Dell PowerEdge XE9680 AI Server

Flagship 6U AI Server Dell PowerEdge XE9680 for LLM Training and HPC

Dell PowerEdge XE9680 is a flagship, ultra-powerful specialized 6U server built for extreme workloads in artificial intelligence (AI), large-scale training and inference of large language models (LLMs), deep machine learning, and high-performance computing (HPC). The platform combines a dual-socket compute node with a massive OEM chassis for graphics accelerators and is rightfully considered the benchmark for AI infrastructure deployment.

The XE9680 is a universal foundation for building GPU clusters of any scale: from a single node for model fine-tuning to multi-rack SuperPOD fabrics with ultra-low latency. Support for multiple generations of AI accelerators (NVIDIA, AMD, Intel) allows the platform to be adapted to a specific task and budget without changing the chassis.

Use Cases

The Dell PowerEdge XE9680 is the optimal solution for organizations requiring:

• Training and fine-tuning of large language models (LLMs)
• Industrial-scale generative AI model inference
• Deep machine learning and computer vision
• High-performance computing (HPC) and scientific simulation
• GPU cluster and AI factory (SuperPOD) construction with high accelerator density
• Working with ultra-large language models (DeepSeek, Qwen, etc.) without quantization within a single node

GPU Architecture and Compute Platform Support

The XE9680’s key advantage is the versatility of its base AI architecture and support for 8 SXM graphics processors with ultra-high-bandwidth inter-chip interconnect. Depending on requirements, configurations with leading AI platforms are available:

• NVIDIA Solutions. Support for NVIDIA HGX H100 (80 GB) platforms based on HBM3 memory or the latest HGX H200 (141 GB) systems with ultra-fast HBM3e memory, providing total VRAM per server of over 1.1 TB. Accelerators are connected via NVLink at up to 900 GB/s per processor. Entry-level AI builds based on 8 × NVIDIA A100 (80 GB) are also available.
• AMD Solutions. A variant supporting AMD Instinct MI300X (192 GB), offering extreme GPU memory capacity for ultra-large language models (such as the DeepSeek or Qwen family) without quantization within a single node.
• Alternative Accelerators. Option to configure with Intel Gaudi3 (128 GB) AI modules.

Compute Node Technical Specifications

Processors. Supports two 4th Gen Intel Xeon Scalable (Sapphire Rapids) or 5th Gen Intel Xeon Scalable (Emerald Rapids) processors. Top-tier CPUs with TDP up to 350 W and up to 64 cores per socket (a total of up to 128 cores per server) are supported.

Memory. 32 DIMM slots for TruDDR5 ECC RDIMM server memory. Maximum RAM capacity reaches 4 TB, with data rates of up to 4800 MT/s or up to 5600 MT/s (depending on processor generation).

Storage Subsystem. The front bay supports up to 8 × 2.5″ NVMe SSDs (PCIe Gen4/Gen5 direct connect) with maximum raw capacity of up to 122.88 TB. Modern variants supporting up to 16 energy-efficient EDSFF (E3.S) NVMe drives are available. Secure internal mirrored M.2 SSD modules are used for OS and hypervisor boot.

I/O and Networking. Up to 10 × PCIe Gen 5.0 x16 full-height expansion slots, available on the front and rear panels. This allows the server to be equipped with an array of high-speed network adapters (e.g., 8-port InfiniBand / RoCE Ethernet cards at up to 200/400 Gbps per port from Mellanox/Broadcom) for building cluster fabrics (SuperPOD) with ultra-low latency.

Cooling, Power, and Management

• Cooling. The XE9680 platform by default relies on high-performance air cooling with massive heatsinks and a redundant array of high-power fans. (Note: for NVIDIA Blackwell architecture chips, a related modification, PowerEdge XE9680L, has been released, focused on Direct Liquid Cooling (DLC).)
• Power. The fault-tolerant power subsystem includes 6 redundant power supplies rated at 2800 W each. Redundancy schemes 3+3 or 5+1 (FTR) are supported with the highest Titanium efficiency rating.
• Remote Management. Comprehensive monitoring of component health, thermal profiles, and power distribution is handled by the embedded iDRAC10 remote management chip with support for the Dell OpenManage ecosystem and Silicon Root of Trust security protocols.

Why GHI SERVERS

• Direct supply from China. We work with CNY and RUB. Turnkey logistics of server equipment from China, including customs clearance and delivery to our Moscow warehouse.
• Custom configuration. We help select the GPU platform type (NVIDIA HGX H100/H200, AMD MI300X, Intel Gaudi3), drive count and type, InfiniBand/Ethernet network adapters, and power supplies for your task and budget.
• Moscow warehouse. Prompt replacement of defective units and warranty service without long downtime.
• 20+ years of expertise. Authorized Dell partner since 2005. Pre-sales engineering, configuration assistance, and post-sales support.

Service and Support

• 24/7 technical support
• Pre-sales engineering and configuration assistance
• OEM warranty and full provenance documentation
• Installation, setup, and deployment of server equipment
• Component replacement and warranty service