Dell PowerEdge XE9680 8-GPU AMD MI300X Server

The Dell PowerEdge XE9680 configured with eight AMD Instinct MI300X accelerators represents one of the most memory-dense and compute-capable server platforms available for enterprise AI and HPC environments. Each AMD Instinct MI300X APU integrates a unified CPU and GPU compute complex with 192 GB of HBM3 memory per accelerator, delivering 5.3 TB/s of aggregate memory bandwidth per GPU. The XE9680 chassis provides the mechanical and thermal infrastructure to sustain full rack-scale performance across all eight accelerators simultaneously, interconnected via AMD Infinity Fabric for high-bandwidth peer-to-peer communication.

The AMD Instinct MI300X is built on a chiplet architecture combining CDNA 3 compute dies with high-bandwidth memory stacks in a single package. This design enables the accelerator to address extremely large model contexts and dataset footprints in-memory, which is a persistent bottleneck in large language model inference and training. With 1,536 GB of total accelerator memory across the eight-GPU node, the XE9680 MI300X configuration can natively host multi-hundred-billion-parameter models without offloading to host DRAM or NVMe storage, significantly reducing latency and improving throughput for AI inference serving.

Designed for enterprise datacenters, cloud service providers, and national research institutions, the Dell PowerEdge XE9680 with MI300X accelerators supports ROCm open software stack compatibility, enabling workloads built on PyTorch, TensorFlow, JAX, and MIOpen to run without proprietary lock-in. The platform is optimized for high-density rack deployments and integrates with Dell OpenManage and AMD ROCm SMI for lifecycle and telemetry management. Omnixon Global supplies this configuration to enterprise and public-sector buyers across UAE, GCC, EMEA, and APAC regions.

Ideal for

• Large language model (LLM) training and fine-tuning for models exceeding 100 billion parameters, enabled by the 1,536 GB aggregate HBM3 memory pool across eight MI300X accelerators
• High-throughput AI inference serving for generative AI applications where low-latency in-memory model residence eliminates swap overhead
• Scientific and national-lab HPC workloads including molecular dynamics, climate simulation, and computational fluid dynamics requiring sustained FP64 throughput
• Multi-tenant AI cloud infrastructure where memory capacity per node allows simultaneous hosting of multiple large model instances
• Data analytics and in-memory database acceleration for financial services, telecommunications, and energy sector workloads demanding high memory bandwidth
• Sovereign AI and government research deployments in EMEA and APAC where open-standard ROCm software stack meets regulatory and procurement requirements

Technical specifications

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