MiTAC Computing Showcases Full-Stack Innovation to Scale Open AI Infrastructure at OCP Global Summit 2026

, /PRNewswire/ -- MiTAC Computing Technology Corporation, a subsidiary of MiTAC Holdings Corporation (TWSE: 3706) and a global leader in high-performance server solutions, today unveiled its latest AI infrastructure innovations at the 2026 OCP Global Summit, Booth D22 , under the theme " Scaling Open AI Infra with Full-Stack Innovation ."

MiTAC Computing Showcases Full-Stack Innovation to Scale Open AI Infrastructure at OCP Global Summit 2026

Large-scale AI clusters and AI factories demand more than compute performance alone. MiTAC Computing takes a full-stack approach, transforming fragmented infrastructure into a cohesive, scalable foundation designed for deployment, management, and expansion.

From Scalable AI Factories to Modular Data Centers: Flexible Infrastructure for Every Scale

As AI workloads expand from training, inference, and RAG to Agentic AI, infrastructure needs vary by scale and deployment environment. MiTAC Computing's AI Infrastructure Stack spans compute, storage, and liquid-cooled servers through open platform firmware, operating systems, AI software ecosystems, and AI Pod management, providing a consistent, scalable foundation for diverse workloads.

For large-scale, centralized AI environments, MiTAC Computing delivers its Scalable AI Factory , combining full-stack capabilities with rack-scale integration to support high-density computing and continued expansion. For distributed sites or rapid buildouts, MiTAC offers modular data center solutions . The TonoForge™ Modular Data Center , developed with Tonomia , incorporates MiTAC G4826Z5 liquid-cooled AI servers and rack-scale integration. Its proven 12-week deployment demonstrates how the modular approach can accelerate the deployment of high-performance GPU AI clusters.

From centralized AI factories to distributed modular data centers, MiTAC gives enterprises the flexibility to build AI infrastructure around their compute requirements, site conditions, and deployment needs.

Breaking Thermal Barriers with Advanced Liquid Cooling

Higher compute density and rack-scale AI factories are making advanced cooling critical to unlocking greater compute performance. MiTAC Computing is showcasing multiple advanced liquid cooling solutions, including a PoC with ZutaCore that integrates the MiTAC G4826Z5 with ZutaCore ® HyperCool ® Waterless Two-Phase Liquid Cooling technology . Using a dielectric fluid that changes phase from liquid to vapor directly at the chip, the technology removes heat generated by CPUs and GPUs at the source.

The technology demonstrates the potential to support more than 4,000W per chip and average heat flux exceeding 250W/cm² . Under defined operating conditions, an annualized model indicates a pPUE as low as 1.03 , highlighting the value of two-phase cooling for increasingly power-dense AI infrastructure.

On October 14, MiTAC Computing will present an Expo Hall Session exploring the evolution of high-density AI systems toward next-generation rack-scale platforms and how advanced liquid cooling can overcome thermal constraints to unlock greater compute performance.

Extending Openness Across the AI Stack

Alongside advances in hardware performance, MiTAC Computing continues to develop its Open Source Firmware Solution . The G4826Z5 demonstrates full-stack integration from bare metal to AI applications. At the platform level, MiTAC's internally developed MiOPF™ supports fast system initialization, while MiOBMC™ provides real-time hardware monitoring and telemetry, giving customers greater firmware control and customization flexibility while reducing dependence on a single vendor.

Across the software ecosystem, the G4826Z5 runs Canonical-certified Ubuntu 26.04 OS and integrates Kubernetes and the DDN Data Intelligence Platform . It also incorporates AMD ROCm™ open software stack to unlock the compute performance of AMD Instinct™ MI355X GPUs . MiTAC Computing is working with Rafay on GPU cluster management to support flexible GPU-as-a-Service (GPUaaS) resource pools , enabling on-demand compute allocation based on AI workload requirements while simplifying deployment and management.

In an Executive Session later today, MiTAC Computing will explore open infrastructure management through technologies such as Open Platform Firmware (OPF), OpenBMC, and Redfish. The discussion will examine how open standards and OCP community collaboration can improve infrastructure interoperability, and how MiCoreView™ , MiTAC's infrastructure management software built on open-source technologies, can further extend management capabilities from the platform level to rack and cluster scale. MiTAC will also share practical insights into deployment, telemetry, data center rack and cluster management, and system orchestration.

New Platforms Target OCP Computing and AI Inference

MiTAC Computing is making the global debut of the C2611Z6 and showcasing the latest G4520G6E3BC-N configuration , expanding its portfolio for high-density OCP computing and enterprise AI inference.

C2611Z6 : A liquid-cooled multi-node server designed for OCP ORv3 racks and powered by the latest AMD EPYC™ 9006 Series Server processors . It integrates four compute nodes in a compact 2OU form factor, with each node supporting 16 DDR5-8000 RDIMM slots and OCP 3.0 expansion for high-density, liquid-cooled infrastructure.

G4520G6 (G4520G6E3BC-N) : A 4U2S GPU server built for enterprise AI inference. Powered by Intel ® Xeon ® 6 processors , the system supports up to eight NVIDIA RTX PRO 4500 / NVIDIA 6000 Blackwell Server Edition GPUs or NVIDIA H200 GPUs . The configuration also features 12 NVMe E3.S drive bays , combining high-density GPU compute and high-speed storage in a single 4U platform for data-intensive inference workloads.

MiTAC Computing invites OCP Global Summit attendees to visit Booth D22 and experience how full-stack innovation is helping scale open AI infrastructure.

Learn more:

AMD Platform Brochure

Intel Platform Brochure