Super Micro Computer, Inc. (SMCI) Q4 2026 Earnings Report, Transcript and Summary

Thank you for standing by. My name is Jen, and I will be your conference operator today. At this time, I would like to welcome everyone to the Super Micro Computer, Inc. Q4 FY '26 Business Update Call. With us today are Charles Liang, Founder, President and Chief Executive Officer; David Weigand, CFO; and Michael Staiger, Senior Vice President of Corporate Development. [Operator Instructions] I will now hand the conference over to Michael Staiger. Please go ahead.

Thank you, Jen. Good afternoon, and thank you for attending Super Micro's call to discuss financial results for the fourth quarter of fiscal 2026, which ended June 30, 2026. With me today, as you know, are Charles Liang, Founder, Chairman and Chief Executive Officer; and David Weigand, Chief Financial Officer. By now, you should have received a copy of the press release from the company that was distributed at the close of regular trading and is available on the company's website. As a reminder, during today's call, the company will refer to a presentation that is available to participants in the Investor Relations section of the company's website under the Events & Presentations tab. We've also published management scripted commentary on our website. Please note that some of the information you'll hear during our discussion today will consist of forward-looking statements, including, without limitation, those regarding revenue, gross margin, operating expenses, other income and expenses, taxes, capital allocation and future business outlook including guidance for the first quarter of fiscal 2027 and the full fiscal year '27. These statements and other comments are based on management's current expectations and assumptions and involve material risks and uncertainties that could cause actual results or events to materially differ from those anticipated, and you should not place undue reliance on forward-looking statements. You can learn more about these risks and uncertainties in the press release we issued earlier today, our most recent 10-K filing for fiscal '25 and other SEC filings. All these documents are available on the IR page of Super Micro's website. We assume no obligation to update any forward-looking statements. Most of today's presentation will refer to non-GAAP financial results and business outlook. For any explanation of our non-GAAP financial measures, please refer to the accompanying presentation or to our press release published earlier today. The non-GAAP measures are presented as we believe that they provide investors with the means of evaluating and understanding how the company's management evaluates the company's operating performance. These non-GAAP measures should not be considered in isolation from, as substitutes for or superior to financial measures prepared in accordance with U.S. GAAP. In addition, a reconciliation of GAAP to non-GAAP results is contained in today's press release and in the supplemental information attached to today's presentation. At the end of today's prepared remarks, we'll have a Q&A session for sell-side analysts. Our fiscal '26 -- '27, excuse me, quiet period begins at the close of business, Friday, September 11th, 2026. I will now turn the call over to Charles.

