Giga Computing Showcases Rack-Scale Infrastructure and Agentic AI at Supercomputing 2026
As the focus of data centers shifts toward massive, integrated deployments to support the rise of Agentic AI, Giga Computing is positioning itself at the center of this transition. At Supercomputing 2026 (SC26) in Chicago, the GIGABYTE subsidiary is showcasing a full-spectrum hardware portfolio designed to handle the entire AI deployment lifecycle—ranging from compact, personal AI supercomputers to large-scale, rack-scale infrastructure and modular 'AI Factories.'
The exhibition highlights a move toward unified compute and liquid cooling, emphasizing how hardware, power management, and partner ecosystems must integrate to meet the demands of hyperscalers and AI cloud providers. A key focus is the expansion of the GIGABYTE GIGAPOD portfolio, which aims to simplify resource management for complex AI workloads through integrated rack-scale solutions.
Expanding Rack-Scale and Liquid Cooling Solutions
To address the complexities of modern data center management, Giga Computing is promoting its G-REX solution. This management rack system integrates compute nodes with simplified liquid and power management (including CDUs, PDUs, and leak detection) into a single interface. This ecosystem is bolstered by partners like UfiSpace, which provides switches for AI training clusters and out-of-band management, and Vast and WEKA, which offer high-speed centralized storage for HPC clusters.
The company is also pushing high-density, liquid-cooled CPU compute racks designed to capture up to 90% of system heat via cold plates on the CPU, memory, and NICs. Notable hardware includes the B685-D80-LS1, built for AMD EPYC™ 9006 CPUs (offering up to 5,120 cores per server), and the G684-A80-LAS1 for Intel® Xeon® 6 CPUs. For environments where facility water connections are unavailable, the newly introduced H295-D80-CS1 offers a 2U, four-node closed-loop liquid cooling system, providing high efficiency without the need for extensive plumbing infrastructure.
Platform-Specific Solutions: AMD, Intel, and NVIDIA
Giga Computing's lineup covers all major silicon architectures, providing specialized hardware for different AI and HPC requirements:
Looking toward future deployments, Giga Computing is highlighting its GAIFA (GIGABYTE AI Factory Accelerator) and GADU (GIGABYTE Accelerated Deployment Unit) reference designs, which are intended to simplify the rapid deployment of modular, containerized AI infrastructure at scale.