Supermicro AS-8126GS-NB3RT-01-G2 AI GPU Server | Dual AMD EPYC 9575F | 8x NVIDIA HGX B300 GPUs | 3TB DDR5 | 800GbE Networking
Supermicro AS-8126GS-NB3RT-01-G2 AI GPU Server
8x NVIDIA HGX B300 GPUs with Dual AMD EPYC 9575F Processors for AI Training and HPC
Note: New Supermicro product with a 3-year warranty. Available exclusively for U.S. end users. Limited quantities are available for immediate shipment.
The Supermicro AS-8126GS-NB3RT-01-G2 is a next-generation AI GPU server purpose-built for large-scale artificial intelligence, machine learning, generative AI, and high-performance computing workloads. Featuring 8 NVIDIA HGX B300 GPUs, dual AMD EPYC 9575F processors, and 3TB DDR5-6400 memory, this platform delivers exceptional compute density for enterprise AI factories, research institutions, and cloud-scale infrastructure.
Powered by 256 total AMD EPYC cores running at 3.3GHz, the server provides the processing performance required to support large language model (LLM) training, deep learning frameworks, scientific simulations, and advanced analytics. The integrated NVIDIA HGX B300 platform leverages NVIDIA Blackwell architecture and high-speed NVLink connectivity to accelerate AI model training and inference at scale.
For data-intensive workloads, the system includes 8× 7.6TB E1.S NVMe SSDs, providing 60.8TB of high-performance storage, alongside 2× 1.9TB M.2 NVMe SSDs for operating system and application deployment. Ultra-fast networking is delivered through an 8-port 800GbE/XDR800 fabric and 4 additional 200GbE/NDR200 ports, enabling seamless communication across AI clusters and supercomputing environments.
Designed for continuous operation in enterprise data centers, the AS-8126GS-NB3RT-01-G2 incorporates redundant 6600W Titanium-level power supplies, advanced cooling, ECC memory protection, and enterprise-grade management capabilities. The platform supports leading AI software frameworks including NVIDIA CUDA, TensorRT, NCCL, RAPIDS, and NVIDIA AI Enterprise.
Product Specifications – Supermicro AS-8126GS-NB3RT-01-G2 AI GPU Server
CLICK HERE FOR DATA SHEET
| Specification Category | Details |
|---|---|
| Product Name | Supermicro AS-8126GS-NB3RT-01-G2 |
| Manufacturer | Supermicro |
| Product Type | AI Training & High-Performance Computing (HPC) GPU Server |
| Form Factor | 8U Rackmount GPU Server |
| System Architecture | NVIDIA HGX B300 Platform |
| Processor Configuration | Dual-Socket Server |
| CPU Model | 2 × AMD EPYC™ 9575F Processors |
| CPU Cores | 128 Cores per CPU (256 Total Cores) |
| CPU Base Frequency | 3.3 GHz |
| Processor Architecture | AMD EPYC 9005 Series (Turin) |
| GPU Configuration | 8 × NVIDIA HGX™ B300 GPUs |
| GPU Architecture | NVIDIA Blackwell Architecture |
| GPU Interconnect | NVIDIA NVLink High-Speed GPU Fabric |
| System Memory | 3 TB DDR5-6400 ECC Registered Memory |
| Memory Technology | DDR5 RDIMM |
| Memory Speed | 6400 MT/s |
| Memory Protection | ECC (Error-Correcting Code) |
| Primary Storage | 8 × 7.6TB E1.S NVMe SSDs |
| Storage Capacity | 60.8TB Raw NVMe Storage |
| Storage Interface | PCIe NVMe |
| Boot Drives | 2 × 1.9TB M.2 NVMe SSDs |
| M.2 Storage Capacity | 3.8TB Total |
| Networking Fabric | 8-Port 800GbE / XDR800 High-Speed Interconnect |
| Network Speed | Up to 800Gbps per port |
| Additional Networking | 2 × Dual-Port 200GbE / NDR200 Network Adapters |
| Additional Network Ports | 4 × 200GbE Ports |
| Cluster Connectivity | Optimized for AI Factories and Multi-Node GPU Clusters |
| Power Supply Configuration | Redundant Power Supplies |
| Power Rating | 6600W Titanium-Level Efficiency |
| Power Redundancy | Yes |
| Cooling System | High-Performance Redundant Cooling Architecture |
| Remote Management | Enterprise Remote Server Management Support |
| Security Features | Secure Boot, TPM Support, Hardware Root of Trust |
| Expansion Architecture | NVIDIA HGX GPU Baseboard with High-Speed PCIe Infrastructure |
| AI Framework Support | NVIDIA CUDA, TensorRT, NCCL, RAPIDS, NVIDIA AI Enterprise |
| Operating System Support | Enterprise Linux Distributions and HPC Software Stacks |
| Deployment Environment | AI Data Centers, Research Labs, Cloud AI Infrastructure |
| Target Workloads | Large Language Models (LLMs), Generative AI, Deep Learning, AI Training, AI Inference, HPC, Scientific Simulation |
| Reliability Features | Redundant Power, ECC Memory, Enterprise-Class Components |
| Storage Use Cases | AI Datasets, Model Checkpoints, High-Speed Data Pipelines, Large-Scale Analytics |
| Network Use Cases | Distributed AI Training, Multi-Node Clusters, AI Supercomputing, HPC Fabrics |
FAQs
1. How many GPUs does the AS-8126GS-NB3RT-01-G2 include?
It features 8 NVIDIA HGX B300 GPUs for AI training, inference, and HPC workloads.
2. Which processors power this server?
The system is equipped with 2 AMD EPYC 9575F processors, providing 256 total CPU cores.
3. How much memory is installed?
The server includes 3TB DDR5-6400 ECC memory for memory-intensive AI and data analytics workloads.
4. What storage configuration does it offer?
It comes with 8× 7.6TB E1.S NVMe SSDs and 2× 1.9TB M.2 NVMe SSDs for high-speed storage and boot functionality.
5. What workloads is this server designed for?
It is optimized for LLM training, generative AI, machine learning, deep learning, AI inference, and high-performance computing (HPC).