Thank you, Michael, and thank you all for joining today's call. Fiscal year 2026 was a historic milestone for Super Micro as we nearly doubled our revenue year-over-year, growing from $22 billion last year to $39 billion fiscal year '26. The world is being transformed by AI, and Super Micro is transforming as well from a USA-based server manufacturer into a leading AI/IT data center total solution company. We design and manufacture our total data center building block solutions, DCBBS, in the USA, with main facilities in USA, Taiwan, Malaysia and the Netherlands. The demand for our AI/IT solutions is even stronger than ever before as we are transforming into a total DCBBS company: A one-stop shop company for customers who want to build their data center or AI factory quicker and better. In our preannouncement, we disclosed over $60 billion in new orders, driving our order book and backlog to new record levels as we enter fiscal year 2027. While Q4 revenue came in at $11.1 billion due to some short-term customer delays in power shortage, cooling, and networking, we know this is purely a timing story. The good news is that now, our customers can easily leverage our unique DCBBS total solution advantages and upcoming new technology and product lines to accelerate their time-to-deployment, we call TTD; and time-to-online, we call TTO, ensuring strong future growth and long-term value for Super Micro for many years to come. Most importantly, our focus on profitability is yielding clear results. For the fourth quarter, I am happy to report a non-GAAP gross margin of 17.6% and $1.70 in non-GAAP diluted earnings per share. This margin expansion mainly came from our strategic focus on balancing customer mix and product mix while having few one-time positive contributions for the quarter. Since early 2026, we added dedicated departments and resources to focus on growing enterprise customer base and have expanded our enterprise CPU-based server, storage, and IoT product lines. Our quick growing inferencing and agentic AI-centric products are also driving healthier profit margins for the company going forward. Another key to this margin expansion is our DCBBS, which delivers total-solution value by seamlessly integrating GPU and CPU servers, enterprise storage, direct liquid cooling solutions, CDU, chilled door, water tower, high-speed data switch, and networking, data center management software, and full life-cycle services. This turnkey ecosystem enables customers to build and scale AI data centers in quarters rather than years, dramatically reducing TCO and accelerating time-to-online and time-to-revenue for customers. We are further elevating this value proposition with our new proactive service model, where our data center management software and field teams will automatically alert and be ready immediately to fix or maintain the failure unit, preventing reduction of computing power at customer data center. As the new software with powerful management features and automatic services attach to our hardware builds, they deepen customer trust and drive long-term value. Our DCBBS is getting very powerful and it will soon contribute significant net income to our business. By early next quarter, more of those software features and service products will be online. On the operational side, we are complementing this high-value strategy by driving higher manufacturing yields through factory automation, design optimization, and our highly versatile building block architecture. At the same time, we remain very focused on logistics and inventory management, significantly reducing inventory reserves and expedite charges. Together, these operational disciplines will help moderate quarter-to-quarter margin fluctuations driven by uneven customer and product mix, supporting our goal of consistent, growing gross margins. Turning to our key product roadmap. Our system building blocks allow us to quickly optimize every major silicon platform. Through our long-term NVIDIA partnership, we are shipping volume SKUs across the GB300 NVL72, HGX B300, B200 NVL4, and RTX 6000 Pro product lines, while preparing first-to-market Vera Rubin VRNVL72, Rubin HGX, and Vera C1 and other high density Vera systems. With AMD, we launched completely new Helios product line and MI450 Total Solution alongside strong EPYC CPU, MI350 and MI355X momentum. Working with Intel, we brought Panther Lake Edge AI systems to market and shipping Xeon 6+ platforms in volume. We also dedicated on developing product for the strong demand of Arm AGI processor-based, code-named Phoenix architectures optimized for high performance-per-watt inferencing workloads, demonstrating our silicon partners' deep confidence in our engineering excellence. To support the massive demand, we continue to expand our physical footprint. In Silicon Valley, we recently announced our new 32-acre DCBBS campus featuring advanced optical photonics networking labs and data center scale manufacturing, which brings our USA footprint to nearly 4 million square feet. Globally, our facilities in Taiwan, Malaysia, and the Netherlands are also ramping strongly to meet demand, putting our total manufacturing capability on track to exceed 6,000 racks per month, including more than 3,000 direct liquid-cooled racks per month. Especially, most of our DLC rack production lines support the most dense, latest 250kW rack platforms. Before I close, a quick update on our capital structure: following our $5.6 billion financing in June, our balance sheet fully supports our component supply and business needs. Thanks to our strong cash position and more favorable customer and product mix, we currently have no plans to utilize our ATM program which we initiated a few months ago. At the same time, we remain focused on building financial efficiency. Building on all of these operational and product advancements, I want to emphasize that our growth momentum is accelerating where it matters most. By expanding hundreds of new enterprise customers and other customers and leading the transition into agentic and specialized AI workloads, Super Micro has become a foundational architect of today's AI backbone. Our DCBBS total solution-spanning CPU and GPU computes, storage, 800G and 1.6T high-speed switch, upcoming optical networking, and our management software suite including SCM, Super Micro Cloud Composer; SDM, Super Micro Data Center Manager, and SOM, Super Micro Orchestration Manager delivers the complete, one-stop shop experience that modern enterprises, Neocloud and any other data center customer's needs. Looking to fiscal 2027, our momentum gives us strong confidence to target our revenue in the range of $65 billion to $72 billion as we are in the process of a historic infrastructure buildout. We are balancing top-line expansion with bottom-line profitability by focusing on growing enterprise customer base, customer mix, DCBBS solutions, and operational discipline. We are shaping the future of AI technology while delivering true technology value to our customers. I am very confident that fiscal 2027 will be our strong and fast growth year again. Thank you, and I will now turn the call to David.

