POWERING THE NEXT AGE OF INTELLIGENCE-RTX PRO 6000 BLACKWELL 96GB

NVIDIA L40 / L40S Enterprise 48GB

NVIDIA L40 / L40S Enterprise 48GB is a curated accelerator for local inference, fine-tuning, visualization, or production AI compute. Its key configuration includes L40S and 48GB.

Request a quote Engineer replies, not a bot.
REPLY WITHIN 24H AUTHORIZED SOURCING ENGINEER SUPPORT
BULK ORDER OR CUSTOM BUILD?TALK TO AN ENGINEER
DATASHEET

Technical
specifications

EDITIONS
Available optionsNVIDIA L40 / NVIDIA L40S
ArchitectureNVIDIA Ada Lovelace
Coolingpassive, server airflow required
MEMORY
Capacity48 GB GDDR6 with ECC
Memory interface384-bit
Memory bandwidth864 GB/s
COMPUTE
CUDA cores18,176
Tensor cores568 (fourth generation)
RT cores142 (third generation)
INTERFACE
Host interfacePCIe 4.0 x16
Form factorfull-height, dual-slot
POWER
L40 total board power300 W
L40S total board power350 W
Workload classinference, training, rendering and digital twins
Silicon & memory
Memory
48 GB
TDP
350 W
Released
2023
InferenceFine-tuningGraphics

Every order, covered.

[ 01 ]Sealed & genuineFactory-sealed, authorized stock. Warranty valid.
[ 02 ]Insured, tracked dispatchLead time is confirmed on your order, not guessed at.
[ 03 ]Real engineer supportTalk to people who run this hardware.
[ 04 ]Returns without drama30 days. DOA? Replacement ships first.

Pairs well with

QUOTEPBA-0501
NVIDIA GeForce RTX 5090
GPUNVIDIA GeForce RTX 509032GB GDDR721,760 CUDA575WPrice on requestQUOTE
QUOTEPBA-0506
NVIDIA RTX PRO 6000 Blackwell 96GB
GPUNVIDIA RTX PRO 6000 Blackwell 96GB96GB GDDR7 ECC600W / 300W editionsPCIe 5.0Price on requestQUOTE
QUOTEPBA-0511
NVIDIA H100 80GB PCIe
GPUNVIDIA H100 80GB PCIe80GB HBM2e2 TB/sPCIe Gen5Price on requestQUOTE
QUOTEPBA-0512
NVIDIA H200 NVL 141GB
GPUNVIDIA H200 NVL 141GB141GB HBM3e4.8 TB/sNVLink bridgePrice on requestQUOTE