At the beginning of May 2026, Super Micro Computer (SMCI) released its financial report for the third quarter of fiscal year 2026. It’s important to clarify SMCI’s fiscal calendar: Q1 covers July–September, Q2 is October–December, Q3 spans January–March, and Q4 runs April–June. Therefore, Q3 of fiscal year 2026 reflects business performance up to March 31, 2026, and the report was published on May 5, 2026, marking a completed event.
Key financial highlights include: revenue of $10.2 billion, representing a year-over-year increase of 122%; non-GAAP earnings per share of $0.84, significantly beating market expectations of $0.62; and non-GAAP gross margin rebounding from a previous low of 6% to 10.1%, far exceeding analysts’ estimates of 6.75%. Notably, quarterly revenue declined by about 19.7% compared to the previous quarter (Q2 FY2026), mainly due to seasonal shipping patterns and inventory clearance ahead of the Blackwell platform transition.
With the financial report’s release, SMCI’s backlog of orders surpassed $13 billion, with Blackwell Ultra-related orders as the primary driver. Following the announcement, SMCI’s stock price jumped 9% in a single day on May 28. Boosted by Dell’s strong earnings and a wave of AI product launches during Computex, SMCI’s share price surged from a 2026 low of roughly $19.60 to over $48, marking a cumulative increase of 145% (this figure reflects the total gain from the start of 2026 to date, not a weekly rise). As of June 3, 2026, the stock was consolidating near $48, approaching a "golden cross" pattern—where the 50-day moving average crosses above the 200-day moving average—a technical signal often indicating sustained upward momentum.
From a broader industry perspective, Gartner data shows global server market spending is projected to reach $382 billion in 2025 and grow to $466 billion in 2026, with AI-related expenditures accounting for 76% starting in 2026. Focusing further, AI-optimized server spending is expected to rise from $280 billion in 2025 to $353 billion in 2026, a year-over-year increase of 26.2%, with a projected five-year compound annual growth rate (CAGR) of 28.2% through 2029.
A Fundamental Shift in Financials: Gross Margin Recovery and Order Visibility
SMCI’s recent rally isn’t simply a case of speculative hype around technology concepts—it’s anchored in a performance report that shows improvement both in scale and quality.
| Key Metric | Q3 FY2026 Data | Change |
|---|---|---|
| Revenue | $10.2 billion | YoY +122%, QoQ -19.7% |
| Non-GAAP EPS | $0.84 | Above expectations ($0.62) |
| Non-GAAP Gross Margin | 10.1% | Previous low: 6%, far above forecast (6.75%) |
| GAAP Gross Margin | 9.9% | Concurrent recovery |
| Backlog Orders | Over $13 billion | Covers baseline for future shipments |
The rebound in non-GAAP gross margin from 6% to 10.1%, well above analysts’ 6.75% estimate, signals that SMCI’s profitability hasn’t suffered a structural breakdown. Combined with a backlog exceeding $13 billion, SMCI has partially locked in its revenue base for upcoming quarters. Despite a sequential revenue drop, the market is more focused on the gross margin recovery and the positive impact of the Blackwell Ultra ramp-up cycle on profitability.
Competitive Landscape: SMCI vs Dell vs HPE
In the global server market for 2025, Dell Technologies leads the OEM segment with $12.5 billion in revenue and a 10.0% market share, driven by accelerated server growth. Super Micro follows closely with a 9.5% market share, also achieving triple-digit growth fueled by accelerated servers. HPE ranks fifth in the overall server market.
These three companies exhibit differentiated competition in the AI server segment:
Dell has established a clear scale advantage in AI servers. In the first quarter of fiscal year 2027 (covering February–April 2026), Dell’s AI server revenue reached $16.1 billion, a staggering 757% year-over-year surge. Single-quarter new AI orders totaled $24.4 billion, with backlog orders hitting a record $51.3 billion, as demand consistently outpaces supply. The company raised its full-year AI server revenue guidance to around $60 billion. Notably, Dell added about 1,000 new enterprise AI customers this quarter, bringing the total to over 5,000, indicating that AI spending is spreading from cloud service providers to a broader enterprise base.
