Skip to configurations
Zipcode / Infrastructure

Serious compute.
Your configuration.

Explore the hardware behind your next idea. Compare GPU systems and build a selection for your infrastructure discussion.

Explore configurations
Hardware, in perspective
From a single tower
to rack-scale compute
Illustrative schematic
10 configurations to exploreCompare. Select. Save.Presentation catalog · specifications subject to confirmation
Find your fit

Explore machines

Start with the platform, the form factor,
or the memory your workload needs.

10 configurations

8U rack serverIllustrative schematic
NVIDIA HGX platform server

NVIDIA HGX B300

8 × NVIDIA Blackwell Ultra SXM

GPU memory
GPU memory requires confirmation
System memory
24 × 96 GB DDR5-6400 ECC · 2,304 GB
GPUs included
Included
TrainingInference
Configuration details
CPU
2 × AMD EPYC 9555 · 64 cores each
Storage
2 × 2 TB M.2 PCIe 4.0; 8 × 3.84 TB U.2/U.3 PCIe 5.0
Networking
8 × ConnectX-8 integrated 800G NICs; dual 1GbE plus management
Power
12 × 3,000 W Titanium PSUs · 6+6 redundancy · 200–240 VAC
  • Storage drive layout and RAID configuration require confirmation.
  • AI Enterprise is not included.

Memory across multiple GPUs is aggregate capacity, not a single shared memory pool. Deployment and software affect usable capacity.

Configured systemRequest quote

Availability, shipping
and taxes to confirm

View full configuration
Tower workstationIllustrative schematic
Grace Blackwell workstation

NVIDIA GB300-based Workstation

1 × NVIDIA B300

+ 1 × NVIDIA RTX PRO 4000 Blackwell SFF · 24 GB GDDR7

GPU memory
252 GB per primary GPU
System memory
496 GB LPDDR5X
GPUs included
Included
InferenceDevelopment
Configuration details
CPU
NVIDIA Grace · 72 cores
Storage
4 × 2 TB M.2 PCIe 5.0 · 2 OS drives in RAID1 and 2 data drives
Networking
2 × 400G QSFP112 ConnectX-8; 1 × 10GbE plus management
Power
1,600 W Titanium PSU · 120 V / 15 A minimum recommended
  • Workstation configuration based on NVIDIA GB300; final system configuration requires confirmation.
  • The primary accelerator is one B300; the additional RTX GPU has separate memory.
  • Power-cable inclusion requires confirmation.
  • AI Enterprise is not included.

Memory across multiple GPUs is aggregate capacity, not a single shared memory pool. Deployment and software affect usable capacity.

Configured systemRequest quote

Availability, shipping
and taxes to confirm

View full configuration
8U rack serverIllustrative schematic
NVIDIA HGX platform server

NVIDIA HGX B200

8 × NVIDIA Blackwell SXM

GPU memory
180 GB per primary GPU
System memory
24 × 96 GB DDR5-6400 ECC · 2,304 GB
GPUs included
Included
TrainingInference
Configuration details
CPU
2 × AMD EPYC 9555 · 64 cores each
Storage
1 × 2 TB M.2 PCIe 4.0; 2 × 7.68 TB U.2/U.3 PCIe 5.0
Networking
Dual 1GbE plus management; high-speed networking requires confirmation
Power
12 × 3,000 W Titanium PSUs · 6+6 redundancy · estimated draw 14.3 kW
  • AI Enterprise is not included.

Memory across multiple GPUs is aggregate capacity, not a single shared memory pool. Deployment and software affect usable capacity.

Configured systemRequest quote

Availability, shipping
and taxes to confirm

View full configuration
4U rack serverIllustrative schematic
PCIe GPU server

NVIDIA H200 NVL

8 × NVIDIA H200 NVL PCIe

GPU memory
141 GB per primary GPU
System memory
24 × 96 GB DDR5-6400 ECC · 2,304 GB
GPUs included
Included
TrainingInference
Configuration details
CPU
2 × AMD EPYC 9555 · 64 cores each
Storage
2 × 2 TB M.2 PCIe 4.0 · OS RAID1
Networking
Dual 10GbE Intel X710-AT2 plus management
Power
4 × 3,200 W Titanium PSUs · 3+1 redundancy · capacity depends on input voltage
  • Includes a five-year AI Enterprise license per GPU.

Memory across multiple GPUs is aggregate capacity, not a single shared memory pool. Deployment and software affect usable capacity.

Configured systemRequest quote

Availability, shipping
and taxes to confirm

View full configuration
Liquid-cooled rack-scale systemIllustrative schematic
NVIDIA DGX system

NVIDIA DGX GB200 NVL72

72 × NVIDIA Blackwell

GPU memory
Memory to confirm
System memory
17 TB LPDDR5X CPU memory
GPUs included
Included
TrainingInference
Configuration details
CPU
36 × NVIDIA Grace · 2,592 Arm Neoverse V2 cores total
Storage
To be confirmed
Networking
To be confirmed
Power
To be confirmed
  • Storage, networking, power, rails and site requirements need confirmation.
  • Mission Control and AI Enterprise subscription inclusion requires confirmation.

