NVIDIA L40S 48GB PCIe GPU Accelerator

NVIDIA L40S 48GB PCIe GPU Accelerator

Brand: NVIDIA | Category: GPUs

SKU: 900-2G133-0200-000 | Part #: 900-2G133-0200-000 | MPN: 900-2G133-0200-000

Contact for Pricing — Request a Quote

Request a Quote Contact Us

About the NVIDIA L40S 48GB PCIe GPU Accelerator

The NVIDIA L40S is a professional-grade GPU accelerator built on NVIDIA's Ada Lovelace architecture, delivering exceptional performance for enterprise AI inference, data analytics, and graphics-intensive applications. With 48GB of GDDR6 memory and full-height, full-length PCIe form factor, the L40S integrates seamlessly into standard datacenter infrastructure while maintaining thermal efficiency and reliability for mission-critical deployments.

Designed for virtualized and bare-metal environments, the L40S supports multi-instance GPU (MIG) partitioning for flexible workload consolidation and advanced memory compression technologies that maximize throughput per watt. The accelerator features dual-slot passive cooling and operates within standard enterprise thermal and power parameters, making it suitable for large-scale production deployments across AI model serving, scientific computing, and content creation pipelines.

Ideal for

  • Large language model (LLM) and generative AI inference serving at scale
  • Enterprise machine learning model training and fine-tuning workflows
  • Real-time data analytics and in-database GPU acceleration
  • Medical imaging processing and computational research simulations
  • High-performance graphics rendering and digital content creation
  • Financial services risk modeling and quantitative analysis

Technical specifications

ManufacturerNVIDIA
ModelNameL40S
ManufacturerPartNumber900-2G133-0200-000
ArchitectureAda Lovelace
GPUMemory48GB GDDR6
MemoryBandwidth960 GB/s
CUDACompute18,176 NVIDIA CUDA Cores
TensorCores568 Tensor Cores
PCIeGenerationPCIe 5.0
FormFactorFull-Height Full-Length (FHFL)
ThermalDesignPower320W
MaxPowerConsumption320W
CoolingDesignDual-slot passive cooling
PrecisionSupportFP32, TF32, BFLOAT16, FP16, FP8, INT8
MultiInstanceGPUYes (up to 7 instances)
DisplayOutputDisplayPort 1.4a
InterfaceConnectivitySingle PCIe 5.0 x16 slot

Available from Omnixon Global. Submit an RFQ and our team will confirm configuration and availability for your order.

Technical Specifications

BrandNVIDIA
CategoryGPUs
SKU900-2G133-0200-000
Part Number900-2G133-0200-000
ConditionNew
ModelNameL40S
ManufacturerPartNumber900-2G133-0200-000
ArchitectureAda Lovelace
GPUMemory48GB GDDR6
MemoryBandwidth960 GB/s
CUDACompute18,176 NVIDIA CUDA Cores
TensorCores568 Tensor Cores
PCIeGenerationPCIe 5.0
FormFactorFull-Height Full-Length (FHFL)
ThermalDesignPower320W
MaxPowerConsumption320W
CoolingDesignDual-slot passive cooling
PrecisionSupportFP32, TF32, BFLOAT16, FP16, FP8, INT8
MultiInstanceGPUYes (up to 7 instances)
DisplayOutputDisplayPort 1.4a
InterfaceConnectivitySingle PCIe 5.0 x16 slot