SYS-822GS-NB3RT
8-GPU NVIDIA HGX B300 server for LLM training and AI infrastructure
2.3 TB HBM3e GPU memory per system
8-GPU design for intensive training and accelerated HPC
Server Products
GPU Servers
GPU servers are high power-density systems for AI training, LLM, inference and accelerated HPC workloads.
Portfolio4 models
FocusAI server
ApproachCustomizable enterprise configuration
GPU Servers
GPU server platforms are optimized for LLM training, fine-tuning and accelerated HPC scenarios with NVIDIA HGX, Grace Hopper and multi-GPU configurations.
Server Selection Criteria
GPU count and GPU-to-GPU bandwidth
CPU-to-GPU ratio
Power and cooling profile
Model size and data flow
Key Use Cases
AI server
LLM training server
Inference server
GPU-accelerated HPC
Image processing and analytics
Server Architecture and Evaluation Topics
This page does not only list products; it also brings together the topics that influence proper server sizing, system architecture and enterprise infrastructure decisions.
Capacity Planning
CPU, memory, expansion capacity and storage tiers are evaluated together based on the workload.
Rack Layout and Power Density
Power density, cooling requirements and cabling layout are made visible from the beginning.
Network Architecture and Integration
Network selection is not treated separately from the server; it is evaluated together with the full solution.
Operations and Deployment
Serviceability, spare-parts approach and deployment process are all part of product selection.
Server Model List
Full product catalog →
AS-8126GS-NB3RT
8-GPU NVIDIA HGX B300 server for AMD-based AI infrastructures
AMD-based 8U dense GPU server architecture
Multi-GPU scale for large-model training
SYS-821GE-TNHR
8-GPU NVIDIA HGX H100/H200 server for AI training and accelerated HPC
HGX H100/H200-based 8-GPU configuration
Enterprise 8U design for AI training and accelerated HPC
AS-4125GS-TNRT2
4U multi-GPU server for AI labs, inference and development environments
4U dense GPU layout and flexible expansion
Balanced platform for AI labs and inference clusters
