NVIDIA 900-2G133-0080-000 L40S 48GB GDDR6 384bit Graphics ca
The NVIDIA L40S (part number 900-2G133-0080-000) is an Ada-generation datacenter GPU built for AI inference, ML workloads, rendering, and enterprise visualization. It targets integrators and datacenter buyers deploying GPU servers in space-constrained or edge environments where power efficiency matters. Note that it uses a passive thermal design and is intended for properly cooled server chassis, not desktop workstations.
Key specifications (table)
| Parameter | Specification |
|---|---|
| Part Number | 900-2G133-0080-000 |
| GPU | NVIDIA L40S (Ada architecture, datacenter class) |
| Memory Capacity | 48 GB GDDR6 |
| Memory Interface | 384-bit |
| Accelerators | Tensor Cores for inference and ML acceleration |
| Cooling | Passive, designed for server chassis airflow (not suitable for air-cooled desktops) |
| Duty Cycle | Engineered for 24/7 enterprise operation |
Good to know
- Designed for edge and space-constrained deployments; verify your server chassis provides adequate airflow for the passive heatsink before ordering.
- Within the NVIDIA datacenter lineup, the L40S sits as a versatile single-slot-class accelerator positioned between pure visualization cards and flagship training GPUs — well suited for mixed inference, rendering, and VDI workloads rather than dedicated large-scale training.
- Compatible with modern AI software stacks and development tools; confirm server PSU and thermal clearances against your platform’s OEM guidelines before volume purchase.
For tray and reel pricing, current stock by part number and date code, and volume discounts on bulk orders, request a bulk quotation.
The NVIDIA L40S (part number 900-2G133-0080-000) is an Ada-generation datacenter GPU built for AI inference, ML workloads, rendering, and enterprise visualization. It targets integrators and datacenter buyers deploying GPU servers in space-constrained or edge environments where power efficiency matters. Note that it uses a passive thermal design and is intended for properly cooled server chassis, not desktop workstations.
Key specifications (table)
| Parameter | Specification |
|---|---|
| Part Number | 900-2G133-0080-000 |
| GPU | NVIDIA L40S (Ada architecture, datacenter class) |
| Memory Capacity | 48 GB GDDR6 |
| Memory Interface | 384-bit |
| Accelerators | Tensor Cores for inference and ML acceleration |
| Cooling | Passive, designed for server chassis airflow (not suitable for air-cooled desktops) |
| Duty Cycle | Engineered for 24/7 enterprise operation |
Good to know
- Designed for edge and space-constrained deployments; verify your server chassis provides adequate airflow for the passive heatsink before ordering.
- Within the NVIDIA datacenter lineup, the L40S sits as a versatile single-slot-class accelerator positioned between pure visualization cards and flagship training GPUs — well suited for mixed inference, rendering, and VDI workloads rather than dedicated large-scale training.
- Compatible with modern AI software stacks and development tools; confirm server PSU and thermal clearances against your platform’s OEM guidelines before volume purchase.
For tray and reel pricing, current stock by part number and date code, and volume discounts on bulk orders, request a bulk quotation.
