Single Node H200X8 GPU Server | NVIDIA H200, AMD EPYC, Intel Xeon | Hyblox
Highlights
 Rack Height: |
8U |
| GPU Configuration: |
8x NVIDIA H200 SXM5 141GB HBM3e |
| Processor (Per Node): |
2x AMD EPYC 9754 (128-core) or Intel Xeon Platinum 8592+ (64-core) |
| Memory (Per Node): |
3TB DDR5-6000 RDIMM ECC (24x 128GB) |
Starting Price: Contact for Price
The H200X8 is a high-density 8U single-node server designed to accelerate AI and HPC workloads at scale. Supporting 8x NVIDIA H200 SXM5 GPUs, the system delivers exceptional memory bandwidth and compute throughput, making it ideal for large language model training, deep learning inference, and simulation-heavy environments.
Built for Enterprise AI and HPC Performance
Powered by either dual AMD EPYC 9754 with 128 cores or Intel Xeon Platinum 8592 Plus with 64 cores, the InferCore H200X8 offers exceptional flexibility and raw CPU performance. It supports up to 3TB of DDR5-6000 RDIMM ECC memory, features high-speed PCIe Gen5 NVMe storage, and is equipped with 400GbE NVIDIA ConnectX-7 networking, delivering an optimal solution for scalable AI infrastructure and modern data center deployments.
Key capabilities include:
- Optimized for AI training, generative AI inference, and LLM model development
- Accelerated performance with NVIDIA H200 GPUs and HBM3e memory
- Ideal for high-performance computing, scientific workloads, and AI research clusters
Supports multi-GPU communication and NVLink interconnects for massive parallelism
Processor & Chipset
Number of Nodes | 1 |
Number of Processors Supported (Per Node) | 2 |
Processor Options | AMD EPYC 9754 (128-core, 2.25GHz) or Intel Xeon Platinum 8592+ (64-core) |
Memory Support Per CPU | 12x DDR5 RDIMM Slots |
Processor Architecture | x86_64 |
Thermal Design Power (TDP) | Up to 360W (per CPU) |
Chipset Manufacturer | AMD / Intel |
Chipset Model | SoC |
Memory (Per Node)
Memory Technology | DDR5 SDRAM |
Memory Standard | DDR5-6000 ECC RDIMM |
Installed Memory | 3TB (24x 128GB) |
Number of Memory Slots | 24 |
GPU & Accelerators
GPU Model | NVIDIA H200 SXM5 |
GPU Memory | 141GB HBM3e per GPU |
Total GPUs | 8 |
Multi-GPU Support | NVLink and NVSwitch enabled |
Storage (Per Node)
Boot Drives | 2x Samsung PM1743 2TB PCIe Gen5 NVMe U.2 |
Model/Data Storage | 6x WD SN861 8TB PCIe Gen5 NVMe U.2 |
Drive Bays | 6x 2.5″ U.2 Hot-Swap |
Storage Interface | NVMe Gen5 |
RAID Support | Software RAID or VROC (optional) |
Networking (Per Node)
High-Speed Networking | 2x NVIDIA ConnectX-7 NDR 400GbE, OSFP |
Management NIC | 2x 10GbE Base-T onboard |
Ethernet Technology | PCIe Gen5 and OCP 3.0 compliant |
I/O Expansions (Per Node)
PCIe Slots | 1x PCIe 5.0 x16 FHHL slot, 1x PCIe 5.0 x16 OCP 3.0 slot |
M.2 Slots | 2x PCIe 5.0 x2 M.2 slots (2280/22110) |
Power Components
Power Supply | 6x 3000W Titanium PSU |
Power Cables | 6x C19 to C20, 2 meters |
Cooling & Environmental
Cooling System | 8x High-performance integrated fans |
Rack Mounting | 8U mounting rails with cable management kit |
Chassis | Supermicro AS-8125GS-TNHR |
About
Built for Enterprise AI and HPC Performance
- Optimized for AI training, generative AI inference, and LLM model development
- Accelerated performance with NVIDIA H200 GPUs and HBM3e memory
- Ideal for high-performance computing, scientific workloads, and AI research clusters
Specifications
Processor & Chipset
| Number of Nodes | 1 |
| Number of Processors Supported (Per Node) | 2 |
| Processor Options | AMD EPYC 9754 (128-core, 2.25GHz) or Intel Xeon Platinum 8592+ (64-core) |
| Memory Support Per CPU | 12x DDR5 RDIMM Slots |
| Processor Architecture | x86_64 |
| Thermal Design Power (TDP) | Up to 360W (per CPU) |
| Chipset Manufacturer | AMD / Intel |
| Chipset Model | SoC |
Memory (Per Node)
| Memory Technology | DDR5 SDRAM |
| Memory Standard | DDR5-6000 ECC RDIMM |
| Installed Memory | 3TB (24x 128GB) |
| Number of Memory Slots | 24 |
GPU & Accelerators
| GPU Model | NVIDIA H200 SXM5 |
| GPU Memory | 141GB HBM3e per GPU |
| Total GPUs | 8 |
| Multi-GPU Support | NVLink and NVSwitch enabled |
Storage (Per Node)
| Boot Drives | 2x Samsung PM1743 2TB PCIe Gen5 NVMe U.2 |
| Model/Data Storage | 6x WD SN861 8TB PCIe Gen5 NVMe U.2 |
| Drive Bays | 6x 2.5″ U.2 Hot-Swap |
| Storage Interface | NVMe Gen5 |
| RAID Support | Software RAID or VROC (optional) |
Networking (Per Node)
| High-Speed Networking | 2x NVIDIA ConnectX-7 NDR 400GbE, OSFP |
| Management NIC | 2x 10GbE Base-T onboard |
| Ethernet Technology | PCIe Gen5 and OCP 3.0 compliant |
I/O Expansions (Per Node)
| PCIe Slots | 1x PCIe 5.0 x16 FHHL slot, 1x PCIe 5.0 x16 OCP 3.0 slot |
| M.2 Slots | 2x PCIe 5.0 x2 M.2 slots (2280/22110) |
Power Components
| Power Supply | 6x 3000W Titanium PSU |
| Power Cables | 6x C19 to C20, 2 meters |
Cooling & Environmental
| Cooling System | 8x High-performance integrated fans |
| Rack Mounting | 8U mounting rails with cable management kit |
| Chassis | Supermicro AS-8125GS-TNHR |
