ASRock Rack

永擎 / 永擎電子股份有限公司

ASRock Rack designs and integrates server motherboards and GPU-dense systems, converting NVIDIA and AMD accelerator platforms into deployable AI infrastructure for cloud and enterprise customers.

TWSE: 7711
Profile as of September 14, 2026
#86 in the AsiaAI Tech Index
How the ranking works →

ASRock Rack is a fast-growing AI-server integrator benefiting from its Pegatron group ties and rapid qualification of new NVIDIA and AMD platforms. Its central challenge is navigating intense price competition, working-capital demands, and heavy reliance on upstream accelerator allocation.

Key figures

Revenue FY2025: NT$24.876 billion
Revenue growth FY2025: Not disclosed
Operating margin FY2025: 3.86% Estimate
Net income FY2025: NT$803.48 million
R&D spend FY2024: NT$735.347 million
Capital expenditure FY2024: NT$6.02 million
Market capitalization Sep 11, 2026: NT$38.07 billion Estimate
FY2024 revenue growth FY2024: 225.3% year over year
Top ten customer concentration 2024-12-31: 94.33% of receivables
Related-party purchases FY2024: NT$1.964 billion

Overview

ASRock Rack is the server-hardware arm of ASRock and sits within the Pegatron group structure. Founded in 2013, it designs and sells server motherboards, barebones, complete server systems, storage platforms, and high-performance computing systems. The company listed on the Taiwan Stock Exchange in November 2025, transitioning from an operating subsidiary to a publicly traded entity.

While independently listed, ASRock Rack is not operationally independent of its corporate ecosystem. It relies heavily on ASRock's distribution channels and related-party operating network, and Pegatron serves as its ultimate parent. This structure provides critical procurement, logistics, and manufacturing support as it competes in the global server market against larger original design manufacturers.

Its core business model focuses on providing the physical infrastructure required for modern data centers. By combining its heritage in motherboard design with advanced system integration, ASRock Rack bridges the gap between semiconductor manufacturers and end-user deployments, offering highly configurable solutions for cloud service providers and enterprise customers.

The AI angle

ASRock Rack supplies the physical infrastructure required for AI compute. It designs and integrates multi-GPU servers, storage servers, and edge systems based on AMD EPYC, Intel Xeon, Arm, and Ampere processors. Its most direct AI products are high-density accelerator systems that house the industry's most advanced silicon.

The company's named offerings include NVIDIA HGX B200 and B300 GPU servers, NVIDIA GB300 NVL72 rack-scale systems, NVIDIA RTX PRO 6000 Blackwell Server Edition systems, and AMD Instinct MI350X GPU servers. For example, its 8U16X-GNR2 uses an NVIDIA HGX B300 NVL16 baseboard with 16 Blackwell Ultra GPUs and 2.3 terabytes of HBM3e memory. Its XGMRACK configuration uses the NVIDIA GB300 NVL72 platform, integrating 36 Grace CPUs, 72 Blackwell Ultra GPUs, more than 20 terabytes of HBM3e memory, and NVIDIA Quantum-X800 InfiniBand networking.

AI demand reaches ASRock Rack through higher-value server boards and integrated systems. While it does not manufacture GPUs, CPUs, high-bandwidth memory, or advanced packaging, its revenue depends directly on demand from cloud providers, AI startups, and enterprise deployments for these compute-dense platforms. The link to AI is direct at the hardware level, as it provides the necessary power, cooling, and connectivity for AI workloads.

Technology and moat

ASRock Rack's competitive capability is server-platform integration rather than proprietary AI silicon. It combines motherboard design, board management controllers, BIOS and firmware, chassis selection, GPU and CPU validation, power delivery, storage, high-speed I/O, thermal design, and liquid-cooling integration into deployment-ready configurations.