Thank you, Charles. We are pleased to report record fiscal year '26 revenue of $39.1 billion, up 78% over fiscal year '25 revenues and (sic) [ of ] $22 billion and record non-GAAP fully diluted EPS of $3.63, up 76% over fiscal year '25 EPS of $2.06. Our fiscal year '26 ending backlog was at record levels with over $60 billion in new orders received during Q4 fiscal year '26, which we expect to fulfill over the coming quarters. Non-GAAP gross margins for fiscal year '26 were 10.9% versus 11.2% in fiscal year '25. Our fiscal year '26 non-GAAP operating margins expanded to 8.1% from 7.1% in fiscal year '25. Our customer base is diversifying, and we had 9 customers in fiscal year '26 with revenues greater than $1 billion each versus 4 such customers in fiscal year '25. Turning to fiscal Q4 fiscal year '26 results, we achieved revenue of $11.1 billion, up 93% year over year and up 9% quarter over quarter. Revenue was near the low end of our guidance range of $11 billion to $12.5 billion due to delays in customer readiness, and we anticipate this revenue to be recognized in subsequent quarters. These AI solutions -- our AI solutions contributed approximately 60% of total revenue in Q4 versus over 80% in Q3 due to the timing of some large AI project ramps. Based on our backlog, we believe greater than 80% of revenues will be AI-related solutions going forward. During Q4, enterprise and channel revenue was $5.6 billion, representing 50% of total revenue, compared with 28% in the prior quarter. Revenue in this segment increased 172% year over year and 98% quarter over quarter. During Q4, we saw a pickup in demand from enterprise and channel customers which were upgrading their compute, storage and network infrastructure with more efficient CPU platforms. OEM appliance and large data center revenue was $5.5 billion, also representing 50% of total revenue, compared with 72% in the prior quarter. Revenue in this segment increased 50% year over year and decreased 26% quarter over quarter. For fiscal year '26, enterprise and channel revenue grew 39% and represented 31% of total revenue. The OEM appliance and large data center revenue grew 104% and represented 69% of total revenue. For fiscal year '26, we had one large data center/CSP customer, which represented 28% of revenue. By geography, the U.S. represented 71% of Q4 revenue, Asia represented 11%, Europe represented 8%, and the rest of the world represented 10%. On a year-over-year basis, revenue in the U.S. grew 259%, Asia decreased 50%, Europe increased 4%, and the rest of the world increased 296%. On a quarter over quarter basis, revenue in the U.S. grew 12%, Asia decreased 13%, Europe increased 25%, and the rest of the world increased 1%. Q4 non-GAAP gross margin was 17.6% versus our guidance of 8.2% to 8.4%. This was up from 10.1% in Q3. Gross margins improved by 750 basis points sequentially due to a better-than-anticipated customer and product mix, including the deferral of several contracts from Q4 fiscal year '26 to Q1 fiscal year '27, and perhaps the subsequent quarter. This favorable mix contributed approximately 75% of the gross margin improvement. Lower tariff costs and lower inventory reserves drove the remaining 25% of the gross margin improvement. Q4 GAAP operating expenses were $455 million, up 44% year over year and 16% quarter over quarter on a non-GAAP basis. On a non-GAAP basis, operating expenses were $357 million, which was up 49% year over year and 28% quarter over quarter. The sequential increases in both GAAP and non-GAAP operating expenses primarily reflected higher headcount-related expenses and sales and marketing expenses. Non-GAAP operating margin was 14.3% in Q4, compared with 7.2% in Q3. Other income and expense for Q4 was a net expense of $19 million, consisting of $61 million in interest and other income, offset by $80 million in interest expense related to our convertible notes and revolving credit facilities. The Q4 tax provision was $290 million on a GAAP basis and $316 million on a non-GAAP basis. The Q4 GAAP tax rate was 19.7%, while the non-GAAP tax rate was 20.1%. For fiscal year '26, the GAAP tax rate was 19.9%, compared with 12.9% in fiscal year '25. The non-GAAP tax rate was 20.4%, compared with 15.4% in fiscal year '25. Q4 GAAP diluted earnings per share was $1.62, compared with our guidance range of $0.53 to $0.67. Non-GAAP diluted earnings per share