprise applications. The non-GAAP operating margin was 10.9%, compared to 14.3% in the fourth quarter.
Here, too, the two-quarter perspective is more useful. In the fourth quarter and the first quarter combined, non-GAAP operating expenses were approximately $741 million, accounting for about 2.4% of revenue, and the non-GAAP operating margin was approximately 12.1%, compared to 5.4% in the same two quarters last year. We intend to continue investing in R&D, field service, and global presence, and expect operating expenses to continue to grow significantly lower than revenue growth.
Other income and expenses were net expenses of $34 million, including $49 million in interest income and other income, offset by $83 million in interest expenses from our convertible bonds and revolving credit arrangements.
The tax preparation was $435 million in GAAP caliber and $455 million in non-GAAP caliber, and the actual tax rate for both was 21.4%.
I'd like to add a few points from the last call about earnings per share, because our capital structure now has an additional variable that I don't think is fully understood. We use two methods to calculate earnings per share, which reflects our mandatory convertible preferred shares issued in the fourth quarter. The instrument pays a quarterly fixed dividend of approximately $75 million and will be converted to common stock in June 2029. During this period, dividends and the distribution of earnings to the participating securities will reduce the profits available for distribution to common shareholders — a total reduction of approximately $312 million this quarter.
GAAP diluted earnings per share were $1.75, leading to $0.89—$0.98. Non-GAAP diluted earnings per share were $1.81; guidance is $1.01—$1.10. GAAP basic earnings per share were $1.92.
The number of GAAP diluted shares rose from 705 million to 749 million shares, and the number of non-GAAP diluted shares rose from 721 million to 764 million shares.
Now let's talk about working capital, which I think helps explain the key elements of how we run our business.
Our cash conversion cycle improved from 149 days to 97 days, an improvement of 52 days in a single quarter. Based on the revenue scale for the first quarter, this difference is equivalent to approximately $10 billion in working capital that the company would have needed additional financing.
The cash flow statement reflects this. Of our $19.9 billion revenue, we use $434 million in cash in operating activities. Throughout fiscal year 2026, we used $6.8 billion in cash on $39.1 billion in revenue. Capital expenditure for the quarter was $47 million, so free cash flow was negative $481 million, compared to negative $7 billion for fiscal year 2026.
I'm going to break down these 52 days, including both improvements and parts that still have room for improvement. Inventory turnover days improved by 45 to 74 days. Accounts receivable turnaround days improved by 23 to 36 days, thanks to faster payouts and customer prepayments. The payables turnaround days were bad for us, falling from 29 days to 13 days, consuming around $3 billion in cash, reflecting when payments to suppliers were made in a supply-constrained environment.
The single biggest driver is customer upfront payments. Deferred revenue increased by approximately $3.1 billion this quarter. Accounts receivable increased by $3.6 billion and inventory increased by $2.5 billion to support our forward revenue, and prepayments funded much of it.
We expect this quarter's cash conversion cycle to be in the range of plus or minus 10% compared to the first quarter, which allows us to properly manage working capital while seizing attractive market opportunities.
We held $6.9 billion in cash and cash equivalents at the end of the quarter. Bank loans and convertible bonds totaled $8.6 billion, and net debt was $1.7 billion, compared to $1.2 billion at the end of the fourth quarter. Inventories of $151 billion, up from the previous $12.9 billion, were established to support anticipated revenue for the coming quarters. We met all contract terms at the end of the quarter.
Let's talk more broadly about capital structure: Our starting point is solid. The $5.6 billion we raised in June was to support the working capital behind the orders we are currently delivering.
As we identify high-quality growth opportunities, we will optimize our capital structure to support these opportunities and do so in a way that strengthens our balance sheet and doesn't incur unnecessary costs. This means matching the term and form of financing to the working capital and assets it supports, and being cautious about when and whether to enter the capital market. I will be transparent with you every quarter about the dilution, expiration structure, and available credits.
We had a record reserve of orders at the end of the quarter, providing greater visibility for the quarter and the rest of the fiscal year.
Steering guide.
For the second quarter of fiscal year 2027, we expect net sales of $19.5 billion to $21.5 billion and gross profit margin of 12.0% — 12.5%. We expect GAAP diluted earnings of $1.59—$1.83 and non-GAAP diluted earnings of $1.66—$1.90 per share.
