Supermicro: NVIDIA HGX B300 GPU Server

SYS-822GS-NB3RT 8U

Supermicro: NVIDIA HGX B300 GPU Server

Pre-Configured Specs
  • 2x Intel Xeon 6767P 64C
  • 32x 64GB DDR5 ECC REG RDIMM
  • 8x NVIDIA HGX B300 288GB GPU
  • 2x 960GB M.2 NVMe SSD
  • 8x 3.84TB 2.5" U.2 NVMe SSD
  • 1x ConnectX-6 100GBe NIC
  • Add On: 3-Year NBD Onsite Warranty & Support
Starting at
$463,999 USD

Powered by Supermicro

The Supermicro HGX B300 GPU Server, available in a robust 8U chassis, is designed to maximize GPU density and performance for the most demanding AI and HPC workloads. Built on NVIDIA Blackwell HGX technology, Supermicro delivers unmatched scalability and energy efficiency for large-scale training, inference, and simulation.

Key advantages of the Supermicro HGX B300 GPU Server include:

  • Proven enterprise reliability – Backed by Supermicro’s track record in delivering mission-critical systems worldwide.
  • Optimized Blackwell performance – Engineered to fully leverage NVIDIA HGX B300 architecture for LLM training, high-throughput inference, and advanced AI pipelines.
  • Flexible deployment – Configurable 10U chassis options enable seamless integration into existing clusters and data center infrastructure.

By combining Supermicro’s engineering innovation with Arc Compute’s deployment expertise, customers gain a powerful and future-ready GPU platform designed to accelerate discovery and innovation.

Ideal Workloads for HGX B300 Servers

NVIDIA HGX B300

LLM Training at Scale

Fifth-gen NVLink fabric removes bottlenecks for multi-trillion parameter pre-training and continual training, delivering faster time-to-accuracy for models.

Inference and RAG

Blackwell Ultra GPUs enable low-latency, large-scale inference, powering retrieval-augmented generation and production AI deployments.

Agentic AI and Reasoning

Optimized for reasoning-heavy pipelines, the HGX B300 supports autonomous agents and advanced decision-making models at scale.

ModelSuperServer SYS-822GS-NB3RT
Form Factor8U rack server
ProcessorDual Socket E2 (LGA-4710) 2x Intel® Xeon® 6700/6500 series processors with P-cores Up to 128C/256T; Up to 504MB Cache per CPU
GPUNVIDIA Blackwell 8-GPU, TDP up to 1200W (HGX B300)
MemorySlot Count: 32 DIMM slots Max Memory (2DPC): Up to 8TB 6400MT/s ECC DDR5 RDIMM
StorageDefault: Total 8 bays 8 front hot-swap E1.S NVMe drive bays M.2: 2 M.2 NVMe slots (M-key; RAID support via S3808N controller)
CoolingAir cooled Liquid cooled options available
Fans12 heavy duty fans with optimal fan speed control
Power Supply6x 6600W Redundant (3 + 3) Titanium Level (96%) power supplies
DimensionsEnclosure: 449 x 356 x 950mm (17.6" x 13.8" x 37.4")
Warranty & Support3-Year Next-Business-Day Onsite Warranty & Support

Trusted Across Industries

2x Aivres HGX B200 GPU systems deployed in Secaucus, NJ
2x Aivres HGX B200 GPU systems deployed in Secaucus, NJ
“Arc Compute made the entire process smooth, from quoting to delivery. Their team was responsive and knowledgeable and helped us get the right configuration. The delivery and support have been excellent end-to-end."
Samuel Vasilevskiy
Datacenter Engineer
65x Supermicro HGX H100 GPU systems deployed in Barrie, ON
65x Supermicro HGX H100 GPU systems deployed in Barrie, ON
“Arc Compute helped us stand up our infrastructure faster than we thought possible and support has been outstanding from day one.”
Alex Smola
CEO
TRUSTED ACROSS INDUSTRIES
Does Not Have Feature

Request a Quote

Submit this form and a member of our team will be in touch.
Prefer to speak right away?  Book a call with our team.
By providing you email address, you agree to the Arc Compute Terms of Service/Privacy Policy.
Does Have Feature
Submission Successful

A member of our team will be in touch with you shortly. If you have any questions in the meantime, feel free to reach out to us at sales@arccompute.io.

Oops! Something went wrong while submitting the form.