WSG-GS8800 4U Eight-GPU AI Training Server

WSG-GS8800 4U Eight-GPU AI Training Server

Maximum accelerator density for large-model training

The WSG-GS8800 is a 4U rackmount server supporting up to eight double-width GPUs with NVIDIA NVSwitch-class interconnect. Engineered with enhanced cooling and redundant power, it delivers datacenter-class performance in an industrial-grade chassis for deep-learning training, digital twins and multi-camera vision workloads.

Key Features

  • Up to 8× double-width professional GPUs (NVIDIA L40S / H-class ready)
  • Dual 4th/5th Gen Intel Xeon Scalable processors
  • Up to 4TB DDR5 ECC memory
  • GPU-to-GPU high-speed interconnect architecture
  • 4+2 redundant 3000W titanium power supplies
  • Industrial-grade shock and vibration resistance

Key Specifications

Form Factor
4U rackmount
GPUs
Up to 8× double-width, up to 700W each
CPU
2× Intel Xeon Scalable, up to 64 cores each
Memory
Up to 4TB DDR5 ECC RDIMM
Storage
Up to 8× 2.5" hot-swap NVMe/SATA
Network
2× 10GbE + OCP 3.0 (up to 200G)
Power
4+2 × 3000W 80 PLUS Titanium

Applications

Large-model training and fine-tuning AI vision model training farms Factory digital-twin simulation
Request a Quote

Related Products

WSG-GS4200 2U Dual-GPU Inference Server

WSG-GS4200 2U Dual-GPU Inference Server

High-throughput inference for production AI
WSG-EG1100 Short-Depth Edge GPU Server

WSG-EG1100 Short-Depth Edge GPU Server

AI inference at the production edge
WSG-GW4000 Industrial GPU Workstation

WSG-GW4000 Industrial GPU Workstation

Desk-side computing for vision labs and engineers