The guidelines assume GAAP operating expenses of approximately $455 million, including approximately $102 million in equity incentive expenses, which do not count towards non-GAAP. We expect other income and expenses (including interest expenses) to be net expenses of approximately $58 million, GAAP and non-GAAP tax rates of approximately 21%, GAAP diluted shares of approximately 753 million shares, 769 million non-GAAP diluted shares, dividends of approximately $75 million to $80 million in preferred shares, and capital expenses of $70-80 million.
For the full fiscal year, we raised our revenue outlook from $65-$72 billion to $72-$82 billion. We expect gross margin of 11.8% — 12.2% for the full year.
Let me relate this to the quarter just announced. The first quarter included approximately $2.3 billion in catch-up revenue from previous periods, resulting in underlying revenue of approximately $17.6 billion. The full year range does not assume that this part of catch-up revenue is repeated. It assumes continued month-on-month growth in underlying demand and reflects a significant improvement in our visibility into customer deployment timelines compared to 90 days ago. Our business continues to grow, and the momentum for fiscal year 2027 is building, despite supply constraints and data center readiness; we focus on disciplined execution, creditworthy counterparties, and expanding higher-margin revenue streams.
Regarding gross profit margin, I want to be honest about the information implied by the year-round range. The first quarter was 12.8%, and the second-quarter guideline was 12.0% — 12.5%. The median value for the full year range implied that the gross margin for the second half of the year was lower than in the first half of the year. This is a structural effect: with large-scale deployments and shipments in order reserves, we expect AI solutions to move to the upper edge of the 70% to 80% range, and the gross margin of this type of business is lower than that of vertical enterprise and channel businesses. Within each category, we expect gross margins to maintain or improve.
What determines this range is not demand, nor our ability to build, but rather the speed at which customer facilities are ready to receive the products we have manufactured, and the availability of key components. Currently, the hard constraints for AI construction across the industry are land, power, and data center housings. We have better visibility into this than in the past and will continue to update it throughout the year.
Let me go back to my original judgment.
This was a strong quarter of growth, and the quality of customers was high. Revenue nearly tripled year over year, gross margin was well above our guidelines, and customer advances provided most of the working capital we needed to deliver. Behind these results is a broad and innovative product portfolio—from DCBS rack-scale and liquid cooling systems to enterprise, storage, edge, and software—that enables us to tackle the most demanding workloads and the widest range of use cases in AI construction.
My future focus is what I described at the beginning: financial discipline, making growth profitable; transparency, so that everyone can see how this business works from season to season, including the time-in-time effects affecting our numbers; deeper communication with all stakeholders; continuing to manage working capital based on improvements in the cash conversion cycle this season; and optimizing the capital structure to support high-quality growth on a stable basis.
I look forward to seeing many of you in the coming months.
Michael, we can go ahead and ask questions.
Q & A session
Moderator: Thank you. Now let's start the question and answer session. The first question comes from Katherine Murphy from Goldman Sachs Research, please ask.
Katherine Murphy (Goldman Sachs Research): Thank you for your questions and I'm happy to talk to you. First, I would like to ask about the revenue outlook after the increase. I understand that land, electricity, and data center housings are currently hard constraints, but I would like to know what percentage of the project in your annual outlook and project reserves can you clearly see who it has bought and how much has been promised? Then I had a question.
Liang Jianhou: Yes, as you know, there is a shortage of parts such as electricity and even GPUs. But because our solutions are superior to the industry, especially with DCBBS — DCBBS not only provides data center hardware modules, but also software and services to help customers build data centers faster and maintain better availability of computing power. Indeed, we continue to get more customers and more shares. We are quite optimistic about the continued growth in market share.
Katherine Murphy (Goldman Sachs Research): As a question, can you talk more about this quarter's customer advance payments and how should we view it as it continues to support your working capital efficiency? Is this a change we can expect associated with DCBS and the market share growth strategy you're talking about?
Liang Jianhou: Yes, as you know, many companies have tight cash flow. But Ultra Micro, because we have a better overall solution — also DCBBS — has attracted a large number of customers. Basically we're seeing good signs. Most of our customers are able to pay on time, and we feel very secure in our continued business growth. Mark, you can add a few things.
MARK LONG: Yes. I think the advantage, as Charles pointed out, is essentially the quality of the customer base. Very large, creditworthy customers, paid very fast and provided us with a very strong cash flow. And the broader customer base is providing more upfront payments, and this discipline allows us to fund more working capital. I think you should expect this to continue. So we feel like we're in a very good position. Since the company raised in the summer and since the cash conversion cycle improved, the company has done a lot, and the efficiency of working capital has improved dramatically.
