Specifications
| Parameter | Value |
|---|---|
| GPU Architecture | NVIDIA Ada Lovelace |
| Memory | 48 GB GDDR6 with ECC |
| Memory Bandwidth | 864 GB/s |
| NVLink | Not supported (PCIe only) |
| Form Factor | Dual-slot, PCIe Gen4 x16 |
| Max Power | 300 W |
| Tensor Cores | 4th Generation — 1,466 TOPS (INT8) |
| RT Cores | 3rd Generation — 209 TFLOPS |
| Use Cases | AI Inference, Visualization, VDI, Rendering |
Pricing by Channel
Authorized Distribution
$7,800 – $9,500
¥53,000 – ¥65,000
DigiKey • Mouser • Avnet • Arrow
⭐ In Stock
⭐ In Stock
China Domestic
¥56,000 – ¥75,000
~$8,200 – $11,000
Leadtek 丽台 • Digital China 神州数码 • JD.com 京东
Second-hand / Refurbished
$5,000 – $7,000
¥30,000 – ¥48,000
eBay • Xianyu 闲鱼 • ServerMonkey
L40S Pricing (NVIDIA 900-2G133-0080-000)
The L40S is an enhanced variant with 1,466 TOPS (INT8), ideal for AI inference. Cloud rental starts at $0.40/hr (Vast.ai) to $3.50/hr (AWS).
New / Authorized
$9,100 – $11,975
¥66,000 – ¥87,000
PNY • Dihuni • ServerSupply
MPN: 900-2G133-0080-000
MPN: 900-2G133-0080-000
Cloud Rental
$0.40 – $3.50/hr
~¥3 – ¥25/hr
Vast.ai • AWS g6e • gpus.io
Median: $1.22/hr
Median: $1.22/hr
Refurbished
$8,950
~¥65,000
ServerSupply • eBay
48GB GDDR6, Factory Refurb
48GB GDDR6, Factory Refurb
Channel Coverage
15 distribution channels monitored across authorized distributors, domestic Chinese retailers, and secondary markets. Pricing refreshed regularly.
Need bulk pricing or specific configuration?
Request L40 Quote →