What Mitac Computing Technology does

MiTAC Computing Technology Corporation is a Taiwan-headquartered subsidiary of MiTAC Holdings Corporation that designs and manufactures high-performance GPU-accelerated servers, multi-node platforms, storage servers, motherboards, and rack-scale AI/HPC infrastructure for hyperscale data centers, HPC facilities, and enterprise customers. Core products include the G4826Z5 (4U liquid-cooled AI server with up to 8 AMD Instinct MI355X GPUs), G4520G6 (4U Intel Xeon 6 platform supporting 8 NVIDIA GPUs), G8825Z5 (8U high-density GPU server with air or liquid cooling), C2811Z5 and C2810Z5 (OCP-compliant multi-node servers with AMD EPYC 9005), and the S8040 motherboard; the portfolio spans AMD EPYC and Intel Xeon 6 CPU platforms with AMD Instinct and NVIDIA GPU accelerators, complemented by proprietary open-source firmware (MiOBMC on OpenBMC, MiOPF on EDK2), a POD management platform (MiCoreView), and Diamond Cooling technology co-developed with Akash Systems that uses synthetic diamond substrates to remove heat up to five times faster than copper.

The company generates revenue primarily through hardware sales — both barebone platforms and fully integrated rack-scale solutions — sold via direct enterprise field sales and a global network of authorized resellers and distributors. Pricing is quote-based and varies by CPU/GPU selection, cooling type, and rack integration scope; multi-year contracts are typical for hyperscale and enterprise deployments. Manufacturing is vertically integrated across facilities in Taoyuan and Hsinchu (Taiwan) and Shunde and Kunshan (China), with sales, engineering, and service offices in Shanghai, Tokyo, Crawley (UK), and Newark, California; the company has 5,001–10,000 employees and operates as a private subsidiary of MiTAC Holdings, with no public equity listing identified.

MiTAC serves hyperscale data centers, HPC and scientific computing (e.g., Qarnot in France with planned Italy expansion), enterprise/cloud IT, broadband network infrastructure (Harmonic), global streaming entertainment, and online gaming. Differentiators include vertical integration from PCBA to L12 rack manufacturing, proprietary cooling and firmware, OCP-compliant open hardware, and a deep co-development ecosystem spanning AMD, NVIDIA, Intel, Canonical, Broadcom, Samsung, Micron, Solidigm, CoolIT, DDN, and Rafay. Recent strategic moves include the Akash Systems Diamond Cooling launch (with a reported $300M initial order), the Tonomia TonoForge modular data center partnership for 12-week sovereign-AI deployments, and validated benchmark wins on Llama 3.1 8B (95.8% scaling efficiency) and Pebble Sonar-powered MI350X inference (28% throughput gain within the same power envelope).

Name

Mitac Computing Technology

Legal name

MiTAC Computing Technology Corporation

Website

https://mitaccomputing.com

Company type

Private

Founded year

2014

Operating status

Operating

Headcount range

251–500 employees

Short description

MiTAC Computing Technology is a Taiwan-based subsidiary of MiTAC Holdings that designs and manufactures GPU-accelerated AI, HPC, and cloud servers, with proprietary liquid and Diamond Cooling, open-source firmware, and rack-scale integration for hyperscale and enterprise data centers worldwide.

Ownership category

akta.pro rank

Product category

AI and HPC Server Hardware

NAICS

Computer and Peripheral Equipment Manufacturing (33411), Computer Systems Design and Related Services (5415), Computer Storage Device Manufacturing (334112)

SIC

Electronic Computers (3571), Computer Storage Devices (3572), Services-Computer Integrated Systems Design (7373)

akta.pro primary industry

AI Server Systems & HGX/Accelerator Platforms (HDAAAAAB)

akta.pro secondary industries

AI Datacenter Power, Cooling & Racks (liquid cooling, power delivery) (HDAAAAAH), Hyperscale/OEM & Open Compute (OCP) Server Platforms (HDACABAI)

Keywords

• AI server hardware
• GPU accelerated servers
• Rack scale integration
• Liquid cooling systems
• ODM server manufacturing