Liang Jianhou: Yes. As we just shared, we've improved our cash flow business performance in two areas. The first is to improve the inventory and production turnaround time from about 100 days to the current 70 days. We believe we can continue to improve production and inventory turnaround time. Second, as Mark said, the quality of our customers is improving. These customers pay on time and even pay back very quickly. So we're pretty solid in this regard.
Moderator: The next question is from Ananda Baruah of Loop Capital, please ask.
Ananda Baruah (Loop Capital): Thank you for taking the questions. Congratulations on the progress made, Mark, and welcome to join. I have two questions. The first one concerns gross profit margin and how to understand its evolution for the rest of the fiscal year. If I misunderstood, please correct: to hit the median value of your guidelines, although it declined year over year in the second half of the year, it remained roughly flat at around 12% for the next three quarters. Please correct me if this isn't right. The real question is, you mentioned that the increase in the share of the AI structure is one of the factors. Can you talk about the variables in gross margin within the AI customer base? I remember Charles talking about AI gross margin improvements in the past, and I wanted to know what the internal dynamics of AI gross margin were for the rest of the fiscal year. Afterwards I had a brief question. Thank you.
Liang Jianhou: Yes, as I just shared, in the past three or four quarters, our production time, production quality, and yield have improved dramatically, which has made our turnaround time faster and our customers are more satisfied because of better TTM and better TTO to help customers go online faster. So the customer approved it. Coupled with our DCBS software and services, it further helps customers. As well as better customers, we continue to get more customers, especially enterprise agent-based AI customers. So these are all having a positive impact on our business.
MARK LONG: Yes, I'd like to add one more point to what Charles said. As far as the full-year guidance is concerned, what you can see is that we have good visibility, our business is now stabilizing, and we have established a foundation in this range of about 12% gross margin. As we continue to diversify our customer base and product portfolio, there is room for further expansion of gross margin. So, in terms of accurate modeling, the judgment in your original question is relatively close. But for us, visibility is very strong.
As Charles pointed out, gross margins have improved across all of our customer categories. That's what really matters. This is the result of what Charles said about excellent engineering, excellent product portfolio, and team execution. So we feel like we're in a very good position on this scale, and we see huge opportunities ahead.
Jianhu Liang: Yes, just like our DCBS, we not only provide hardware components for modular data center solutions, but also software to help customers maintain availability rates. So these have almost become one of our business models. Customers love our products, and more and more customers love our solutions.
Ananda Baruah (Loop Capital): That's a very valuable background. The short question is, considering all the structural side effects of this fiscal year, what are reasonable expectations for CPU business opportunities? That's all I have, thank you.
Liang Jianhou: Yes, as you know, proxy AI is becoming very popular. As a result, we have gained a large number of customers, including many corporate clients. They're buying our CPU-based solutions, including Intel CPU-based ones, AMD-based ones, and Nvidia Vera, right? Our Nvidia Vera solution is ready, including the ARM solution. So we're seeing significantly stronger CPU demand now, and the same is true looking forward to the future.
Moderator: The next question comes from Joseph Cardoso of J.P. Morgan Chase Research. Please ask.
Joseph Cardoso (J.P. Morgan Chase Research): Hello, thank you for asking, Mark, and I look forward to working with you. My first question, from your presentation, doesn't seem to include more catch-up or deferred delivery in the guidelines. But I'd like to confirm first, are there any deferred projects that haven't been delivered yet but are not included in the outlook? The second part of the problem is that when we remove this part, you still implied that growth in the second half of the year was slower than in the first half. Can you please sort out the developments affecting this and whether this is related to the likely decline in the share of largest customers after entering the second half of the year?
MARK LONG: Yes. I think the key point about deferred revenue, looking back at history, is that there is a set of delayed deliveries and deployments that were carried over from previous quarters. In a typical quarter, you'll have revenue recognition based partly on delivery, and there may also be one or two deployment delays. But the amount of deferred revenue at the time was higher than typical. When it was released, it sparked a series of results.
Revenue has risen, and as we mentioned, this has diluted gross profit margins overall. Looking ahead, however, we don't expect the same type and scale of delayed deployment to be reflected in deferred revenue. The majority of deferred revenue is either customer upfront payments or service and software revenue that we usually defer confirming for multiple periods.