HPE adopts a more conservative, profit-focused strategy in AI. Since Q1 FY2023, the company has signed about $6.8 billion in cumulative AI orders, with over 60% coming from sovereign and enterprise clients. Compared to the other two, HPE’s AI server revenue proportion is lower (about 13% of total revenue), and its order conversion cycle is longer due to data center readiness, power supply, and cooling constraints.
SMCI differentiates itself through liquid cooling design, rapid custom delivery capabilities, and a modular server architecture tailored for the Blackwell series chips (DCBBS: Data Center Building Block Solutions). According to Goldman Sachs’ 2025 industry report, Dell is capturing high-end market share from SMCI. However, Dell’s $51.3 billion backlog and $60 billion full-year AI server revenue guidance provide a clear future revenue pipeline for the entire AI hardware sector, boosting demand logic across the board—including SMCI.
Whether SMCI can consistently maintain gross margins above 10% will be a key indicator of its core competitive strength and recovery.
AI Server Supply Chain Overview: From Upstream Chips to Downstream Assembly
The AI server industry chain can be divided into five layers. Understanding the players and their interdependencies in each segment helps identify current profit transmission paths and potential bottlenecks.
Upstream—Chips and Advanced Packaging
NVIDIA is at the heart of this layer. Its latest data center GPU architecture, Blackwell Ultra, is expected to enter mass production in the second half of 2026, with AI chip capacity directly setting shipment limits for downstream server manufacturers. On the manufacturing side, TSMC is the main foundry for NVIDIA and AMD’s AI chips and the primary provider of CoWoS advanced packaging capacity. Capacity research estimates TSMC’s CoWoS monthly output will expand from about 115,000 wafers at the end of 2026 to 175,000 by the end of 2027 and 220,000 by the end of 2028. Currently, over 80% of TSMC’s CoWoS capacity supports AI applications.
Midstream—Server ODM Manufacturing
Taiwanese firms dominate this segment. Hon Hai (2317) holds about 40% of the global AI server market share and commands 52% in NVIDIA Blackwell GB200 full cabinet assembly. Chairman Liu Young-way has set a "strong growth" outlook for 2026, targeting annual revenue of NT$900 billion. Quanta (2382) expects triple-digit percentage growth in annual AI server revenue and has upgraded its general server outlook to double-digit growth, with large cloud customer order visibility extending beyond 2027. Wistron (3231) projects high double-digit percentage revenue growth for the year and has raised its 2026 capital expenditure to NT$60 billion. Inventec (2356) anticipates server business revenue to grow by over 30% annually, with server revenue share likely exceeding 50%.
The ODM Direct model continues to gain share in the overall server market. The trend of large CSPs developing proprietary ASICs is accelerating, and some ODMs have expanded their AI cabinet capacity from GPU solutions to ASIC solutions, with the GPU-to-ASIC product mix expected to approach an 8:2 ratio.
Lower Midstream—Brand Manufacturers
SMCI, Dell, and HPE are all branded server manufacturers, but their business models differ. SMCI combines ODM-style rapid customization with direct delivery of comprehensive AI infrastructure solutions to enterprise clients. Dell has raised its AI server revenue forecast for fiscal year 2027 to about $60 billion, with AI backlog orders reaching $51.3 billion, suggesting that AI server procurement cycles are still expanding. HPE’s AI strategy is more focused on margin.
Peripheral—Cooling, Power, Chassis, and Liquid Cooling Systems
SMCI has strong ties to the Taiwanese supply chain. Its server shipments directly drive demand for cooling modules (such as Auras 3324, Chaun-Chi 3017), power supplies (like Delta Electronics 2308, Lite-On Technology 2301), backplanes, and server chassis (such as Ching Cheng 8210). For example, the GB300 NVL72 cabinet has a thermal design power (TDP) of 142 kW, making liquid cooling not just optional but standard.
