We take cryptocurrency — so the coins you hold can buy real hardware.Pay in crypto, spend it on real hardware. Free insured shipping worldwide over $399. Plain, unbranded boxes, sent from Germany.Free insured shipping over $399. Your rate is locked for 30 minutes once checkout opens.Rate locked 30 minutes. Every item sealed, serial-checked and covered by a 24-month warranty.Sealed, 24-month warranty.

Saved items Sign in
NVIDIA Tesla V100 32 GB HBM2 GPU in SXM2 form factor with passive cooler

NVIDIA

NVIDIA Tesla V100 32 GB HBM2 graphics card

SKU CH-9ZB7-Z1Q93 MPN V100

Data-center GPU with 32 GB HBM2 memory delivers up to 100 CPU-equivalent performance for AI and HPC workloads

  • Chipset ManufacturerNVIDIA
  • GPUV100
  • Memory TypeHBM2
  • OpenGLOpenCL
  • Form FactorSXM2
  • 32 GB HBM2 memory enables large-model AI training
  • Passive cooler suits dense data-centre racks
  • Volta architecture accelerates AI, HPC and graphics workloads
  • SXM2 form factor fits compatible server boards
  • OpenCL, OpenACC and DirectCompute APIs supported

Data-center GPU with 32 GB HBM2

The NVIDIA Tesla V100 is a compute-focused graphics card built on the Volta architecture for AI, HPC and graphics workloads. It delivers the performance of up to 100 CPUs in a single GPU for data scientists, researchers and engineers. The card uses a passive cooler and the SXM2 form factor. It supports OpenACC, OpenCL and DirectCompute APIs.

SXM2 form factor requires compatible server board

This accelerator uses the SXM2 socket, so it must be installed in a server motherboard or chassis designed for that interface. The passive cooler relies on system-level airflow rather than an onboard fan. Because it draws power and cooling from the host platform, the build must provide adequate slot spacing and directed airflow across the heatsink.

Sustained compute with passive cooling

Under continuous load the GPU maintains performance through the high-bandwidth HBM2 memory and the Volta architecture without active fans. The passive design eliminates moving parts that can wear out during long training or simulation runs. System integrators must ensure the enclosure supplies sufficient airflow to keep temperatures stable over extended periods.

Highlights

  • 32 GB HBM2 memory enables large-model AI training
  • Passive cooler suits dense data-centre racks
  • Volta architecture accelerates AI, HPC and graphics workloads
  • SXM2 form factor fits compatible server boards
  • OpenCL, OpenACC and DirectCompute APIs supported

Specifications

BrandNVIDIA
Chipset ManufacturerNVIDIA
GPUV100
Memory TypeHBM2
OpenGLOpenCL
Form FactorSXM2

Questions about this item

When does the 32 GB HBM2 memory capacity become the limiting factor?

The 32 GB HBM2 limit is reached when training datasets or simulation meshes exceed the onboard capacity, forcing data to spill over slower system memory and reducing throughput.

How much heat does the passive cooler need to dissipate during continuous operation?

The passive cooler relies on forced airflow from the server chassis to remove heat; sustained all-day loads require the host system to supply sufficient air volume across the heatsink.

What server infrastructure must already be present to install the SXM2 module?

The SXM2 module requires a compatible NVIDIA HGX or OEM server baseboard that provides the socket, power delivery, and NVLink interconnects.

Also in this aisle

People compared these

Same shelf, same checkout — eight coins and a 30-minute rate lock.