Liang Jianhou: Yes. The good part is that since we launched DCBBS, some customers and even people within us have been in the process of adapting. So they have some controversy about when to confirm revenue. But now, our DCBS is a much more mature business. Not only are our insiders familiar with DCBS now, but customers also understand what services they buy and what they don't buy, so there should be no more coverage issues in this area. As it became more clear and mature, so did our deferred revenue.
MARK LONG: Yes. Regarding the second question, that the guidelines suggest a slowdown in growth in the second half of the year, all you actually heard was our set of expected ranges based on supply constraints and an attempt to determine where the data center readiness for large-scale deployments will take us. But as we mentioned in our presentation, demand is still extremely strong, particularly interest in the DCBS product portfolio.
Joseph Cardoso (J.P. Morgan Chase Research): Understood, very clear. My second question is related to this and is probably a two-part one. You talked about the tight supply of the entire ecosystem entering the second half of the year. Can you talk about the tightest constraints in the outlook for the second quarter and the second half of the year? Is it mainly the GPU, or is it something else in terms of components? The second problem is that in this context of shortages, price is clearly a strong transfer factor. Can you help us understand how much of the growth in the 2027 outlook comes from shipments and how much from cost transfers? Thank you.
Liang Jianhou: Yes, great question. We are giving conservative annual revenue figures of $72-$82 billion because we are concerned about shortages of some key components and the macroeconomy. Therefore, if the shortage of parts does not become a major problem, and if the macroeconomy is healthy, then the second half of the year will undoubtedly be significantly larger than the first half. Of course we hope so, which is why we are very optimistic that the second half of the year will be significantly better than the first half. If that were the case, $72-$82 billion would be very, very conservative.
MARK LONG: Yes. I think your view on ASP growth is indeed driven by the cost of key components, sometimes rising on a quarterly basis. Our outlook for the second half of the year does not include significant growth in ASP. So we're taking a relatively conservative approach in this regard; we're really looking at the details of our customers' deployments, what they need, what product combinations, what our production capacity is, and how we manage these. As Charles points out, our demand far exceeds the availability of critical components.
Moderator: The next question comes from Ruplu Bhattacharya of Bank of America Research, please ask.
Ruplu Bhattacharya (Bank of America Research): Hello, thank you for taking the question. Mark, congratulations on your new appointment and look forward to working with you. First question I'd like to ask you. Looking ahead to the next 12 months, do you think Ultraman will have enough cash and cash flow to fund the investments it needs to support the strong growth it sees? What circumstances will motivate you to further finance, and do you have a preference between debt and equity? I have a question.
Liang Jianhou: Great question. As you know, at the end of last quarter, we had $6 billion in cash.
MARK LONG: Yes.
Liang Jianhou: We've never had this much cash in our accounts. Indeed, we're not worried about the current cash flow. Even if strong growth continues in the second half of the year, if the macroeconomy is healthy, our cash flow position is stronger than ever.
MARK LONG: Right. As Charles pointed out, starting with $69 billion in cash put us in a very good position. As we mentioned, our working capital efficiency has improved, and our ability to fund working capital needs by focusing on high-quality customers and partial customer prepayments has also been enhanced. This puts us in a good position to optimize our capital structure. If a very attractive opportunity for incremental funding arises, we can make trade-offs. We have a working capital of $1.8 billion and opportunities to enter other markets, including the credit market and equity market, as you pointed out.
Currently, we don't have any plans to consider equity financing. As for the debt market, I think we'll look at the conditions available at the time and what kind of arrangement is most conducive to the company's continued success, and provide us with an appropriate leverage structure.
Moderator: Due to time constraints, we'll just accept the last question. This question comes from George Notter of Wolfe Research, please ask.
Brendan Rogers (Wolfe Research, on behalf of George Notter): Hi everyone, I'm Brendan Rogers, asking questions on behalf of George Notter. What I'm asking is, how has your communication with customers changed over the past 90 days to gain better visibility? Is this a feature of the current market, where customers are extending the forecasting window they give you, or is it more a shift in the way you manage your business?
Liang Jianhou: Yes. As we mentioned, we have strong undelivered orders, so we are in a position to choose healthy customers and order quality. Customer quality is very important to us. So in a sense, we have a chance to choose.
MARK LONG: Yes. As Charles points out, we've had better visibility from many of our customers, some very large, some medium, and we've gained visibility in every region.
Liang Jianhou: And corporate customers.
MARK LONG: Yes. The growth of our business is another factor that will bring greater visibility and stability to our overall revenue stream.
Moderator: This concludes the Q&A session. This concludes today's call. Thanks to everyone who participated, you can now disconnect.