GIGABYTE introduces its GIGAPOD ecosystem, featuring direct liquid cooling solutions, support for NVIDIA Blackwell GPUs, and turnkey deployment services for high-performance data centers.
Giga Computing, GIGABYTE’s server-focused subsidiary, showcased its vision for the data centers of the future at the ISC 2025 conference, fully equipped to meet the challenges of large-scale artificial intelligence. At the heart of its proposal lies GIGAPOD, a modular and scalable rack architecture designed to integrate with leading platforms such as NVIDIA HGX, AMD Instinct, and Intel Gaudi 3.
Infrastructure Ready for Generative AI
Specifically designed for high-density AI training and inference environments, GIGAPOD enables the deployment of GPU clusters at scale with centralized management. Giga Computing’s proposal extends beyond hardware: it includes the GIGABYTE POD Manager (GPM) software platform, allowing for efficient monitoring, management, and optimization of data center resources.
With its flexible design, GIGAPOD can be installed in racks with air cooling or direct liquid cooling (DLC), facilitating a gradual transition or a comprehensive rollout from scratch.
Direct Liquid Cooling: Energy Efficiency in Action
A major focus of Giga Computing’s exhibition has been the demonstration of its G4L3 Series servers with DLC, integrated in a 4+1 rack configuration. This technology allows for a significant reduction in energy consumption while maintaining optimal temperatures for intensive computational workloads such as language model training or digital twins.
Thanks to strategic partnerships with top-tier technology partners, Giga Computing offers validated system-level solutions and accelerates deployment times for integrators and cloud providers.
End-to-End Services: From Consulting to Maintenance
Beyond hardware, Giga Computing has announced its evolution toward a comprehensive AI data center provider model. This includes L12 services, encompassing initial consulting and infrastructure planning through system integration, operational deployment, and ongoing support. These capabilities are particularly tailored for hyperscalers, enterprise data centers, and cloud service providers.
Optimized Platforms for AI Across the Board
Giga Computing has designed servers compatible with the leading acceleration architectures in the market:
NVIDIA RTX PRO™ Server Edition, featuring NVIDIA Blackwell GPUs and MGX systems, aimed at workflows with agent-based AI, real-time rendering, and enterprise applications.
AMD Platform, equipped with EPYC 9005 processors and Instinct MI350 GPUs, complemented by AMD Pensando™ NICs for high-performance networking.
- Intel Platform, based on 5th generation Xeon CPUs and Gaudi 3 accelerators, ideal for enterprise-scale AI.
Blackwell NVL72: 72 GPUs in One Cabinet
The star of the show has been the NVIDIA GB300 NVL72, an AI inference system that integrates 72 NVIDIA Blackwell GPUs, 36 Grace CPUs, connectivity with SuperNICs ConnectX-8, and comprehensive liquid cooling, all within a single high-performance cabinet.
This deployment exemplifies Giga Computing’s capability to lead the implementation of solutions designed for the most demanding models of today’s landscape, from LLMs to multimodal generative AI.
Keynote on the Future of Data Centers
As part of the official ISC 2025 program, Giga Computing announced a presentation titled:
“GIGAPOD: Redefining the Future of the Data Center Beyond Servers and Software”
- Location: HPC Solutions Forum, Hall H, Stand L01 (Ground Floor)
- Date and Time: Wednesday, June 11, 3:00 to 3:20 PM
A Bold Step Towards the Smart Data Center
With the introduction of GIGAPOD and its integrated solutions, Giga Computing demonstrates its commitment to scalability, energy efficiency, and operational flexibility, three key pillars for data centers aiming to successfully navigate the new era of artificial intelligence.
Interested organizations can visit Giga Computing at Stand D40 of ISC 2025 to learn more about their AI infrastructure solutions and explore collaboration opportunities.
Source: Gigabyte