Headquarters

HQ city

Taoyüan

HQ country

Taiwan

HQ region

Asia

Offices8 records

Markets served

GTM type

B2B

Offering type

Hardware or Manufacturing

Cost components

Supply Chain, Personnel, Technology or R&D, Operations, Infrastructure, Marketing or Sales

Revenue model

1. Hardware Sales (AI Servers, GPU Servers, Multi-node Servers, Storage Servers): MiTAC Computing generates revenue primarily through the sale of high-performance server hardware including AI/GPU servers, multi-node servers, general purpose servers, storage servers, and motherboards. The company offers both barebone platforms and fully integrated rack-scale solutions. Revenue is driven by the growing demand for AI training and inference infrastructure, HPC clusters, and enterprise data center modernization.
2. Rack-Scale Integration and Turnkey Solutions: MiTAC offers bundled rack-scale solutions including pre-configured liquid-cooled AI racks (e.g., 52U High-Density AI Liquid-Cooled Rack with AMD Instinct GPUs), deployment-ready OCP servers, and modular data center solutions. These turnkey offerings carry premium pricing for integrated cluster deployments targeting hyperscale and enterprise AI data centers.

Pricing tiers

Go-to-market motion1 record

Distribution channels3 records

Marketing channels9 records

Core offering

MiTAC Computing designs, manufactures, and integrates high-performance GPU-accelerated servers and rack-scale AI/HPC infrastructure solutions for hyperscale data centers, HPC facilities, and enterprise customers. The portfolio spans AI servers (G4826Z5, G8825Z5, G4520G6), OCP-compliant multi-node servers (C2811Z5, C2810Z5), general-purpose servers, storage servers, and motherboards powered by AMD EPYC, Intel Xeon 6, AMD Instinct, and NVIDIA accelerators. The offering is complemented by proprietary Diamond Cooling, direct liquid cooling architectures, open-source firmware (MiOBMC, MiOPF), the MiCoreView POD management platform, and end-to-end rack-scale integration services delivering turnkey deployments.

Product overview

MiTAC Computing Technology offers a comprehensive portfolio of high-performance server solutions organized around AI servers, multi-node servers, general purpose servers, storage servers, and motherboards. The product line includes GPU-accelerated AI servers (G4826Z5, G4520G6, G8825Z5, TN85B8261) designed for AI training and inference workloads; multi-node servers (C2810Z5, C2811Z5, D50DNP) for cloud and HPC deployments; and enterprise platforms (R2520G6, GC73A-B8046). These are complemented by cooling technology solutions including direct liquid cooling, Diamond Cooling with Akash Systems, and rack-scale integration services providing end-to-end deployment from supply chain to validation. Software offerings include open-source firmware MiOBMC (OpenBMC-based) and MiOPF (EDK2-based), plus MiCoreView POD management platform. The company also provides modular data center solutions through the TonoForge partnership with Tonomia for rapid 12-week deployments.

Differentiator

Problem solved

Functional benefit

Brands

• TYAN: Registered trademark of MiTAC Computing Technology Corporation or MiTAC International Corporation, used for server products.

