Skip to content
NVIDIA HGX H200 8-GPU Server

NVIDIA HGX H200 8-GPU Server

Eight H200 SXM GPUs with 141 GB HBM3e each on NVLink; chassis and configuration agreed per order

Model
HGX H200 8-GPU Server
Condition
New, original packaging
Quantity
Ask on WhatsApp

Price on request. Tell us the quantity and destination and we reply within one working day.

How we supply this item

Supplied through licensed partners. Server, networking and data-centre hardware is supplied through licensed partners of InTechnologies FZCO. The partner issues the contract and invoice; we handle sourcing, consolidation in Dubai and delivery coordination. Export compliance

Export compliance. This product is subject to US export regulations. How we handle it is described on the export compliance page. Export compliance

An HGX H200 server is built around NVIDIA’s eight-GPU HGX baseboard with Hopper-generation H200 SXM GPUs. Each GPU carries 141 GB of HBM3e with 4.8 TB/s of memory bandwidth, and the eight GPUs are joined by NVLink with NVSwitch at 900 GB/s per GPU, so the whole board, about 1.1 TB of GPU memory, behaves as one large accelerator for training and for serving models that do not fit on smaller systems. Per GPU, NVIDIA lists 3,958 TFLOPS of FP8 and 1,979 TFLOPS of FP16/BF16 Tensor Core throughput with sparsity, and 67 TFLOPS of FP64 Tensor Core for HPC work.

The same HGX H200 board is shipped by several OEMs in 6U to 8U air-cooled chassis and in liquid-cooled versions. Processors, system memory, NVMe storage, network adapters and power supplies differ from vendor to vendor and from build to build, so we do not fix them on this page: tell us your workload, rack power and cooling, and we return matching configurations from the vendors available for your delivery window.

InTechnologies FZCO arranges HGX H200 servers for business customers in the UAE, the GCC and Central Asia. The contract and invoice are issued by our licensed partners; we handle sourcing, consolidation in Dubai Airport Freezone and DDP or DDU delivery. H200 is controlled under US export regulations, and every order goes through end-user verification before availability is confirmed.

Specifications

NVIDIA HGX H200 8-GPU Server: Specifications
GPUs8× NVIDIA H200 SXM (Hopper) on an HGX baseboard
GPU memory141 GB HBM3e per GPU, about 1.1 TB in total
Memory bandwidth4.8 TB/s per GPU
GPU-to-GPU interconnectNVLink with NVSwitch, 900 GB/s per GPU
FP8 Tensor Core3,958 TFLOPS per GPU (with sparsity)
Server vendor and chassis—
Processors—
System memory—
Storage—
Network adapters—
Power supplies and cooling—

— the vendor, chassis and these components are agreed per order. Message us your workload, quantity and rack power, and we will send the configurations we can supply. Ask on WhatsApp

Questions buyers ask

Why are the vendor, CPU and memory shown as a dash?

Because the configuration is agreed per order. Several OEMs build servers on the same HGX H200 board, and availability changes by quarter. Send us a message with your workload, quantity and site conditions, and we will come back with the vendors and configurations we can supply.

How does an HGX H200 server differ from a DGX H200?

The GPUs and NVLink are the same. DGX is NVIDIA’s own complete system with a fixed configuration and NVIDIA software and support; an HGX server is built by an OEM, so the chassis, processors, memory, storage and networking can be chosen.

Is the H200 export-controlled?

Yes. Servers with H200 GPUs fall under US export regulations. Our licensed partner reviews an end-user statement and, where required, applies for a licence before the order is accepted.

What is the lead time?

It depends on the vendor, GPU allocation and export processing. We give a delivery window with each quotation.

NVIDIA

NVIDIA HGX B300 8-GPU Server

Eight Blackwell Ultra SXM GPUs with 2.1 TB HBM3e on NVLink 5; chassis and configuration agreed per order