GIGABYTE’s enterprise arm builds some of the industry’s densest GPU computing platforms, from single servers to complete rack-scale AI clusters. XENON has partnered with GIGABYTE for years, custom-building GPU servers for Australian and New Zealand research organisations, and integrates the current generation across AI, HPC and visualisation workloads.
AI Factories
GIGAPOD is GIGABYTE’s turnkey rack-scale AI infrastructure: factory pre-integrated servers, networking, power and cooling, scaling from a single GPU server to 8 racks and 256 GPUs. Flagship nodes include the G894 8U server with NVIDIA HGX B300 (8 Blackwell Ultra GPUs) and 4U direct-liquid-cooled G4L3 systems packing up to 64 GPUs per rack, with GIGABYTE POD Manager (GPM) software running operations. GIGABYTE also builds liquid-cooled compute trays for NVIDIA GB300 NVL72 rack-scale systems.
HPC
GIGABYTE HGX-class servers take Intel Xeon or AMD EPYC hosts, and GIGAPOD supports NVIDIA HGX, AMD Instinct and Intel Gaudi accelerators, so XENON can match the platform to your codes rather than the other way around. Direct liquid cooling and single-phase immersion options, with reference deployments at KDDI and the German Aerospace Center, keep dense clusters efficient.
Data Management
GIGABYTE S-Series storage servers provide high-density NVMe and SATA/SAS capacity as the storage layer within GIGABYTE-based clusters and AI data pipelines.
Services
GIGAPOD ships factory-validated, and XENON adds local design, integration, burn-in, deployment and ongoing support, so a rack-scale AI cluster arrives as a working system, not a pile of boxes.
Servers & Workstations
For deskside AI, the GIGABYTE AI TOP ATOM packs an NVIDIA GB10 Grace Blackwell Superchip, 1 petaflop of FP4 AI performance and 128 GB of unified memory into a compact system that runs models up to 200 billion parameters. Above it sits the W775 deskside supercomputer on NVIDIA DGX Station GB300 architecture, with roughly 775 GB of coherent memory and up to 20 petaflops of AI performance. GIGABYTE E-Series short-depth edge servers cover inference at the edge.
Talk to XENON’s experts about building beyond on GIGABYTE.