Products and services

• G4826Z5 AI Server 4U liquid-cooled AI server supporting up to 8 AMD Instinct MI355X GPUs with direct liquid cooling architecture, designed for AI training and inference workloads.
• G4520G6 GPU Server 4U GPU-optimized platform featuring Intel Xeon 6 processors with support for NVIDIA GPUs, designed for generative AI and LLM workloads.
• G8825Z5 AI Server 8U high-density GPU server supporting up to 8 AMD Instinct GPUs (MI325X, MI350X, MI355X) with air or liquid cooling options.
• C2811Z5 Multi-node Server OCP-compliant liquid-cooled multi-node server powered by AMD EPYC 9005 Series processors with DDR5 memory and NVMe E1.S storage interfaces for HPC environments.
• C2810Z5 Multi-node Server Air-cooled multi-node server powered by AMD EPYC 9005 processors for cloud and HPC deployments.
• TN85B8261 AI Server 2U server with AMD EPYC 9005/9004 Series processors supporting 4 GPUs for AI and HPC workloads.
• D50DNP Multi-node Server 2U multi-node server with 5th/4th Gen Intel Xeon Scalable Processors for cloud and HPC deployments.
• R2520G6 General Purpose Server 2U enterprise server with Intel Xeon 6 processors for cloud infrastructure and general IT workloads.
• GC73A-B8046 General Purpose Server Compact 1U single-socket server powered by AMD EPYC 8005/8004 processors with 12 NVMe U.2 bays for edge, cloud, and storage deployments.
• S8040 Motherboard Single-socket AMD EPYC 8005/8004 motherboard with PCIe Gen 5 and 10GbE networking for data-intensive workloads.
• Diamond Cooling Technology Proprietary cooling technology using diamond's thermal conductivity for up to 10°C GPU temperature reduction and 22% higher FLOPs per watt, developed with Akash Systems.
• Liquid Cooling Solutions Direct liquid cooling (DLC) architecture and warm liquid cooling solutions for high-density AI workloads, supporting configurations from 64 to over 256 GPUs.
• Rack-scale Integration Services End-to-end rack design, manufacturing, and integration services covering supply chain management, hardware installation, testing, and custom configuration for data center deployment.
• 52U High-Density AI Liquid-Cooled Rack Pre-configured 52U rack integrating 12 MiTAC G4826Z5 AI servers with 96 AMD Instinct MI355X GPUs, delivering 50% higher GPU density per rack compared to standard configurations.
• MR1100 Series Liquid-Cooled Rack Liquid-cooled rack supporting up to 256 GPUs for sustainable HPC and AI workloads.
• MR2200 Series HPC Rack High-performance computing rack portfolio with air-cooled and liquid-cooled servers powered by AMD EPYC 9005 processors.
• TonoForge Modular Data Center Modular data center solution developed in partnership with Tonomia for rapid deployment of liquid-cooled AI infrastructure in 12 weeks.
• MiOBMC Firmware Open-source firmware stack built on OpenBMC for AI and HPC data center environments, providing baseboard management controller functionality.
• MiOPF Firmware Open-source BIOS firmware based on EDK2, part of MiTAC's open-source firmware solutions for enterprise server platforms.
• MiCoreView POD Management Platform POD management platform for rack-scale AI infrastructure monitoring and management.

Quantifiable outcome

• 50% higher GPU density per rack (52U High-Density AI Liquid-Cooled Rack integrating 96 AMD Instinct MI355X GPUs)
• +11 more outcomes

Named customers5 records

Segments4 records

Ideal customer profiles3 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

Integration18 records

AI capability5 records

Feature6 records

Partnerships

16 partnerships are on record, tiered flagship, core and minor.

