PantheonGet Early Access

NVIDIA

NVIDIA H100 NVL (PCIe Server)

NVIDIA H100 NVL PCIe server — 94 GB HBM3 per card with NVLink bridges, Hopper compute for facilities that want PCIe rather than SXM.

Available
Air-cooled
NVIDIA
Supermicro
PCIe server
Hopper
Condition: New. Lead time: 2–4 weeks.Request a quote →

NVIDIA H100 NVL PCIe server, built on a 5U PCIe Gen5 platform (e.g. Supermicro SYS-521GE-TNRT) that takes up to eight dual-wide cards. Each H100 NVL provides 94 GB of HBM3 and pairs over an NVLink bridge for 900 GB/s card-to-card, delivering Hopper-class training and inference without an SXM baseboard — the drop-in path for standard air-cooled racks. Dual Intel Xeon processors host the GPUs. Configuration and availability confirmed at quote.

Specifications

GPU architectureNVIDIA H100 NVL (Hopper)
GPUs per serverUp to 8× H100 NVL (dual-wide PCIe Gen5)
GPU memory94 GB HBM3 per card · ~3.9 TB/s per card
CPU2× Intel Xeon Scalable
GPU interconnectNVLink bridge · 900 GB/s per card pair
Scale-out networkingPCIe Gen5 NICs, configurable to 400 Gb/s
Form factor5U PCIe server, air-cooled
CoolingAir-cooled
GPU card power350–400 W per card (configurable)
ConditionNew
Warranty3-yr on-site hardware warranty

Specifications shown are headline figures. Confirm full configuration details when you request a quote.

Last updated

Request a quote for NVIDIA H100 NVL (PCIe Server)

Tell us your requirements and we'll get back to you with availability and pricing.

or reach us directly at