Quantitative Anchors for Market Size
Three sets of data provide the quantitative foundation for assessing the AI server market’s fundamentals, all sourced from verifiable third-party institutions.
| Source | Forecast Target | Market Size | Key Implication |
|---|---|---|---|
| Gartner | AI-optimized servers, 2025 | $280 billion | Market baseline |
| Gartner | AI-optimized servers, 2026 | $353 billion | YoY +26.2% |
| Gartner | AI-optimized servers, 2023–2029 | CAGR 28.2% | Five-year compound growth |
| Gartner | Global servers, 2025 | $382 billion | Includes non-AI segment |
| Gartner | Global servers, 2026 | $466 billion | AI spend share: 76% |
| IDC | Global servers, full year 2025 | $444.1 billion | YoY +80.4%, record high |
Dell’s backlog grew from $43 billion to $51.3 billion, and the addition of 1,000 enterprise AI customers in a single quarter signals that demand for AI servers is expanding from model training to inference deployment. This structural shift may sustain hardware demand for several fiscal years to come.
Objective Risk Analysis: Key Variables to Watch
While SMCI and the AI server industry show clear growth drivers, objective risks must also be considered.
Can Gross Margin Remain Above 10%?
SMCI’s non-GAAP gross margin rebounded from a trough of 6% to 10.1%, marking a critical fundamental signal for this rally. If future quarters see gross margin drop below 8%, it would suggest SMCI remains reliant on price competition to gain market share, undermining its "profit quality recovery" narrative. It’s essential to closely monitor SMCI’s pricing strategy during the Blackwell Ultra ramp-up. If gross margin stays above 10% as scale expands, competitive pressure is easing; otherwise, a reassessment is warranted.
Independent Board Review and Export Compliance Issues
SMCI is undergoing an independent board review directly related to export compliance. There’s no clear timeline yet. If the final outcome points to substantive export control violations, SMCI’s ability to ship to China and certain other markets could be restricted. This risk cannot be mitigated by current revenue data and remains a systemic variable requiring ongoing attention.
Chip Supply Rhythm and Blackwell Ultra Shipment Uncertainty
SMCI’s server business is tightly linked to NVIDIA GPU shipment schedules. If Blackwell Ultra’s mass production ramp in the second half of 2026 falls short of expectations, the conversion of SMCI’s $13 billion backlog into actual revenue will slow accordingly. Any revision to NVIDIA’s next-generation AI chip shipment guidance will be a key indicator for SMCI’s revenue growth prospects.
Sustainability of ODM Competitive Dynamics
The share of the global server market held by ODM Direct models continues to rise. Trends such as large CSPs developing proprietary chips, brand manufacturers like Dell expanding rapidly, and NVIDIA’s own system-level integration could alter the profit structure and market share landscape for Taiwanese ODMs. These are supply chain dynamics that warrant long-term monitoring.
Conclusion
Super Micro’s cumulative 145% gain from its 2026 low (year-to-date) is driven by three layers of fundamental improvement: rapid revenue growth ($10.2 billion, YoY +122%), non-GAAP gross margin recovery to 10.1%, and a backlog exceeding $13 billion providing short- and medium-term visibility. Compared to Dell and HPE, SMCI stands out in liquid cooling design, Blackwell platform adaptation, and rapid custom delivery, but gross margin quality and regulatory risks remain key factors affecting its valuation.
The expansion of the AI server industry is supported by third-party data from Gartner and IDC: the AI-optimized server market is $280 billion in 2025, expected to grow to $353 billion in 2026, with a five-year compound growth rate of 28.2%. As AI computing shifts from model training to large-scale inference deployment, the persistence of server hardware procurement cycles will be a central theme for the coming quarters and years. Investors should focus on three quantifiable signals in financial reports: gross margin trends, backlog order digestion speed, and the actual shipment pace of Blackwell Ultra, and combine these with their own risk tolerance to make independent decisions.