was $1.70, compared with our guidance range of $0.65 to $0.79. The results exceeded our guidance primarily due to higher gross margin. For fiscal year '26, GAAP diluted earnings per share was $3.26, compared with $1.68 in fiscal year '25. Non-GAAP diluted earnings per share was $3.63, compared with $2.06 in fiscal year '25. The GAAP diluted share count increased sequentially from 692 million shares in Q3 to 705 million shares in Q4. The non-GAAP diluted share count increased from 709 million shares to 721 million shares over the same period. Cash provided by operating activities in Q4 was $747 million, compared with cash used in operating activities of $6.6 billion in the prior quarter. For fiscal year '26, cash used in operating activities was $6.8 billion, compared with cash provided by operating activities of $1.66 billion in fiscal year '25. Q4 closing inventory was $12.9 billion, up from $11.1 billion at the end of Q3. CapEx totaled $28 million in Q4, resulting in free cash flow of $722 million. For fiscal year '26, CapEx was $162 million, compared with $127 million in fiscal year '25 as we invested in our expanding capacity globally. During the quarter, we completed public equity offerings, raising $5.6 billion after offering expenses, comprising $1.4 billion of common stock and $4.2 billion of mandatory convertible preferred shares. The proceeds from these offerings will be used primarily to support increased working capital needed to support our new orders. At quarter-end, cash and cash equivalents totaled $7.5 billion. Bank borrowings and convertible note debt totaled $8.7 billion, resulting in net debt of $1.2 billion, compared with net debt of $7.5 billion at the end of the prior quarter. Turning to the balance sheet and working capital metrics, the cash conversion cycle increased by 43 days, from 106 days in Q3 to 149 days in Q4. Days of inventory increased by 13 days to 119 days from 106 days in the prior quarter, as we built inventory in anticipation of higher revenues in fiscal year '27. Days' sales outstanding decreased by 26 days to 59 days from 85 days in Q3 as we realized collections from some large customers. Days payables outstanding decreased by 56 days to 29 days versus 85 days in Q3 due to a significant reduction in our average days' payables between Q3 and Q4 due to the completion of some large AI GPU projects and timing of payments to suppliers. Going forward, we expect the cash conversion cycle to normalize based on terms that we have in our current backlog. Now turning to our outlook for Q1 fiscal year '27, we expect net sales to be in the range of $14.5 billion to $15.5 billion, GAAP diluted net income per share to be between $0.89 and $0.98, and non-GAAP diluted net income per share to be between $1.01 and $1.10. Based on the expected customer and product mix, we expect gross margin to be in the range of 10.4% to 10.8%. We successfully completed the issuance of $4.2 billion of mandatory convertible preferred shares in Q4. Due to that issuance, our GAAP and non-GAAP EPS is computed based on the 2-class method in which we allocated a portion of our net income for participating convertible preferred shares. This impacts our Q1 FY '27 EPS guidance and should be considered in all EPS calculations going forward. Please see the earnings release tables for further details. GAAP operating expenses are expected to be approximately $453 million, including approximately $127 million in stock-based compensation expenses, which are excluded from non-GAAP operating expenses. The outlook for Q1 of fiscal year 2027 fully diluted GAAP EPS includes approximately $106 million in expected stock-based compensation expenses, net of tax effects of $32 million, which are excluded from non-GAAP diluted net income per common share. We expect other income and expense, including interest expense, to result in a net expense of approximately $45 million. The company's projections for Q1 fiscal year '27 GAAP and non-GAAP diluted net income per common share assume a GAAP tax rate of 20.1%, a non-GAAP tax rate of 20.5%, and a fully diluted share count of 745 million shares for GAAP and 761 million shares for non-GAAP. Capital expenditures for Q1 are expected to be in the range of $50 million to $60 million. And for the full fiscal year '27, we expect net sales to be in the range of $65 billion to $72 billion. Michael, we're ready for Q&A now.