This validation work is complex and difficult to replicate quickly for a specific platform, as it requires coordination across processor, accelerator, networking, memory, operating-system, and customer qualification cycles. The company publicly demonstrates this capability through rapid support for successive accelerator generations, including B200, B300, GB300, and expected Rubin GPU platforms.

However, the company's moat relies on engineering execution and platform breadth rather than a proprietary patent portfolio. It remains fundamentally dependent on NVIDIA and AMD for platform availability and design requirements. Its competitive distinction is not ownership of the core compute intellectual property, but its ability to quickly and reliably bring those technologies to market in configurable server systems.

Five-pillar assessment

Scale and market position: A growing AI-server integrator within the Pegatron group, though smaller in manufacturing scale than tier-one original design manufacturers like Quanta or Supermicro.
Technology and R&D: Strong execution in motherboard design, platform integration, and liquid cooling, without proprietary AI silicon or a disclosed patent moat.
Supply-chain centrality: Deeply tied to NVIDIA and AMD accelerator roadmaps, and reliant on the ASRock and Pegatron group for distribution and procurement.
Financial momentum: Rapid revenue growth driven by AI-server demand, though working-capital intensity and inventory funding remain significant operational factors.
Governance and quality: Transitioned to public-market governance with an independent audit committee, though it remains closely tied to its parent group.

Supply chain and relationships

ASRock Rack depends heavily on a narrow set of semiconductor designers, primarily NVIDIA and AMD, for AI accelerators and platform qualification. It sources server processors from AMD, Intel, Arm, and Ampere, while procuring high-speed interconnect components, embedded memory, power systems, thermal components, and printed circuit boards from various specialized suppliers. The company also relies on related-party procurement within its corporate group, purchasing significantly from ASIAROCK TECHNOLOGY LIMITED.

On the customer side, ASRock Rack serves cloud service providers, AI startups, and enterprise customers across North America, Europe, Asia-Pacific, and Greater China. While it does not publicly name its largest end customers in audited filings, customer concentration is exceptionally high. The top ten accounts represented over 94% of the company's receivables at the end of 2024, indicating that a small number of large deployments drive the majority of its revenue.

Suppliers

NVIDIAGPU and platform ecosystem for AI server designs
AMDEPYC processors and Instinct MI350X GPU platforms
IntelProcessor and networking component ecosystem
ArmCPU architecture ecosystem
Ampere ComputingAmpereOne Arm-based CPU ecosystem
ASIAROCK TECHNOLOGY LIMITEDRelated-party procurement

Partners

ASRock Europe B.V.Related-party European distribution
ZutaCoreLiquid-cooling collaboration
DexterityPhysical-AI supercomputer collaboration

Competitors

Super Micro ComputerInferredCompetes in AI GPU servers and motherboards
Gigabyte TechnologyInferredCompetes in Taiwan-origin server boards and systems
WiwynnInferredCompetes for cloud and AI server deployments
Quanta Cloud TechnologyInferredCompetes in cloud server manufacturing
InventecInferredCompetes in server original design manufacturing
ASUSInferredCompetes in enterprise servers and motherboards

Geopolitics and risk

ASRock Rack's most significant concentration risk is geographic. Its headquarters, management, core product development, and group relationships are all centered in Taiwan. This exposes the company to regional cross-strait disruption, potential shipping interruptions, and the growing trend of customer contingency planning that seeks to diversify hardware procurement away from single-region dependencies.

Additionally, the company must navigate a complex geopolitical landscape regarding technology transfers. Its high-end AI systems incorporate U.S.-origin GPUs, CPUs, networking components, and software ecosystems. This makes its exports to restricted destinations strictly subject to changing U.S. export controls. ASRock Rack is entirely dependent on the compliance decisions and export licenses of its upstream suppliers, particularly NVIDIA and AMD, which dictate where and to whom its most advanced AI server configurations can be sold.

