NVIDIADGX B300
Enterprise-grade AI infrastructure platform built for the era of AI inference
Equipped with 8 NVIDIA Blackwell Ultra SXM GPUs, delivering up to 144 PFLOPS of FP4 inference performance and 2.1TB of total GPU memory. The all-new 10U chassis is the first to support NVIDIA MGX racks, and with full-stack AI software, it enables every enterprise to run AI workloads at the scale of hyperscale cloud providers.
Accelerate every businessAI Technology
The NVIDIA DGX B300 is an AI infrastructure solution purpose-built for generative AI workloads. Powered by NVIDIA Blackwell Ultra GPUs, it delivers 1.5x higher dense FP4 compute performance and 2x improved attention performance compared to the DGX B200. Its new form factor integrates seamlessly into modern data centers, streamlining and simplifying enterprise AI deployments.
The Era of ReasoningAI Factory
From real-time inference compute to energy-efficient design, from rack standards to full-stack software, DGX B300 delivers end-to-end capabilities for enterprise AI deployment.
Real-time compute power for LLM inference
Powered by 8 NVIDIA Blackwell Ultra SXM GPUs, this single platform accelerates large language model training and inference with up to 144 petaFLOPS of FP4 inference performance—seamlessly handling generative AI workloads.
Efficient and sustainable design
Supports both AC and DC power options with a focus on superior energy efficiency and watts-per-watt performance, helping enterprises control data center operating costs while deploying AI computing power.
The new rack-level infrastructure standard
Featuring a new 10U form factor, it is the first to support deployment in NVIDIA MGX racks, seamlessly integrating into modern data centers and offering flexible deployment options for scalable AI factory expansion.
Full-stack AI software, ready to use out of the box
Pre-installed with NVIDIA AI Enterprise software suite, Mission Control operations platform powered by Run:ai technology, and DGX OS operating system. Compatible with Red Hat Enterprise Linux, Rocky, and Ubuntu.
NVIDIA DGX B300Technical Specifications
Complete parameter overview. Contact our technical consultant for selection advice.
| GPU | 8x NVIDIA Blackwell Ultra SXM |
| CPU | Intel Xeon 6776P Processor |
| Total GPU Memory | 2.1 TB |
| Performance | FP4 Tensor Core: 144 PFLOPS (sparse) | 108 PFLOPS (dense)* FP8 Tensor Core: 72 PFLOPS (sparse) ** |
| NVLink Switch System | 2x |
| NVLink Bandwidth | 14.4 TB/s aggregated bandwidth |
| Network | 8 OSFP ports corresponding to 8 single-port NVIDIA ConnectX-8 VPI Supports up to 800 Gb/s NVIDIA InfiniBand/Ethernet 2 dual-port QSFP112 NVIDIA BlueField-3 DPUs Supports up to 400 Gb/s NVIDIA InfiniBand/Ethernet |
| Manage Network | 1GbE onboard network interface card (NIC), RJ45 port 1GbE RJ45 Host Baseboard Management Controller (BMC) |
| Storage | Operating System: 2x 1.9 TB NVMe M.2 Internal Storage: 8x 3.84 TB NVMe E1.S |
| Power Consumption | Approximately 14 kW |
| Software | NVIDIA AI Enterprise — Optimized AI Software Suite NVIDIA Mission Control — AI Data Center Operations and Collaboration Platform Powered by NVIDIA Run:ai NVIDIA DGX OS — Operating System Supports Red Hat Enterprise Linux, Rocky, and Ubuntu |
| Rack Unit | 10U |
| Support | 3-Year Business Standard Hardware and Software Support Services |
* Specifications are displayed in sparse and dense modes. ** represents the sparse mode; dense mode has half the specifications of sparse mode. Specification parameters are sourced from the official NVIDIA product page; actual configurations are subject to factory settings.
Product BrochureDownload
Preview of the NVIDIA DGX B300 product brochure. Click the button to generate and download the PDF for internal review and selection archiving.
DGX B300 features 8 NVIDIA Blackwell Ultra SXM GPUs, delivering up to 144 PFLOPS of FP4 inference performance and 2.1 TB of total GPU memory. It offers 1.5× denser FP4 compute performance and 2× faster attention performance compared to DGX B200. The new 10U chassis is the first to support deployment on NVIDIA MGX racks, with AC/DC power options and pre-installed NVIDIA AI Enterprise software stack.
| GPU | 8x NVIDIA Blackwell Ultra SXM |
| CPU | Intel Xeon 6776P Processor |
| Total GPU Memory | 2.1 TB |
| FP4 Performance | 144 PFLOPS (sparse) | 108 PFLOPS (dense) |
| FP8 Performance | 72 PFLOPS (sparse) |
| NVLink Bandwidth | 14.4 TB/s aggregated bandwidth |
| Network Computing | 8x ConnectX-8 VPI (up to 800 Gb/s) |
| Storage | 2x 1 9.0TB M.2 + 8x 3.84TB E1.S NVMe |
| Power / Rack Unit | Approx. 14 kW / 10U |
| Software | NVIDIA AI Enterprise / Mission Control / DGX OS |
Brochures are generated in real-time from page content. If your browser blocks the download, allow pop-ups and try again.
Ready to deploy DGX B300?
Huaying Technology provides end-to-end professional services for DGX B300 configuration consulting, rack planning, and deployment implementation.
Purchase Intent
Submit your requirements and contact details. A sales consultant will confirm pricing, stock availability, and delivery options with you. Product prices are subject to quotation.