Great. Jen?

[Operator Instructions] Your first question comes from the line of Ananda Baruah with Loop Capital.

I have 2, if I could. And congrats on the strong results and the ongoing improvement in profitability here. And let me start just with that. Charles, Dave, what's a good way to think about what fiscal year '27 gross margins can be? You benefited from mix in June. It sounds like you're absorbing some of that mix from deal pushout in September. It's still a nice improvement apples-to-apples in gross margin guide. Can you walk us through how we should think about sort of the puts and takes on the margins, mix, CPU, things like that. And number one, how should we think about what's the useful estimate for gross margin in fiscal year '27 and what may be the progression? And then I have a quick follow-up.

Okay. Thank you for the question. Yes, I mean we will very carefully control our balance between revenue and profitability. As you know, high-volume GPU margin is usually much lower. CPU, storage, IoT, enterprise application, on the other hand, have a higher margin. So we will try to balance between the 2 verticals. Especially last 12 months, we have continued to grow sales force in enterprise and application, server application, storage. So looking forward, we will consistently grow our overall gross margin. Although we will still grow very aggressive, very faster than GPU, but we will focus much more than before on enterprise and CPU storage. And also DCBBS product line is getting mature. So we are shipping more and more DCBBS hardware and also software service and some switch as well. So DCBBS will be our long-term much better profit margin product line.

And so just to clarify before my second question, Charles, I believe I heard you say you anticipate margins to improve from September quarter levels given the factors that you just mentioned. Is that -- did I hear you correctly there?

September, what…

Yes. So we guided to 10.4% to 10.8% for September. And we're doing everything we can, as Charles mentioned, to find the best margins that we can.

Okay. That's great. And then the follow-up is maybe just to dovetail off of Charles' CPU remarks. It sounds like sort of you had 80% of revenue was AI in your Q3, 60% in Q4, looking for 80% again in Q1. Charles talked about taking on more CPU servers, storage and networking. What is the useful way to think about what that sort of 20% that's not AI? What is that? And then if I could just squeeze in quickly, any update on the Board investigation? The Board inquiry, I assume it's coming close to completion, but any update there would be great, too. And that's it for me.

Yes. It depends on customer mix. And when large data center order a lot for sure, AI percentage will be higher. But when David says 80% will be AI, I believe that including 2 market segments. One is traditional AI. The other one is application AI, agentic AI or edge AI. So the pure AI will be about 60% to 70%, while another 10% to 20% AI will be CPU-based AI or kind of agent AI, edge AI. And the other 20% will be pure traditional server storage, IoT. So targeted 100%.

As to your second question, Ananda, we expect to provide an update shortly, and that's all we can share on this call.

Your next question comes from the line of Joseph Cardoso with JPMorgan.

This is MP on for Joseph Cardoso. For my first question, I just wanted to double-click on your robust orders, which you saw during the quarter. You mentioned $60 billion plus orders. Maybe anything in terms of customer concentration within that order growth which you saw as well as any more incremental color in terms of what really was the driver behind such a robust uptick in orders there? And I have a follow-up.

Yes. I mean, around $60 billion, I would like to say 70% AI, pure AI. The other 30% is either CPU or CPU-based AI kind of edge AI application. So overall, I believe our profit margin mix will be getting better.

Okay. Got it. And for my follow-up, I just wanted to ask in terms of you mentioned that you achieved some success in terms of customer diversification where you have, I think, 9 customers, which are $1 billion plus during FY '26. Anything in terms of the nature of those customers, maybe in terms of Neocloud versus enterprises versus sovereign AI customers? Any more incremental color there will be helpful.

Yes. So we have a lot of emerging Neoclouds and CSPs and so they were -- and some enterprise customers that were in that mix that we mentioned.

Yes. The question is CPU-based AI, for example, NVIDIA now also has Vera CPU-based AI, that's from NVIDIA and AMD CPU-based AI, Arm-based, right, and Intel-based. So now AI is kind of majority still GPU-based, but still the CPU-based AI is also growing quickly, especially for agentic AI application.

Your next question comes from the line of Asiya Merchant with Citi.

Two, if I may. One of them was just, was there any change in buying patterns, specifically for the large DC and CSP customers? I understand that there was a shipment delay into 1Q. From what I understand, the guide incorporates that the shipment from 4Q would be shipped into fiscal 1Q. But are you sensing any change in the buying patterns from these large data center/CSP customers that you're predominant in? And -- because there seems to be some investor concern that maybe these customers are going more directly to ODMs than they have been typically to the likes of Super Micro. And then I have a quick follow-up.

