NVIDIA HGX B300 eight-GPU platform with 2.1 TB HBM3e, fifth-generation NVLink, and 1.6 TB/s platform networking for OEM AI servers.
Fast Shipping
100k+ SKUs
8/5 Support
Certified
NVIDIA HGX B300 8-GPU AI Platform | HGX-B300
NVIDIA HGX B300 8-GPU AI Platform (HGX-B300)
NVIDIA HGX B300 is an eight-GPU HGX platform built for OEM and system-integrator AI servers. It combines 8x NVIDIA Blackwell Ultra SXM GPUs on a high-bandwidth NVLink baseboard and provides a standardized accelerated-compute foundation that server manufacturers pair with host CPUs, system memory, networking, storage, power, and cooling.
Availability / RFQ: NVIDIA lists HGX B300 as shipping now through the HGX partner ecosystem. Exact server SKU, CPU platform, thermals, support, and lead time depend on the OEM implementation. Contact Catalyst Data Solutions to discuss current status, compatibility, configuration requirements, and sourcing options.
End-user / export compliance notice: Advanced AI accelerators and systems may be subject to applicable U.S. export controls, end-user/end-use restrictions, destination requirements, and OEM/channel policies. Catalyst may request purchaser, final end-user, intended-use, and deployment-location information before a quote or fulfillment can be approved. An RFQ does not guarantee allocation or shipment.
Technical Specifications
| SKU | HGX-B300 |
|---|---|
| Product Type | HGX 8-GPU AI Platform |
| Architecture | NVIDIA Blackwell Ultra |
| GPU Configuration | 8x NVIDIA Blackwell Ultra SXM GPUs |
| Total GPU Memory | 2.1 TB HBM3e |
| NVLink | Fifth-generation NVIDIA NVLink |
| Total NVLink Bandwidth | 14.4 TB/s |
| Platform Networking Bandwidth | 1.6 TB/s platform networking |
| Deployment Model | OEM HGX server integration |
| Availability Status | Available / RFQ |
Deployment and Compatibility Considerations
- Select the OEM server first because HGX is the GPU platform/baseboard, not a complete independent server.
- Confirm host CPU, system memory, networking, local storage, power, firmware, cooling, and service requirements with the OEM.
- Validate scale-out network design, optics, cable map, and switch fabric for the intended cluster size.
- Plan AI storage throughput and checkpoint/data paths around training or inference workload requirements.
Common Use Cases
- OEM AI servers
- Large-model training
- Reasoning and inference
- HPC and accelerated analytics
- Enterprise and cloud AI factories
Frequently Paired Infrastructure
- Validated HGX OEM server platforms
- NVIDIA ConnectX networking
- NVIDIA BlueField DPUs
- Spectrum-X or Quantum InfiniBand
- AI storage and high-density rack infrastructure
Official NVIDIA Documentation
Frequently Asked Questions
Is NVIDIA HGX B300 a complete server?
No. NVIDIA HGX B300 is the NVIDIA eight-GPU platform that OEMs integrate into complete servers with host CPUs, memory, storage, networking, cooling, power, and management.
Is HGX B300 shipping?
Yes, Contact Catalyst Data Solutions to explore your deployment options and lead time on HGX B300.
| product type | HGX GPU Platform |
|---|---|
| Platform / Architecture | NVIDIA Blackwell Ultra |
| Accelerator Configuration | 8x NVIDIA Blackwell Ultra SXM GPUs |
| GPU Memory | 2.1 TB HBM3e |
| Availability Status | Available / RFQ |
