Uvation MarketplaceMarketplace

Supermicro SYS-422GS-NB3RT-ALC GPU Server

Compact 4U liquid-cooled server uniting eight NVIDIA Blackwell Ultra GPUs with dual Intel Xeon 6700-series processors for AI training, inference and HPC.

GPUs
8x
GPUs
HBM3e
2.3 TB
HBM3e
Max DDR5
8 TB
Max DDR5

Intel Xeon 6 and HGX B300,
Liquid-Cooled

Supermicro’s SYS-422GS-NB3RT-ALC combines eight Blackwell Ultra GPUs and dual Intel Xeon 6700-series processors in a direct-to-chip liquid-cooled 4U rackmount chassis.

86 P-Cores per Socket, Liquid-Cooled

Each Intel Xeon 6700-series processor offers up to 86 cores, 172 threads and 336 MB cache, cooled at up to 350 W TDP.

Choose Speed or Capacity in Memory

32 DDR5 ECC RDIMM slots deliver 4 TB at 6400 MT/s, or 8 TB at 6000 MT/s with 2DPC.

Fast Local Data on E1.S NVMe

Eight front hot-swap E1.S NVMe bays and two M.2 NVMe boot slots with S3808N RAID keep data close.

One NVSwitch Fabric Linking Eight GPUs

Fifth-generation NVLink with NVSwitch connects all eight Blackwell Ultra GPUs at up to 1.8 TB/s GPU-to-GPU bandwidth.

Supermicro SYS-422GS-NB3RT-ALC 4U chassis in a data center aisle
8×
HGX B300 SXM GPUs
288 GB
HBM3e per GPU
1.8 TB/s
GPU-to-GPU NVLink
4U / D2C
Cold-plate liquid cooled

Workloads

Designed for LLMs, generative AI, deep learning and scientific computing in a dense, liquid-cooled 4U footprint.

LLM and Generative AI Development

Eight Blackwell Ultra GPUs and 2.3 TB of HBM3e give large language and generative models room to train and run.

Deep Learning Across GPU Clusters

Eight 800 Gb/s ConnectX-8 SuperNICs with InfiniBand OSFP ports extend training and inference across scalable multi-node clusters.

HPC and Scientific Workloads

Dual Xeon 6700-series CPUs with up to 86 cores each and 8 TB of DDR5 drive simulation and research computing.

Engineered to dominate

Built for Sustained, High-Density AI

Direct-to-chip cold plates cool the CPUs and HGX B300 platform, keeping eight GPUs and dual Xeon processors working inside 4U.

Cutaway view of the liquid-cooled SYS-422GS-NB3RT-ALC chassis
336 MB
Cache per CPU
6400 MT/s
DDR5 speed at 1DPC
96%
Titanium PSU efficiency
800 Gb/s
8× ConnectX-8 InfiniBand
Cutaway view of the liquid-cooled SYS-422GS-NB3RT-ALC chassis
336 MB
Cache per CPU
6400 MT/s
DDR5 speed at 1DPC
96%
Titanium PSU efficiency
800 Gb/s
8× ConnectX-8 InfiniBand

The rack, itemized.

SUBSYSTEM

SPECIFICATION / CONFIGURATION DETAILS

Processor

2× Intel Xeon 6700P (up to 86 cores and 172 threads per processor)

Operating System

Ubuntu 24.04.3, RHEL 9.4, Rocky Linux 9.4, Oracle Linux 9.4, SLES 15 SP6, VMware ESXi 9.0

Chipset

System on Chip (SoC)

Accelerators / GPU

NVIDIA SXM: HGX B300 8-GPU (288GB HBM3e per GPU, ~2.3 TB total HBM3e)

Memory

32 DDR5 ECC RDIMM slots; up to 4 TB (at 6400 MT/s) or 8 TB (at 6000 MT/s)

Primary Storage

8 × front hot-swap E1.S NVMe drive bays

Secondary Storage

2 × M.2 NVMe boot drives

Security

Trusted Platform Module (TPM) 2.0; Silicon Root of Trust (RoT) – NIST 800-193 Compliant

Chassis

CSE-GP405TS-R000NP2

Power Supply

4x 6600W Redundant (2 + 2) Titanium Level (96%) power supplies

Ports

8× OSFP; 2× 10GbE; USB; VGA; Mini DisplayPort, TPM Header

Slots / PCIe Architecture

2 PCIe 5.0 x16 FHHL double-width slots; 2 PCIe 5.0 x16 LP slots

Bezel

Not specified (separate front security bezel not specified by Supermicro)

Form Factor

4U Rackmount

Network Card

8x NVIDIA ConnectX®-8 SuperNICs (up to 800Gb/s)

Warranty

3-Year Labor, 1-Year Parts & 1-Year Advance Cross-Shipment Limited Manufacturer Warranty

Supermicro SYS-422GS-NB3RT-ALC 4U server

Titanium Power and
800G InfiniBand Fabric

Four 6,600 W supplies in 2+2

Redundant Titanium-level power supplies run at up to 96% efficiency to power eight GPUs and dual CPUs.

Eight 800 Gb/s InfiniBand OSFP ports

Eight ConnectX-8 SuperNICs provide high-throughput InfiniBand links for scaling AI clusters, alongside dual 10GbE LAN.

Four PCIe 5.0 x16 expansion slots

Two full-height double-width and two low-profile PCIe 5.0 x16 slots leave room for more networking or storage cards.

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.