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
Single-slot low-profile PCIe card with passive heatsink and gold edge connector

NVIDIA

NVIDIA L4 Tensor 900-2G193-0000-001 24 GB ai accelerator card

SKU CH-11SS-B2N5W MPN 900-2G193-0000-001

Accelerates AI inference with 24 GB memory and 72 W TDP in a single-slot PCIe 4.0 card

  • InterfacePCIe Gen4 x16 64GB/s
  • Chipset ManufacturerNVIDIA
  • GPUL4 Tensor Core GPU
  • Memory Size24GB
  • Memory Clock300GB/s
  • System RequirementsPartner and NVIDIA-Certified Systems with 1–8 GPUs
  • Low 72W TDP reduces energy cost per inference job
  • PCIe Gen4 x16 64GB/s interface feeds data to 24GB memory at 300GB/s
  • FP8 Tensor Core throughput of 485 teraFLOPs accelerates generative AI workloads
  • 1-slot low-profile form factor fits dense 1–8 GPU server configurations
  • Dual NVENC, quad NVDEC and quad JPEG decoders speed video transcoding pipelines

AI Accelerator Card 24 GB

The NVIDIA L4 Tensor is a 24 GB AI accelerator card built on the Ada architecture. It connects through a PCIe 4.0 x16 interface and fits in a single low-profile slot. The board has a maximum thermal design power of 72 W and is qualified for partner and NVIDIA-Certified systems that support one to eight GPUs.

Upgrade From Earlier AI Cards

The card delivers FP32 performance of 30.3 teraFLOPs and tensor throughput of 120 teraFLOPs for TF32, 242 teraFLOPs for FP16 and BFLOAT16, and 485 teraFLOPs for FP8 and INT8. Memory bandwidth reaches 300 GB/s across 24 GB of GPU memory. Video engines include two NVENC encoders, four NVDEC decoders and four JPEG decoders. The jump is worth it when workloads need higher tensor throughput and media decode density in a 72 W envelope.

Suits Dense Inference And Media Servers

This accelerator suits inference serving, generative AI and video transcoding pipelines that benefit from the listed tensor rates and decoder count in a single-slot form factor. Buyers who need more GPU memory than 24 GB or who require multi-GPU scaling beyond eight cards should look at alternative platforms. Systems without certified partner validation may not meet the stated system requirements.

Highlights

  • Low 72W TDP reduces energy cost per inference job
  • PCIe Gen4 x16 64GB/s interface feeds data to 24GB memory at 300GB/s
  • FP8 Tensor Core throughput of 485 teraFLOPs accelerates generative AI workloads
  • 1-slot low-profile form factor fits dense 1–8 GPU server configurations
  • Dual NVENC, quad NVDEC and quad JPEG decoders speed video transcoding pipelines

Specifications

BrandNVIDIA
ModelL4 Tensor
Part Number900-2G193-0000-001
InterfacePCIe Gen4 x16 64GB/s
Chipset ManufacturerNVIDIA
GPUL4 Tensor Core GPU
Memory Size24GB
Memory Clock300GB/s
System RequirementsPartner and NVIDIA-Certified Systems with 1–8 GPUs
PowerMax thermal design power (TDP): 72W
Form Factor1-slot low-profile, PCIe
FeaturesFP32: 30.3 teraFLOPs
TF32 Tensor Core: 120 teraFLOPS*
FP16 Tensor Core: 242 teraFLOPS*
BFLOAT16 Tensor Core: 242 teraFLOPS*
FP8 Tensor Core: 485 teraFLOPs*
INT8 Tensor Core: 485 TOPs*
GPU memory: 24GB
GPU memory bandwidth: 300GB/s
NVENC / NVDEC / JPEG decoders: 2 / 4 / 4
Shipping weight0.36 kg
Package size267 × 142 × 76 mm

Questions about this item

What slot width does the 1-slot low-profile form factor require?

The card occupies a single PCIe slot and uses a low-profile bracket, so it fits in 1U servers or workstations that accept half-height cards.

Which power connector must be seated for the 72W TDP?

Insert the card into a PCIe 4.0 x16 slot; the 72W maximum draw is supplied entirely through the slot, so no external power cable is needed.

When should the passive heatsink be cleaned to maintain the 72W thermal limit?

Inspect the heatsink fins every 6–12 months and remove dust with compressed air; restricted airflow raises GPU temperature and can trigger throttling.

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.