Okay. Yes, for sure. I mean the large data center always have power readiness, data center readiness concern, especially liquid cooling. So our customer base overall have a similar concern as well. But still, basically, the order shipment for September quarter, December quarter have been quite strong. And also, I mean, the Super Micro business model is a little bit special, indeed quite special. We have OEM business, but we also cover ODM business. So we have lots of data center -- large data center customer now and especially kind of Neocloud. And at the same time, we are growing very aggressively for enterprise server, traditional server and storage. So overall, we are kind of, again, both ODM and OEM, we will continue to grow in both ways.

Okay. And for my follow-up, the liquid-cooled data centers, are you able to provide what percentage of revenues those are? And if you can give any further details on verticals between enterprise versus these large data center/CSP customers that you have for the liquid-cooled data centers?

As you know, I mean, we are one of the very early liquid cooling technology leaders. In 2024, for example, we shipped, I guess, 80% plus liquid cooling to the market. And now more and more platforms are liquid cooling ready. And including GPU liquid cooling and CPU liquid cooling, like Vera Rubin and even Vera. Vera is CPU-based. A lot of our Vera CPU base will be liquid cooling as well and some AMD, Intel CPU, also liquid cooling. So overall, liquid cooling will continue to grow quickly and very soon will dominate the data center business, I believe.

Your next question comes from the line of Katherine Murphy with Goldman Sachs.

Charles, you noted that you're making investments into the sales force to address the enterprise opportunity specifically. Can you talk about the progress that you've made here and what further investments need to be made in both go-to-market and in the product features and capabilities of Super Micro's portfolio in order to better address this opportunity? And if this run rate OpEx level is the right way to think about the full year?

Sure. So I'll address the question on the operating expenses. So there are certainly some expenses that we expect to go up and others that we expect to come down. And so we think that the levels that we have will -- are at the proper level. And if you look at our historical growth rate in operating expenses, it's less than half of our revenue growth rate.

Yes. As a technology company, our investment in the new technology continues to be very aggressive. For example, the high-speed switch, the optical technology. So overall, our data center total solution with our DCBBS as a centric kind of focus will be continually strong.

And could you talk more about the sales force and how you're engaging with this expanded enterprise customer set, understanding that this is a broader opportunity than the types of engagements that Super Micro may have had in the past?

Yes. This is Mike Staiger. I just want to chime in on the sales force and some of the sales force changes. You probably saw that we elevated a few of our individuals, Matt Thauberger is Chief Revenue Officer; Vik Malyala is Chief Business Officer. And there's been a focus on efficiency and aligning the sales force with a solution sale element to address the AI opportunity ahead, which is supportive of better margins. So there's definitive action in place to make those improvements, and we'll keep you posted as we go out through the year as we expand and make those changes to address the market opportunity.

As a technology leading company, before we are mostly focused on engineering, production, customer service. But now we're getting focused much more on the enterprise and growing our overall balance, especially enterprise, as you know, the profit margin is always better, right? So we are growing our sales force aggressively now.

Your next question comes from the line of Ruplu Bhattacharya with Bank of America.

David, given the pace of GPU platform transitions, how are you managing inventory risk around each new generation? And what gives you the confidence that the record order backlog that you now have won't result in significant inventory exposure if customer deployment schedules or platform configurations change? I'm asking this because Super Micro has had some issues in the past. And I have a follow-up.

Sure. So I think everyone in the industry, in our industry, has to watch out for changes in technology. But with -- what we found was with prices rising so fast, a lot of times now, some of the old inventory does get resold at favorably. Nonetheless, as you point out, you don't want to get caught having to hold that inventory. There is risk in that. So what we do is we try to ensure as much as possible that we have noncancelable POs. And we also try to match our procurement along with the shipment schedules as much as possible.

Other than that, most of our products are designed based on building block solution. So a lot of our subsystems are compatible or optimized for different product lines or even different generations of product. So that will help us a lot in maintaining inventory when technology generation changes.

