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
Full-height dual-slot graphics card with silver cooler shroud and four DisplayPort outputs

NVIDIA

NVIDIA RTX PRO 4500 32GB GDDR7 AI Accelerator Card 900-5G147-2250-000

SKU CH-DGDE-KP3C6 MPN 900-5G147-2250-000

A full-height dual-slot AI accelerator with 10496 CUDA cores, 328 Tensor cores, 32 GB GDDR7 ECC memory on a 256-bit bus, four DisplayPort 2.1b outputs and 200 W power draw

  • InterfacePCI Express 5.0 x16
  • Chipset ManufacturerNVIDIA
  • GPURTX PRO 4500
  • CUDA Cores10496
  • Tensor Cores5th Generation
  • Memory Size32GB
  • 32 GB GDDR7 memory with ECC protects large AI workloads from data errors
  • 10496 CUDA cores deliver high parallel throughput for compute-intensive tasks
  • 328 fifth-generation Tensor cores accelerate matrix operations in training and inference
  • PCI Express 5.0 x16 interface provides fast host bandwidth for data transfer
  • 200 W power draw in a dual-slot full-height form factor fits dense server deployments

AI Accelerator with 32 GB GDDR7 ECC

The NVIDIA RTX PRO 4500 is an AI accelerator card built on the Blackwell architecture. It delivers 10496 CUDA cores and 328 fifth-generation Tensor cores for training and inference workloads. The 32 GB GDDR7 memory with ECC operates on a 256-bit interface. It targets developers and researchers who need high memory capacity with error correction.

Active dual-slot cooler at 200 W

A full-height dual-slot design with an integrated fan dissipates up to 200 W in typical server chassis. The active cooler keeps thermals managed during sustained compute loads. Power is delivered through the PCI Express 5.0 x16 slot without auxiliary connectors. Acoustic output scales with workload intensity in standard rack airflow.

PCIe 5.0 x16 interface requirement

The card requires a PCI Express 5.0 x16 slot to achieve full bandwidth for AI workloads. Systems limited to PCIe 4.0 or narrower links will constrain data throughput. Four DisplayPort 2.1b outputs support up to four 3840 x 2160 displays at 165 Hz or two 7680 x 4320 displays at 100 Hz. Buyers without a compatible 5.0 x16 slot cannot utilise the accelerator at its rated interface speed.

Highlights

  • 32 GB GDDR7 memory with ECC protects large AI workloads from data errors
  • 10496 CUDA cores deliver high parallel throughput for compute-intensive tasks
  • 328 fifth-generation Tensor cores accelerate matrix operations in training and inference
  • PCI Express 5.0 x16 interface provides fast host bandwidth for data transfer
  • 200 W power draw in a dual-slot full-height form factor fits dense server deployments

Specifications

BrandNVIDIA
Model900-5G147-2250-000
InterfacePCI Express 5.0 x16
Chipset ManufacturerNVIDIA
GPURTX PRO 4500
CUDA Cores10496
Tensor Cores5th Generation
Memory Size32GB
Memory Interface256-bit
Memory TypeGDDR7 with ECC
DirectXDirectX 12
OpenGLOpenGL 4.6
DisplayPort4 x DisplayPort 2.1b
Ray Tracing Support4th Gen Ray Tracing Cores
Max Displays Per Board4 x 3840 x 2160 @ 165 Hz
2 x 7680 x 4320 @ 100 Hz
CoolerWith Fan
AI FeaturesAI Accelerated
Power Consumption200W
Dimensions (L x H)4.4" x 10.5"
Slot WidthDual slot, full height
Shipping weight1.54 kg
Package size387 × 191 × 83 mm

Questions about this item

How do I install the RTX PRO 4500 in a PCI Express 5.0 x16 slot, and what step is often missed?

Insert the dual-slot, full-height card (4.4" x 10.5") firmly into a PCI Express 5.0 x16 slot, secure the bracket screw, then connect the required power cables for the 200 W TDP — the step most often missed is attaching those power connectors before closing the chassis.

What maintenance will the GDDR7 memory with ECC and the active cooler need over time?

The GDDR7 ECC memory requires no user maintenance; schedule quarterly cleaning of the active fan heatsink with compressed air to prevent dust build-up that would reduce airflow across the dual-slot cooler.

At what point does the 200 W power limit throttle sustained AI workloads on the 10496 CUDA cores?

When the GPU hits its 200 W power ceiling, the Boost clock drops to stay within that envelope, so sustained AI compute on the 10496 CUDA cores and 328 Tensor cores will throttle once total board power reaches 200 W.

New to this? Buying server hardware with crypto — what to check, what it costs to pay, and what is never asked for.

Also in this aisle

People compared these

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