Three rack-server concepts, from a two-GPU target to an eight-GPU HGX platform target. Start with your workload, deployment environment and infrastructure requirements.
Concept designs and target configurations. Rack heights are design targets; these systems have not completed engineering validation and are not available to order.
S2 / 2U TARGET / TWO GPUS
ZXO S2
A 2U rack-server concept targeting two RTX PRO Server Edition GPUs.
Target form factor
2U rack server
Target GPUs
2 × NVIDIA RTX PRO 6000 Blackwell Server Edition
GPU memory
96 GB GDDR7 ECC per GPU
GPU power
Up to 600 W per GPU, configurable
The cooling variant, power setting, GPU spacing and chassis airflow are not finalized. The 2U target does not establish mechanical or thermal compatibility. GPU memory remains separate on each card.
A 4U rack-server concept targeting four H200 NVL PCIe GPUs.
Target form factor
4U rack server
Target GPUs
4 × NVIDIA H200 NVL
GPU memory
141 GB HBM3e per GPU
GPU power
Up to 600 W per GPU, configurable
GPU format
PCIe, dual-slot, air-cooled
H200 NVL supports two- or four-way NVLink bridge configurations. ZXO bridge selection, card spacing, topology, chassis cooling and software compatibility remain unvalidated. GPU count does not establish application scaling or a unified memory pool.
An 8U-class exterior concept targeting an NVIDIA HGX H200 8-GPU platform.
Target form factor
8U-class rack server
Target platform
NVIDIA HGX H200 8-GPU
Target GPUs
8 × NVIDIA H200 SXM
GPU memory
141 GB HBM3e per GPU
GPU power
Up to 700 W per GPU, configurable
HGX H200 uses an SXM GPU baseboard with NVLink and NVSwitch, not eight standard PCIe add-in cards. The exterior concept does not establish HGX integration. Baseboard sourcing, chassis fit, cooling, power delivery and firmware require validation.
GPU specifications are component references. Rack height, integration and operating requirements remain engineering targets.
ZXO server targets — engineering validation pending
Model
Rack target
GPU target
Memory per GPU
Maximum GPU power
S2
2U
2 × RTX PRO 6000 Blackwell Server Edition
96 GB GDDR7 ECC
Up to 600 W per GPU
S4
4U
4 × H200 NVL, PCIe
141 GB HBM3e
Up to 600 W per GPU
S8
8U-class
HGX H200 8-GPU platform with H200 SXM
141 GB HBM3e
Up to 700 W per GPU
GPU power is configurable and excludes the rest of the system. Memory per GPU is not a guarantee of a single application-visible pool. Final CPU, host memory, storage, networking, rack depth, rails, redundancy and facility requirements have not been announced.
Renders describe exterior design intent. Mechanical fit, thermal design, electrical service, software and system compliance require validation. No ZXO system is claimed to be NVIDIA-Certified.
Availability
Component sourcing and final configurations are unconfirmed. Pricing, warranty, delivery regions and lead times have not been announced. This site accepts inquiries, not orders or payments.
Plan the deployment
Share your workload, GPU needs, installation country, rack constraints and available power and cooling. An inquiry does not reserve hardware or confirm compatibility.