Overview
AIME GX8-B300 is an 8x NVIDIA B300 Blackwell Ultra HGX server powered by dual AMD EPYC Turin processors. With 288GB HBM3e per GPU and NVLink interconnect, this system targets LLM training at the 100B+ parameter scale, scientific simulation, and datacenter AI inference for the most demanding enterprise workloads.
Key Features
- 8x NVIDIA B300 Blackwell Ultra GPUs (NVLink interconnect)
- Dual AMD EPYC Turin processors
- 288GB HBM3e memory per GPU
- HGX form factor with high-density GPU interconnect
- Enterprise-grade ECC memory and redundant PSU
Ideal For
Enterprises, AI research labs, and HPC centers building private cloud AI infrastructure for large language model training, scientific computing, or high-throughput enterprise inference.
Category
HGX Server
Contact for pricing
Prices may vary. Verify on vendor site.
Quick Specs
- Category
- HGX Server
Tags
model-trainingllm-inferencehpc