Okay. As a follow-up, can I ask now that the business is scaling towards $70 billion of annual revenue, right? How should we think about working capital intensity and operating cash conversion in fiscal '27? I think Charles said something about this in his prepared remarks. I didn't fully catch that. But David, do you expect in fiscal '27, the growth to be self-funded now from operating cash flow? Or will the company need incremental external financing to support inventory and receivables beyond the rates that you recently had?

Sure, Ruplu. So I think as I mentioned in my prepared comments that we do expect the cash conversion cycle to improve. And it's not -- the reason for that is when we look at our backlog, we have improved terms and -- which will help us on our cash flow conversion. So therefore, we expect that this will allow us to carry a greater volume of business. And so we're going to do everything possible to utilize our balance sheet, which is much stronger. And if you look at our current assets and our current liabilities, it's stronger than most companies that you'll see out on the market. And so we expect to use the strength of our balance sheet as well as our good customer base to help us fund our growth.

Yes. Once we keep between $65 billion to $72 billion, I guess our cash flow now is pretty enough. But if there are a chance to grow much higher revenue, then we may need more cash flow, for example, $80 billion or beyond $80 billion. So there are some possibilities like that, but we will carefully control.

Your next question comes from the line of George Notter with Wolfe Research.

I wanted to ask if you're seeing any relief or shift in sort of the AI pricing environment overall? And then sort of just like the balancing revenues and margins commentary, is that -- should we take that to mean that you guys are walking away from some low-margin deals right now? Or how are you sort of managing some of the proof point deals that you guys have done on the next-generation racks in the past? And then I have a follow-up.

Yes, that's why we forecast between $65 billion to $72 billion. So we'd like to support as many customers as we can, but the business has to be healthy. The margin had to be at least meet the minimum financial kind of demand.

Got it. Okay. And then just to dive a little bit deeper in the traditional server and storage benefits right now. Is a lot of that sort of stand-alone CPU demand? What is sort of like the attach rate or the sort of synergies with the AI side of the business look like? And then how are margins sort of apples-to-apples trending in that business?

Yes, very good question. Yes, in the last many years, we fully focused on GPU market, AI market. But when company become bigger, I mean, yes, we circle back to focus on enterprise CPU-based market as well, including enterprise and kind of industrial PC IoT storage-based application. So we are going to make our balance between gross revenue and net profit become a much more healthy balance.

And I think, by the way, I'll add to that. I think we did a reasonable job year-over-year because we grew our top line by 78%, and we grew our bottom line by almost the same amount. So I think that shows on a year-over-year basis, what goals we're after.

Your next question comes from the line of Nehal Chokshi with Northland Capital Markets.

Congrats on amazing gross margin results. Charles, do you see the value add that Super Micro can add to NVIDIA ecosystem being different from the x86 ecosystem, somewhat implied by a discussion in the earnings deck around the Arm AGI CPU platform?

Yes, still lots of chances we can add our value. For example, our DCBBS solution that offers customers a complete data center build-out support, not just GPU, CPU storage, but all the major components for data center. And other than that, lots of agentic AI applications, we have a lot of optimization. For example, Vera-based solution, the Rubin HGX-based and lots of other workstation base. We see still a lot of room where we can differentiate our product from others.

I guess what I'm trying to drive at is that NVIDIA is designing full systems. And where you guys come in is helping end customers customize those full systems. But with the Arm AGI CPU platform, perhaps there's more full system design help that Super Micro can bring to the table relative to NVIDIA ecosystem.

Yes. I mean, yes, for example, better time to market, right? Whenever CPU, GPU available, we -- with our architecture, we are able to provide a better time to market and quality, not just design quality, production quality, deployment quality and service kind of work with customers for the whole data center deployment and bring data center to operation and maintain high availability, make sure customers have a minimal failure system. So we see indeed more and more customers appreciate our partnership. So it's not just buy and go. It's kind of buy and work together.

Okay. Great. And then, David, just real quickly, you mentioned that the backlog has improved terms with respect to cash conversion cycle. Is those improved terms because of customers? Or is it because you're seeing a higher percentage of that backlog represent repeat orders and repeat orders potentially have more favorable terms?

