Gigabyte G292-Z24-AAX1 4-GPU AMD MI300X Server SKU

The Gigabyte G292-Z24-AAX1 is a purpose-built 2U server platform engineered to house four AMD Instinct MI300X accelerators, delivering a unified high-bandwidth memory and compute architecture designed for the most demanding AI training, inference, and HPC workloads in enterprise datacenters. Each AMD MI300X APU integrates CDNA 3 GPU compute dies alongside CPU-class host dies and 192 GB of HBM3 memory per accelerator, yielding up to 768 GB of aggregate accelerator memory across the four-GPU configuration — enabling very large language models and scientific datasets to reside entirely within accelerator memory without host offloading.

The G292-Z24-AAX1 platform is built on AMD EPYC processor infrastructure and leverages high-bandwidth interconnects between accelerators, making it suitable for tightly coupled parallel workloads where inter-GPU communication latency is critical. The system supports PCIe Gen 5 fabric and is designed with enterprise-grade power delivery and thermal management to sustain the sustained compute throughput required by generative AI model training, large-scale inference serving, and numerical simulation pipelines.

As part of Gigabyte's enterprise server portfolio, the G292-Z24-AAX1 is positioned for datacenter operators across UAE, GCC, EMEA, and APAC regions who require a validated, production-ready platform for AMD MI300X-based deployments. The system is available through Omnixon Global, supporting enterprise procurement workflows for organizations scaling AI infrastructure across hyperscale and sovereign cloud environments.

Ideal for

• Large language model (LLM) pre-training and fine-tuning requiring 100B+ parameter models resident entirely within accelerator HBM3 memory
• High-throughput generative AI inference serving for real-time enterprise applications including RAG pipelines and multi-modal workloads
• Computational fluid dynamics (CFD) and finite element analysis (FEA) simulations in engineering and energy sector HPC environments
• Genomics, molecular dynamics, and drug discovery workloads leveraging unified memory to process large biological datasets without CPU-GPU data staging
• Enterprise AI infrastructure consolidation, replacing multi-node clusters with fewer high-memory accelerator nodes to reduce datacenter footprint and networking complexity
• Sovereign AI cloud buildouts in GCC and APAC regions requiring locally operated, high-capacity GPU server platforms certified for regulated industries

Technical specifications

Available from Omnixon Global. Submit an RFQ and our team will confirm configuration and availability for your order.