Risk matrix
Risk	Severity	Why it matters
NVIDIA and AMD accelerator allocation	High	GPU-server shipments depend on receiving sufficient qualified accelerator modules.
U.S. export-control exposure	High	AI systems incorporate U.S.-origin components subject to changing export rules.
Taiwan cross-strait disruption	High	Headquarters and core operations are centered in Taiwan.
Customer concentration	High	The top ten customers account for 94.33% of FY2024 year-end receivables.
Working-capital intensity	High	Growth requires substantial inventory funding, impacting operating cash flow.
AI-server price competition	High	Larger original equipment manufacturers limit pricing power.
Related-party procurement dependency	Medium	Disruption within the group supply chain can affect fulfillment and margins.
Thermal and data-center deployment complexity	Medium	Customer deployment delays can defer server recognition and increase support obligations.

Governance and ownership

ASRock Rack is a public company, independent of state ownership or control. Its corporate governance sits within a private Taiwanese technology group structure, with ASRock Incorporation acting as the direct parent and Pegatron Corporation as the ultimate parent.

The company is led by professional management, including Chairman Lung-Lun Hsu and General Manager Wei-Xu Sha. Ahead of its public-market listing on the Taiwan Stock Exchange in November 2025, ASRock Rack transitioned to a more formal listed-company governance structure. This included establishing an audit committee composed of independent directors in October 2024, replacing its previous supervisor system. While it now operates with public financial reporting, detailed disclosures regarding ultimate beneficial ownership percentages and board composition remain limited.

What to watch

Whether FY2026 revenue growth remains above the FY2025 base while operating margins recover.
Whether NVIDIA HGX B300, GB300 NVL72, and Rubin platforms translate into commercial shipment volume.
Whether AMD Instinct MI350X systems become a meaningful second AI-accelerator revenue stream.
Whether top-ten-customer receivable concentration falls below the 94.33% reported at the end of 2024.
Whether inventory growth converts into operating cash flow.

Recent developments

June 1, 2026Announced AI infrastructure based on the NVIDIA Vera CPU at Computex 2026.
March 16, 2026Displayed liquid-cooled AI platforms built around NVIDIA Rubin GPUs at NVIDIA GTC 2026.
November 19, 2025Listed on the Taiwan Stock Exchange under ticker 7711.
October 13, 2025Announced a NVIDIA HGX B300 server with waterless liquid cooling at OCP Global Summit 2025.
May 20, 2025Introduced an AI server based on AMD Instinct MI350X GPUs at Computex 2025.
March 19, 2025Presented NVIDIA Blackwell Ultra, GB300 NVL72, and B200 liquid-cooled AI-server configurations at NVIDIA GTC 2025.

Coverage on AsiaAI.FYI

We haven't written about ASRock Rack yet. Subscribe below to get it when we do.

Guides that cover ASRock Rack

Taiwan AI Server Supply Chain

About this profile

Compiled with AI-assisted research from company filings, market data, and published reporting as of September 14, 2026, then reviewed by AsiaAI.FYI. Figures marked Estimate are not company-reported. Check primary filings before relying on any number.

Confidence: B. Audited FY2024 financials and product platforms are well sourced, but FY2025 details, customer names, and segment breakdowns are limited.

Main sources: ASRock Rack audited standalone FY2024 financial statements; Taiwan Stock Exchange listing information; Taiwan market company-profile data; Company press releases and product announcements.

All 44 sources

Position in the AI stack

The hardware integration layer, designing the motherboards, chassis, and cooling systems that house AI accelerators.

Legal name: ASRock Rack Incorporation
Headquarters: Taipei City, Taiwan
Founded: 2013-01
Leadership: Lung-Lun Hsu / 許隆倫, Chairman
Wei-Xu Sha / 沙韋旭, General Manager
Employees: Average 317 (FY2024)
Ownership: Publicly listed; part of Pegatron Corporation
Listings: TWSE: 7711
Business: AI server motherboards and integrated systems
Sector: Servers, PCs & ODMs
Website: asrockrack.com