Uvation MarketplaceMarketplace

NVIDIA DGX B300
AI Server

An air-cooled 10U AI system with eight NVIDIA Blackwell Ultra GPUs, dual Intel Xeon Platinum 6776P processors and 2.1 TB of GPU memory.

8×
Blackwell Ultra GPUs
2.1 TB
GPU memory
800
Gb/s Per OSFP port

Built for the Era of Enterprise AI

NVIDIA DGX B300 combines Blackwell Ultra GPU compute, high-speed networking, and integrated AI software to support large-scale training and inference in modern data centers.

An AI Factory for the Era of Reasoning

Eight Blackwell Ultra GPUs linked by fifth-generation NVLink at 14.4 TB/s give enterprises one platform for LLM training and inference.

Flexible AC or DC Power Options

Choose an AC or DC version of the air-cooled chassis, powered by twelve 3.2 kW power supplies.

Fast Local Storage and Secure Drives

Eight 3.84 TB NVMe E1.S drives with self-encrypting cache, plus two 1.92 TB M.2 NVMe boot drives,
keep data close.

Integrated Blackwell Ultra AI Architecture

Eight Blackwell Ultra GPUs and dual Intel Xeon processors connect through fifth-generation NVLink with 14.4 TB/s of aggregate bandwidth.

NVIDIA DGX B300 AI Server architecture
8×
Blackwell Ultra SXM GPUs
2.1 TB
GPU memory
14.4 TB/s
Aggregate NVLink bandwidth
4 TB
Max system memory

Workloads

Run demanding enterprise AI workloads on a unified platform built for large-model training, high-throughput inference, and reasoning-intensive applications.

Large Language Model Training and Fine-Tuning

2.1 TB of GPU memory and 14.4 TB/s of NVLink bandwidth help teams train and fine-tune large language models.

Generative AI and Large-Scale Inference

Eight Blackwell Ultra GPUs and up to 800 Gb/s per OSFP port serve generative AI workloads at scale.

Data-Intensive AI Pipelines

Up to 4 TB of system memory and eight 3.84 TB NVMe E1.S drives keep large datasets close to compute.

Engineered to dominate

Engineered for High-Performance Enterprise AI

DGX B300 combines Blackwell Ultra GPU compute, 2.1 TB of GPU memory and 800 Gb/s networking in an air-cooled 10U system.

Exploded view of the NVIDIA DGX B300 AI Server
2×
Intel Xeon Platinum 6776P
12× 3.2 kW
Power supplies
2×
Dual-port BlueField-3 DPUs
800 Gb/s
Per ConnectX-8 OSFP port
Exploded view of the NVIDIA DGX B300 AI Server
2×
Intel Xeon Platinum 6776P
12× 3.2 kW
Power supplies
2×
Dual-port BlueField-3 DPUs
800 Gb/s
Per ConnectX-8 OSFP port

The rack, itemized

SUBSYSTEM

SPECIFICATION / CONFIGURATION DETAILS

Processor

36x NVIDIA Grace CPUs; 2,592 Arm Neoverse V2 cores

Operating System

NVIDIA DGX OS / Ubuntu

Chipset

NVIDIA Grace Blackwell Ultra Superchips

Accelerators / GPU

72x NVIDIA Blackwell Ultra GPUs

Memory

37 TB Total Fast Memory; 20 TB GPU Memory

Primary Storage

1× 1.92 TB M.2 NVMe boot drive per compute tray

Secondary Storage

4 × 3.84 TB E1.S NVMe RAID 0 per compute tray

Security

BMC with Redfish/IPMI; self-encrypting E1.S drives

Chassis

Liquid-cooled NVIDIA MGX rack-scale system

Power Supply

8 power shelves per rack, each with six 5.5 kW PSUs (up to 33 kW per shelf)

Ports

ConnectX-8 + BlueField-3 + BMC RJ45 connectivity

Slots / PCIe Architecture

9 × L1 NVIDIA NVLink switch systems; fifth-generation NVLink

Bezel

NA

Form Factor

Rack-scale, liquid-cooled

Network Card

8× single-port NVIDIA ConnectX-8 VPI adapters + 2× dual-port NVIDIA BlueField-3 DPUs

Warranty

3-year business-standard hardware and software support

NVIDIA DGX B300 AI Server power and networking

Power and Networking for
Enterprise AI

12 × 3.2 kW power supplies

Twelve 3.2 kW power supplies feed the air-cooled 10U system, available in AC or DC versions.

8× ConnectX-8 VPI at up to 800 Gb/s

Eight OSFP ports with ConnectX-8 VPI adapters deliver up to 800 Gb/s each for cluster networking.

2× dual-port BlueField-3 DPUs

Two dual-port BlueField-3 DPUs connect over QSFP112, alongside onboard 1GbE and two 1GbE RJ45 ports.

Nuclear-Ready, Plug and Play AI Factories

Bypass grid constraints with self-contained, high-density inference ready infrastructure directly adjacent to clean, base load nuclear power.

Global Deployment Engine

Any Infra, Anywhere in the world.

Global Shipping and Deployment Capability treats advanced infrastructure as mobile, modular freight.

Packaged complex computing nodes, cooling manifolds, and power distribution units into certified, standard form factors.

360° Service Ecosystem

Unifying Legacy with Next-Gen Compute.

Vertically integrated service ecosystem that serves as an operational bridge between legacy environments and next-generation, AI-native infrastructure.

Finance and Leasing

Capital-as-a-Service for High-Density Compute

Convert upfront capital expenditures (CapEx) into predictable operating expenses (OpEx).

Uvation allows enterprises and sovereign entities to deploy multi-megawatt AI factories and advanced GPU clusters without liquidity constraint.