MiTAC’s bold play at COMPUTEX 2025: Redefining sustainable AI with liquid-cooled server power

At COMPUTEX 2025 in Taipei, MiTAC Computing Technology Corp., a prominent subsidiary of MiTAC Holdings Corp., made a definitive pivot from being a traditional hardware manufacturer to becoming a full-spectrum provider of AI and data center infrastructure solutions. Leveraging its experience in high-performance computing, artificial intelligence, and sustainable infrastructure, the company introduced a comprehensive range of next-generation products designed to address the growing complexity and energy demands of AI-powered data centers.

This marks MiTAC’s first appearance under its consolidated brand identity at COMPUTEX, and it seized the opportunity to position itself as a strategic partner for organizations seeking intelligent, energy-efficient, and scalable server platforms. With a central theme of “Empowering Unlimited Possibilities: AI x Sustainability,” MiTAC is aligning its offerings with the twin goals of technological acceleration and environmental responsibility—both key drivers of modern enterprise and hyperscale computing.

What Are MiTAC’s Key Innovations in AI Servers and Liquid Cooling Systems?

At the heart of MiTAC’s exhibit were its AI inference and HPC servers, purpose-built for real-time data mobility, model training, and edge-to-cloud deployment. Among the highlights was the MiTAC G4527G6 (MGX 4U), which made its global debut featuring eight NVIDIA RTX PRO 6000 Blackwell GPUs. The platform was further enhanced with the unveiling of the first-ever NVIDIA MGX PCIe Switch board integrated with ConnectX-8 SuperNIC and dual Intel Xeon 6767P CPUs, enabling powerful, parallel processing for advanced inference workloads.

Alongside this, MiTAC also showcased the G8825Z5 (8U) model powered by AMD EPYC 9005 processors and AMD Instinct 350 GPUs—marking a strategic move to support generative AI and large language model computation. These servers reflect MiTAC’s growing role in enabling AI democratization by making powerful AI-ready infrastructure accessible for enterprise-grade workloads.

The company also placed a spotlight on its full-rack liquid cooling architecture. Unlike legacy air-cooled designs, MiTAC’s solutions combine Liquid-to-Air and Liquid-to-Liquid cooling configurations to cater to varying thermal demands. These systems, supported by top-tier coolant distribution units (CDUs) and MiTAC’s in-house management software, are engineered to manage workloads exceeding 2000 kW, demonstrating the firm’s readiness for thermal challenges posed by modern AI workloads.

Why Is Liquid Cooling Gaining Traction in AI and HPC Infrastructure?

Liquid cooling has rapidly evolved from an experimental deployment to a foundational requirement for next-gen data centers, especially in AI and HPC environments. MiTAC’s Direct Liquid Cooling (DLC) solution, featured in its new multi-node HPC server C2820Z5, exemplifies this shift. Engineered for ORv3 racks and powered via 48VDC systems, these servers not only boost computational performance but also reduce overall power usage, thereby enhancing data center efficiency while meeting global sustainability benchmarks.

With the ability to cool high-power CPUs more effectively and reduce system-level thermal risk, DLC is fast becoming an industry standard. MiTAC’s DLC servers are customized for verticals such as genomics, financial modeling, AI simulation, and edge inferencing—industries that require low latency, sustained throughput, and eco-conscious deployment.

How Does MiTAC’s Collaboration with AMD and NVIDIA Strengthen Its Offerings?

One of the most strategic pillars of MiTAC’s COMPUTEX 2025 presence is its collaboration with leading silicon vendors. The appearance of AMD GPU Market Development Director Ted Marena and commentary from Ken Lin, Senior VP of AMD Taiwan, reflected a strong alignment between MiTAC and AMD’s GPU and CPU roadmaps. The launch of AMD Instinct MI350 GPUs, tailored for extreme AI workloads and integrated with MiTAC’s liquid-cooled platforms, illustrates a shared vision of scalable performance and energy optimization.

Similarly, MiTAC’s commitment to NVIDIA’s latest MGX platform and Blackwell GPU architecture signals a future-oriented approach. The co-innovation seen in PCIe switching and NIC technologies shows MiTAC’s ability to remain at the forefront of data mobility and parallel processing.

These joint developments highlight how MiTAC’s server platforms are becoming reference designs for AI cluster deployments globally, particularly as organizations prioritize environmental goals alongside performance.

What Strategic Vision Drives MiTAC’s Data Center Transformation?

According to Rick Hwang, President of MiTAC Computing, the company is embracing open innovation as a core philosophy. By adhering to Open Compute Project (OCP) standards and deepening technical integration with partners, MiTAC is delivering end-to-end infrastructure that meets emerging AI requirements while simplifying deployment and management.

More than just hardware innovation, MiTAC is rethinking data center architecture from a platform perspective. Its server portfolios now reflect a system-wide design approach—encompassing compute, thermal, power, and software orchestration—to reduce complexity and improve TCO (Total Cost of Ownership) for global clients. The inclusion of integrated telemetry and system-level diagnostics demonstrates MiTAC’s commitment to lifecycle value and operational transparency.

How Is MiTAC Positioning Itself for the Future of AI and HPC?

MiTAC’s trajectory is clearly aligned with the global acceleration of AI adoption. By combining high-efficiency servers, liquid cooling expertise, and strong ecosystem partnerships, the company is building foundational blocks for the AI-driven digital economy. With cloud service providers, research institutions, and private enterprises increasingly seeking integrated infrastructure with minimal energy footprints, MiTAC’s differentiated platform strategy places it in a strong position to capture this demand.

The emphasis on sustainability is not a marketing theme but a design imperative. As AI workloads grow exponentially, energy efficiency and thermal management will determine the viability of deployments at scale. MiTAC’s ability to address this through system-level engineering and long-term innovation cycles is a signal of maturity rarely found in mid-tier server vendors.

Looking ahead, MiTAC Computing is expected to deepen its involvement in AI-native computing infrastructure, offering tailored platforms that can be rapidly deployed in modular data centers, edge environments, and private clouds. Its continued focus on quality, performance, and integration—backed by robust global support—positions the company as a key enabler of the next wave of sustainable AI and HPC growth worldwide.