Yes. So you broke up just a little bit, Nehal, but let me answer what I thought I heard you ask, and that is we really had a combination of 2 things. We had new customers come in. And -- but we also had existing customers that we're already selling to. And we tightened -- really, we tightened the terms of those contracts. So that's what gives us a little bit of visibility into our cash conversion cycle.

Your next question comes from the line of Brandon Nispel with KeyBanc Capital Markets.

I wanted to ask about DCBBS. You had previously guided for that to be about 20% of gross profit for this year. Could you maybe update us on how that contributed to revenue and gross profit for this year and how you're thinking about that for 2027? And I have a follow-up.

Yes. Thank you for the question. I mean, yes, DCBBS is a big project. I mean we provide all the data center hardware and also the software -- management software and deployment networking, make sure customers have highest availability and efficient maintenance. So it's a kind of a combination of all. So for example, management software, I mean, earlier next quarter, we will provide a proactive service packaging. That's the feature of the service to maintain customers' maximum availability, make sure all the servers they invest are working instead of failure and waiting there, for example. So I mean, we see a very good -- very big room to grow, including kind of switch, high-bandwidth switch and whole networking design management tool. So we see a big room to grow. 20% should be not far away.

Got it. And David, unpacking your comments around gross margins with 75% coming from mix and 25% coming from tariff and inventory write-downs. It's about $700 million for mix and $230 million from the other bucket on my math. So within mix, I'm curious what did vendor rebates look like this quarter? And then within tariffs, did you book a tariff rebate?

Yes, that's a great question. And let me say, we did not book a tariff rebate in our numbers. We are actively pursuing refunds, but we did not take a benefit for those until we see them. And -- but we do -- we're not -- I think the rest of the industry is expecting that the tariffs may go back up, maybe not be in the same fashion, but they may go back up. So that's why we still kind of look at this as perhaps a onetime benefit, but time will tell. So I think that we had -- but as to rebates, your question about rebates, we did have, I think, a little bit higher rebates because we did -- we had a different mix of business, more rebate-laden this quarter -- this past quarter.

Got it. I appreciate the color. And if I could do one more. On the gross margin guidance, I think it's 10.6% for the first quarter. If we were to normalize that for tariffs and inventory write-downs, how do you see that from a year-over-year standpoint? I know from a reported basis, it's up, but last year, you definitely had more tariffs in the numbers and definitely more inventory write-downs that hurt those results. So I was wondering if you could sort of help us from a normalized year-over-year perspective in the first quarter.

Yes. So my comparisons were more quarter-to-quarter where we came down a lot on tariffs and on excess and obsolete inventory. Year-over-year, I think the same things are going to hold true with tariffs coming down a lot in this quarter with the suspension of the IEEPA tariffs. And I think there's no question that we had good results on our E&O this quarter. And so that was what we would consider a nonrecurring event.

Your final question comes from the line of Victor Chiu with Raymond James.

I wanted to circle back on one of Charles' previous comments. Can you just provide some color around how much of your backlog and end demand are being impacted by the shift towards agentic and inference workloads from traning workloads? Are you observing kind of this inflection right now in your results? And I guess, how does the inflection kind of impact the mix of shipments between CPUs and GPUs in the medium term and kind of going forward?

Yes. Kind of -- it's basically a complicated mix. I would have to say still 70-something percent GPU, maybe high 20% CPU. But still, some GPU now is agentic GPU or kind of edge GPU. So it's kind of -- in terms of profit margin, the edge GPU is between traditional GPU and CPU. So it's a complicated combination. But overall, with our DCBBS, it is growing quickly. So I believe we should be able to maintain the profit margin we plan for.

I got it. And I just -- are you seeing kind of a shift at all? Any changes in this? Or -- I understand mix that you're kind of alluding to, but is this -- how does that compare to maybe a year ago quarter in terms of kind of the workloads influencing that?

Yes, long-term GPU percentage will continue to grow, I believe. But again, a lot of GPU will become -- will be used in application, agentic AI, enterprise AI. So overall GPU market, I believe, will be not just big, but also get into every vertical. And that's what we believe at this moment.

We have reached the end of the Q&A session. This concludes today's call. Thank you for attending. You may now disconnect.