• Pebble (MyPebble Inc.)flagshipStrategic or Co-development Partner · 26 June 2026Pebble and MiTAC conducted a joint demonstration showing that Pebble's Sonar power management software increased AI inference throughput by 28% on a power-constrained AMD Instinct MI350X cluster (running MiTAC G8825Z5 servers serving Meta's Llama 3.1 405B model) without drawing additional power. Sonar recovered ~15% of wasted power headroom and brought two additional GPUs online within the same 6kW power envelope.
• XPerf Inc.coreStrategic or Co-development Partner · 18 June 2026XPerf Inc. partnered with MiTAC Computing to validate AMD Instinct MI325X AI air-cooled racks (featuring G8825Z5 servers) using XPerf's ClusterReady validation platform. The validation demonstrated 20% higher performance than AMD's publicly reported FLUX.1-dev benchmark results and produced the first reported Llama 3.1 8B pre-training results on MI325X. ClusterReady achieved 95.8% scaling efficiency and 10x improvement in multi-node communication bandwidth, compressing rack validation from weeks to hours/days.
• TonomiaflagshipStrategic or Co-development Partner · 3 June 2026MiTAC partnered with Tonomia to launch the TonoForge Modular Data Center, enabling deployment-ready liquid-cooled server racks through the TonoForge system. This partnership reduces traditional data center deployment timelines to 12 weeks, enabling decentralized sovereign AI infrastructure globally. Tonomia completed development of TonoForge with strong engineering support from MiTAC Computing and assistance from AMD, entering commercial release in June 2026.
• AMDcoreStrategic or Co-development Partner · 2 June 2026MiTAC collaborates with AMD extensively across multiple product lines including servers powered by AMD EPYC 9005/9004/8005 processors and AMD Instinct MI355X/MI350X/MI325X GPUs. At COMPUTEX 2026, MiTAC showcased 52U liquid-cooled AI racks integrating 96 AMD Instinct MI355X GPUs delivering 50% higher GPU density. At ISC 2026, the G4826Z5 liquid-cooled server with AMD Instinct MI355X GPUs was highlighted. AMD liquid-cooled AI clusters scale from 64 to 256 GPUs.
• NVIDIAcoreStrategic or Co-development Partner · 2 June 2026MiTAC offers NVIDIA MGX architecture servers supporting up to 8 NVIDIA H200 NVL or RTX PRO 6000 Blackwell GPUs, and NVIDIA-qualified GPU systems. At NVIDIA GTC 2026, MiTAC showcased flexible AI hardware based on NVIDIA MGX architecture. Strategic collaborations with industry leaders including DDN and Rafay for enterprise-scale AI data pipelines and management platforms.
• IntelcoreStrategic or Co-development Partner · 2 June 2026MiTAC collaborates with Intel on servers powered by Intel Xeon 6 processors, including the G4520G6 GPU-optimized platform with NVIDIA GPU options, R2520G6 enterprise servers, and OCP-compliant platforms. Products target AI and HPC workloads with scalable, energy-efficient configurations for sustainable data centers. MiTAC holds a license to manufacture certain Intel server boards (acquired April 6, 2023).
• CanonicalcoreTechnology or Integration · 2 June 2026MiTAC integrates Canonical Ubuntu 26.04 LTS across its AI hardware portfolio, powering secure, scalable, and production-ready AI infrastructure. Ubuntu is featured as part of MiTAC's 'AI Factory' rack-scale solutions, providing a validated OS platform for enterprise AI deployments.
• BroadcomcoreTechnology or Integration · 2 June 2026Broadcom network chips are integrated into MiTAC's rack-scale AI solutions, contributing to high-efficiency network architectures supporting 400Gb/s and 800Gb/s networking for ultra-fast data transfer in AI training clusters.
• MicroncoreTechnology or Integration · 2 June 2026Micron memory is incorporated into MiTAC's AI and HPC server platforms as part of the 'AI Together' ecosystem partnership platform, supporting high-performance compute workloads.
• SolidigmcoreTechnology or Integration · 2 June 2026Solidigm storage solutions are part of MiTAC's ecosystem partnerships for AI infrastructure, contributing to high-capacity storage for data-intensive AI workloads.
• DDNcoreTechnology or Integration · 2 June 2026DDN storage solutions are featured in MiTAC's NVIDIA MGX-based AI server solutions, providing high-performance storage for AI data pipelines and enterprise-scale AI deployments.
• RafaycoreTechnology or Integration · 2 June 2026Rafay is featured in MiTAC's NVIDIA GTC 2026 showcase for enterprise AI server solutions, providing Kubernetes management and MLOps platform integration for AI infrastructure deployments.
• Samsung (Memory)coreTechnology or Integration · 2 June 2026Samsung memory components are integrated into MiTAC's AI cluster rack solutions featuring technologies from collaborators at Supercomputing 2025, supporting high-bandwidth memory for GPU-accelerated AI workloads.
• CoolIT SystemscoreTechnology or Integration · 29 April 2026CoolIT cooling systems are integrated into MiTAC's rack-scale AI cluster solutions, providing liquid cooling technology for high-density GPU deployments at Supercomputing 2025 and OCP summits.
• FuturenestminorStrategic or Co-development Partner · 29 April 2026Futurenest is listed as a technology partner in MiTAC's 'AI Together' ecosystem partnership platform announced at COMPUTEX 2026, contributing to MiTAC's diversified AI infrastructure ecosystem.
• Akash SystemsflagshipStrategic or Co-development Partner · 3 March 2026Akash Systems and MiTAC launched the world's first Diamond Cooled AI servers powered by AMD Instinct MI350X/MI355X GPUs. Akash's Diamond Cooling technology (using synthetic diamond substrates for 5x faster heat transfer than copper) is manufactured by MiTAC Computing into G8825Z5 Diamond-cooled AI servers, delivering up to 10°C GPU temperature reduction, 22% higher FLOPs per watt, and throttle-free performance at over 95°F inlet temperatures. Akash received a $300M initial launch order, projecting up to $1M incremental value per server over 4 years.

