Products / SRV-04
GPU Servers · SRV-04

HGX H200 / B200 Node (8 HBM GPUs)

8-GPU node with HBM memory (1.1–1.5 TB) for foundation-model training and high-performance clusters.

On request

Estimated delivery: 60 days. Confirmed on order acceptance.

Request a proposal
Warranty and support includedTPV Virtual Banco Santander · Bank transfer

Overview

HGX platform with NVIDIA H200 or B200 GPUs connected over NVLink, for organisations that train or serve foundation models. Configuration, networking (400 Gb/s InfiniBand or Ethernet), storage and price are defined in a proposal. Subject to manufacturer availability and export-control regulations.

Included

  • Tailored technical and commercial proposal
  • 3-year manufacturer warranty
  • Installation and commissioning

Specifications

GPU8x NVIDIA H200 141 GB HBM3e o B200 180 GB HBM3e, NVLink
CPU2x Intel Xeon / AMD EPYC
Memory2–4 TB DDR5
Network8x 400 Gb/s InfiniBand o Ethernet
Form factor8U–10U rack, refrigeración por aire o líquida