Uvation MarketplaceMarketplace

NVIDIA DGX SuperPOD with DGX Rubin NVL8 Systems - 72-System Scalable Unit

NVIDIA DGX SuperPOD with DGX Rubin NVL8 Systems combines 72 DGX systems in a multi-rack deployment built for large-scale AI computing. Each system integrates eight Rubin GPUs, high-speed networking, and NVMe storage.

4x
NVIDIA Rubin GPUs per module
2x
NVIDIA Vera CPUs
14.4
NVLink Bandwidth

Scale AI Across a Unified Infrastructure

Bring GPU compute, high-speed interconnects, and storage together in a scalable architecture designed for demanding AI workloads.

72-System Compute Foundation

Connect 72 DGX Rubin NVL8 systems, each equipped with eight Rubin GPUs, in one scalable deployment.

High-Bandwidth Fabric

Quantum-X800 InfiniBand supports the compute fabric, with ConnectX-9 and BlueField-4 providing system-level connectivity.

Integrated Storage Architecture

Combine local NVMe storage in each DGX system with certified partner shared storage for deployment-level capacity.

72 Systems, One Scalable AI Platform

Bring GPU compute, CPU processing, memory, storage, and high-speed interconnects together in a rack-scale AI platform designed for multi-rack deployment.

NVIDIA DGX SuperPOD with DGX Rubin NVL8 Systems - 72-System Scalable Unit rack architecture
8
Rubin GPUs per system
2×
Vera CPUs
2.3 TB
GPU memory per system
Quantum-X800
InfiniBand compute fabric

Workloads

Deploy a large-scale GPU infrastructure foundation for computationally intensive AI development, training, and inference.

Large-Scale Model Training

576 interconnected NVIDIA Rubin GPUs across 72 systems pool massive computing power and fast memory into a unified AI factory, allowing teams to build and train frontier models without running into communication delays.

Distributed AI Computing

Dual Intel processors per server, ultrafast network adapters, dedicated management chips, and NVIDIA Mission Control software keep all 72 systems synchronized, handling cluster coordination, quick data saves, and smooth operations.

High-Performance AI Inference

Generating over 167,000 response tokens per second and reading billions of prompt tokens per second across the cluster, the platform delivers instantaneous answers and scales up to support millions of simultaneous users.

Engineered to dominate

Built for Multi-Rack AI Scale

A 72-system deployment combining Rubin GPU compute, NVMe storage, and dedicated compute-fabric connectivity across a unified infrastructure.

Exploded view of the NVIDIA DGX SuperPOD with DGX Rubin NVL8 Systems - 72-System Scalable Unit
72
DGX Rubin NVL8 system
576
Rubin GPUs across 72 systems
2.3 TB
GPU memory per system
2 ×
Intel Xeon 6776P processors per system
Exploded view of the NVIDIA DGX SuperPOD with DGX Rubin NVL8 Systems - 72-System Scalable Unit
72
DGX Rubin NVL8 system
576
Rubin GPUs across 72 systems
2.3 TB
GPU memory per system
2 ×
Intel Xeon 6776P processors per system

The rack, itemized

SUBSYSTEM

SPECIFICATION / CONFIGURATION DETAILS

Processor

2× Intel Xeon 6776P per DGX system

Operating System

DGX OS; Ubuntu supported per DGX rack

Chipset

N/A

Accelerators / GPU

8× Rubin GPUs per DGX system; 72 systems per scalable unit

Memory

2.3 TB GPU memory per DGX system

Primary Storage

2× 2 TB NVMe M.2 per DGX system

Secondary Storage

8× 4 TB NVMe E1.S per DGX system; certified partner shared storage

Security

DGX BMC and BlueField-4 infrastructure security

Chassis

Multi-rack MGX-based deployment

Power Supply

DC-busbar compute racks with rack power shelves

Ports

ConnectX-9 compute-fabric and BlueField-4 storage/management connections per DGX system

Slots / PCIe Architecture

Per node 6th-generation NVLink; cluster fabric per RA

Bezel

N/A

Form Factor

Multi-rack solution; 72 DGX Rubin NVL8 systems per scalable unit

Network Card

ConnectX-9 and BlueField-4; Quantum-X800 InfiniBand compute fabric

Warranty

NVIDIA Enterprise Support; Contract-Dependent

NVIDIA DGX SuperPOD with DGX Rubin NVL8 Systems - 72-System Scalable Unit power and networking

Power Fabric

Approximately 24 kW per DGX system

Power usage per liquid-cooled DGX Rubin NVL8 server

Multi-rack solution

Organizing 72 DGX Rubin NVL8 systems into a single SuperPOD scalable unit

DC busbar compute racks

Supported by rack power shelves, busbars, and PDUs across the deployment

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.