Uvation MarketplaceMarketplace

NVIDIA DGX Rubin NVL8

A liquid-cooled 2RU DGX system pairing eight NVIDIA Rubin GPUs and 2.3 TB of HBM4 memory with dual Intel Xeon 6776P processors.

8×
NVIDIA Rubin GPUs/system
2x
Intel Xeon 6776P CPUs/system
2.3 TB
Total HBM4 GPU memory/system

Eight Rubin GPUs in a 2RU System

NVIDIA DGX Rubin NVL8 brings eight Rubin GPUs, sixth-generation NVLink
and dual Intel Xeon 6776P processors together in one liquid-cooled
2RU rack system.

Sixth-Generation NVLink Across Eight GPUs

Eight NVIDIA Rubin GPUs connect through sixth-generation NVLink, sharing 2.3 TB of HBM4 GPU memory for large AI models.

Dense Compute in Just Two Rack Units

A liquid-cooled 2RU chassis, powered through a DC busbar at about 24 kW, packs eight-GPU DGX compute into a compact footprint.

ConnectX-9 and BlueField-4 Networking Built In

Eight ConnectX-9 VPI adapters on OSFP ports and a BlueField-4 DPU on 400G QSFP112 ports link the system to AI clusters.

One Sixth-Gen NVLink Fabric,
Eight Rubin GPUs

Sixth-generation NVLink connects all eight Rubin GPUs, giving workloads access to 2.3 TB of HBM4 GPU memory.

NVIDIA DGX Rubin NVL8 eight-GPU system
8×
NVIDIA Rubin GPUs
2.3 TB
HBM4 GPU memory
6th Gen
NVLink interconnect
2×
Intel Xeon 6776P CPUs

Workloads

Train, fine-tune and serve large AI models on a compact, liquid-cooled DGX platform built around Rubin GPUs.

Large-Scale Model Training

Eight Rubin GPUs with 2.3 TB of HBM4 memory and sixth-generation NVLink give training jobs room for larger models.

High-Throughput AI Inference

Eight Rubin GPUs and eight ConnectX-9 VPI adapters serve large models across nodes for production inference.

Data-Heavy AI Pipelines

Eight 4 TB NVMe E1.S data drives and two 2 TB M.2 OS drives keep datasets close to the GPUs.

Engineered to dominate

Rubin-Class Compute in 2RU

DGX Rubin NVL8 combines eight Rubin GPUs, 2.3 TB of HBM4, ConnectX-9 networking and liquid cooling in a 2RU footprint.

Exploded view of the NVIDIA DGX Rubin NVL8 liquid-cooled system
~24 kW
System power (DC busbar)
8×
ConnectX-9 VPI adapters
2× 400G
BlueField-4 QSFP112 ports
8× 4 TB
NVMe E1.S data drives
Exploded view of the NVIDIA DGX Rubin NVL8 liquid-cooled system
~24 kW
System power (DC busbar)
8×
ConnectX-9 VPI adapters
2× 400G
BlueField-4 QSFP112 ports
8× 4 TB
NVMe E1.S data drives

The rack, itemized.

SUBSYSTEM

SPECIFICATION / CONFIGURATION DETAILS

Processor

2× Intel Xeon 6776P

Operating System

NVIDIA DGX OS; Ubuntu; Red Hat Enterprise Linux; Rocky Linux

Chipset

N/A

Accelerators / GPU

8× NVIDIA Rubin GPUs; sixth-generation NVLink

Memory

2.3 TB HBM4 GPU memory; system RAM: Information Not Available

Primary Storage

2× 2 TB NVMe M.2 OS drives

Secondary Storage

8× 4 TB NVMe E1.S data drives

Security

BMC and NVIDIA BlueField-4 infrastructure security; exact options N/S

Chassis

Liquid-cooled 2RU DGX system

Power Supply

DC busbar; approximately 24 kW system power

Ports

8× OSFP for ConnectX-9; 2× 400G QSFP112 for BlueField-4; 1GbE BMC RJ45; 1GbE onboard RJ45

Slots / PCIe Architecture

6th-generation NVLink; 4 NVLink switch systems (NVIDIA product page)

Bezel

N/A

Form Factor

2RU rack system

Network Card

8× ConnectX-9 VPI; BlueField-4 DPU

Warranty

Three-year business-standard hardware and software support

NVIDIA DGX Rubin NVL8 front view

Busbar Power and
ConnectX-9 Scale-Out

DC busbar, about 24 kW

Draws approximately 24 kW of power through a DC busbar

8× ConnectX-9 VPI on OSFP ports

Scale-out networking for multi-node AI clusters.

BlueField-4 DPU with 2× 400G QSFP112

Adds data and management connectivity

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.