Memory across multiple GPUs is aggregate capacity, not a single shared memory pool. Deployment and software affect usable capacity.

Configured systemRequest quote

Availability, shipping
and taxes to confirm

View full configuration
10U rack serverIllustrative schematic
NVIDIA DGX system

NVIDIA DGX B300

8 × NVIDIA Blackwell Ultra SXM

GPU memory
Memory to confirm
System memory
To be confirmed
GPUs included
Included
TrainingInference
Configuration details
CPU
2 × Intel Xeon Platinum 6776P
Storage
2 × 1.92 TB M.2 boot; 8 × 3.84 TB E1.S self-encrypting NVMe cache
Networking
8 × OSFP 800 Gb/s ConnectX-8; 2 × dual-port 400 Gb/s BlueField-3; host and BMC GbE
Power
12 × 3.2 kW AC PSUs · 200–240 VAC · 14.5 kW consumption
  • AC PSU configuration; facility power requirements need confirmation.
  • DGX OS 7 based on Ubuntu 24.04 LTS.
  • Mission Control and AI Enterprise subscription inclusion requires confirmation.

Memory across multiple GPUs is aggregate capacity, not a single shared memory pool. Deployment and software affect usable capacity.

Configured systemRequest quote

Availability, shipping
and taxes to confirm

View full configuration
Liquid-cooled rack-scale systemIllustrative schematic
NVIDIA DGX systemRevision pending

NVIDIA DGX GB300 NVL72

72 × NVIDIA Blackwell GB300

GPU memory
Memory to confirm
System memory
To be confirmed
GPUs included
Confirm configuration
TrainingInference
Configuration details
CPU
36 × NVIDIA Grace · 2,592 Arm Neoverse V2 cores total
Storage
To be confirmed
Networking
72 × ConnectX-8 OSFP single-port 800 Gb/s InfiniBand; 18 × dual-port BlueField-3 200 Gb/s; BMC RJ45
Power
To be confirmed
  • Configuration revision is pending; final specifications require confirmation.
  • Mission Control and AI Enterprise subscription inclusion requires confirmation.

Memory across multiple GPUs is aggregate capacity, not a single shared memory pool. Deployment and software affect usable capacity.

Configured systemRequest quote

Availability, shipping
and taxes to confirm

View full configuration
Liquid-cooled rack-scale systemIllustrative schematic
NVIDIA DGX system

NVIDIA DGX Vera Rubin NVL72

72 × NVIDIA Rubin

GPU memory
Memory to confirm
System memory
Up to 54 TB LPDDR5X platform capacity; installed capacity unconfirmed
GPUs included
Included
TrainingInference
Configuration details
CPU
36 × NVIDIA Vera · 3,168 Olympus cores / 6,336 threads total
Storage
To be confirmed
Networking
ConnectX-9; BlueField-4; Quantum-X800 InfiniBand / Spectrum-X Ethernet; up to 28.8 TB/s scale-out
Power
To be determined; site survey required
  • GPU memory capacity requires confirmation.
  • System-memory capacity is a platform limit, not a confirmed installed amount.
  • Mission Control and AI Enterprise subscription inclusion requires confirmation.

Memory across multiple GPUs is aggregate capacity, not a single shared memory pool. Deployment and software affect usable capacity.

Configured systemRequest quote

Availability, shipping
and taxes to confirm

View full configuration
Form factor to be confirmedIllustrative schematic
NVIDIA DGX systemDetails pending

NVIDIA DGX B200

To be confirmed

GPU memory
Memory to confirm
System memory
To be confirmed
GPUs included
Unknown
Configuration details
CPU
To be confirmed
Storage
To be confirmed
Networking
To be confirmed
Power
To be confirmed
  • Detailed hardware configuration requires confirmation.

Memory across multiple GPUs is aggregate capacity, not a single shared memory pool. Deployment and software affect usable capacity.

Configured systemRequest quote

Availability, shipping
and taxes to confirm

View full configuration
Form factor to be confirmedIllustrative schematic
NVIDIA DGX systemDetails pending

NVIDIA DGX Rubin NVL8

To be confirmed

GPU memory
Memory to confirm
System memory
To be confirmed
GPUs included
Unknown
Configuration details
CPU
To be confirmed
Storage
To be confirmed
Networking
To be confirmed
Power
To be confirmed
  • Detailed hardware configuration requires confirmation.

Memory across multiple GPUs is aggregate capacity, not a single shared memory pool. Deployment and software affect usable capacity.

Configured systemRequest quote

Availability, shipping
and taxes to confirm

View full configuration

All configurations are for discussion. Pricing, GPU inclusion, final specifications, availability, shipping, taxes and procurement eligibility must be confirmed. Selecting a machine makes no commitment.

Your shortlist

The details, side by side.

Compare up to 4 systems. Saved on this browser; copy or download a summary when you’re ready.

No machines selected yet.
Add a configuration above to start comparing.

Browse machines ↗

This is a presentation tool. Saving or downloading a selection does not send an inquiry, reserve hardware, place an order, or confirm financing.