Scale indicators2 records

Recent moves6 records

Expansion highlights7 records

Direct peers

• Wiwynn: Wiwynn is a white-box server ODM spun out of Wistron, focused on hyperscale cloud and AI/ML infrastructure. Its direct liquid-cooled and high-density GPU servers overlap almost completely with MiTAC's AI server and rack-scale portfolio.
• Supermicro: Supermicro is the closest direct peer, specializing in GPU-optimized AI/HPC servers and liquid-cooled rack-scale platforms using NVIDIA and AMD accelerators. It targets the same hyperscale, HPC, and enterprise AI customers as MiTAC and competes head-to-head on GPU density, liquid cooling, and time-to-deploy.
• Inventec: Inventec is a major Taiwan ODM producing AI servers, hyperscaler systems, and storage platforms for global brands. Its server business directly competes with MiTAC on GPU-accelerated platforms and rack-scale integration for enterprise and cloud customers.
• Wistron: Wistron designs and manufactures AI servers, hyperscale ODM systems, and rack-level solutions, overlapping with MiTAC's core product lines. As a fellow Taiwan-based ODM with global manufacturing, it is a direct comparable on enterprise and cloud AI server deals.
• ASUSTeK Computer (ASUS): ASUS's enterprise server business (ASUS Servers and ESC series) offers GPU-accelerated AI servers and HPC platforms that compete with MiTAC's AI server lineup. Both are Taiwan-based and target enterprise, HPC, and AI training customers.
• GIGABYTE Technology: GIGABYTE's G-Series servers (including G593 and G294) provide GPU-dense AI and HPC platforms that directly compete with MiTAC's AI servers. As another Taiwan-headquartered systems builder, GIGABYTE is a frequent competitor on enterprise AI and HPC tenders.
• Quanta Computer: Quanta is a leading Taiwan-based ODM that designs and manufactures GPU servers, hyperscaler systems, and rack-scale AI infrastructure. It is a direct competitor to MiTAC's ODM business, especially for cloud and AI workloads at hyperscale customers.

Broad incumbents

• Hewlett Packard Enterprise (HPE): HPE's ProLiant and Cray HPC systems, along with HPE Private Cloud AI, compete directly with MiTAC's GPU and HPC servers in enterprise and research deployments. HPE offers a broader services and software stack but overlaps materially on AI server hardware.
• Dell Technologies: Dell's PowerEdge server portfolio, including the PowerEdge XE9680/XE8640 and integrated Dell AI Factory, competes with MiTAC's AI servers in enterprise accounts. As a broad incumbent with a wider portfolio, Dell overlaps with MiTAC on GPU servers but also bundles storage, networking, and services.
• Lenovo: Lenovo's ThinkSystem and ThinkEdge servers, including GPU-rich SR650/SR670 and ThinkSystem SR780a V3, target the same AI training and inference workloads as MiTAC. As a broad incumbent with ThinkShield and TruScale services, Lenovo competes across multiple buyer tiers.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat6 records

Key risks6 records

Key highlights7 records

Customer concentration

• LinkedInlinkedin.com/company/35512
• Xx.com/MiTACcomputing

Compliance1 record

Revenue estimate

Valuation estimate

Number of profiles

Profiles1 record

Funding overview

Funding rounds

Investors

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Mitac Computing Technology