Giga Computing Delivers an Optimized Rack-scale Solution Portfolio to Streamline Deployment of AI Factories

Configured to Deliver Performance in Rack-level Power Architecture

April 9, 2026 - Giga Computing, a leader in accelerated computing and infrastructure solutions, today announced its portfolio of rack-scale solutions that are optimized and tailored for high-performance workloads where throughput is paramount. These new integrated systems move the discussion from which platforms are best to how many racks are needed to achieve performance goals.

As AI and HPC continue to redefine the boundaries of what is possible, it has become clear that networking, specifically bandwidth and latency, plays a pivotal role in enabling these advancements. Delivering this capability within a predictable, rack-level architecture is now essential. While high-speed interconnects remain critical, overall system performance must be considered holistically. Improving performance per watt continues to drive next-gen processor design; however, these gains have been accompanied by a significant rise in power demands. As a result, data center racks and infrastructure have evolved dramatically, data center racks have moved from 10-20kW configurations to 50-100kW and beyond. 

No longer is the focus on the management of an individual server, its now about the rack management and the clustering of resources. This new infrastructure requires compute, storage, and memory resources to be pooled and dynamically allocated in an efficient and highly utilized manner. This system-level approach has unified networking, centralized orchestration, and optimized topology. Racks can be built with a focus on resources such as CPU/memory, GPUs, and storage. So, multiple rack configurations must be created to ensure an optimized and tightly integrated system.

Giga Computing has designed and built rack-scale configurations that include advanced cooling technologies (such as management nodes and manifolds), network switches, CDUs, PDUs, and optional management software.

Sample of the Rack-scale Projects:

•        DLB2-CB3: NVIDIA GB300 NVL72
•        DL83-GP6: liquid cooling, 8x NVIDIA HGX B300 and 16x x86 processors
•        DL83-GP0: liquid cooling, 32x AMD Radeon AI PRO R9700S and 8x x86 processors
•        DL83-BL0: liquid cooling, 100x AMD EPYC™ 9005 Series processors

Giga Computing’s shift to rack-scale orchestration provides the high-density infrastructure required for modern AI and HPC workloads. By integrating advanced cooling with pooled resource management, these configurations deliver the throughput and efficiency necessary to meet the power demands of next-generation data centers.

For a complete list, visit: https://www.gigabyte.com/us/Enterprise/GIGAPOD-Rack-Scale