NVIDIA L40S rental: early access
Ada Lovelace · PCIe (being confirmed)
We’re collecting early-access requests. We’re placing our capacity in Uzbekistan. We don’t publish prices yet. We’ll confirm availability in our reply.
L40S
Request L40SWe’ll reply with terms for your workload. Nothing to pay before the contract.
Estimate usage- Memory per GPU
- 48 GB GDDR6
- Memory bandwidth
- 0.86 TB/s
- GPU interconnect
- PCIe 4, no NVLink
NVIDIA L40S specifications
- Architecture
- Ada Lovelace
- Announced
- 2023
- Memory
- 48 GB
- Memory type
- GDDR6
- Memory bandwidth
- 0.86 TB/s
- FP32
- 91.6 TFLOPS
- FP16/BF16, dense
- 362.05 TFLOPS
- FP8, dense
- 733 TFLOPS
- FP4, dense
- none
- FP64
- not in NVIDIA’s card specs
- RT cores
- yes
- NVENC/NVDEC
- NVENC 3, NVDEC 3
- MIG
- none
- GPU interconnect
- PCIe 4, no NVLink
- Power (TDP/TGP)
- 350 W
- Form factor
- PCIe (being confirmed)
Memory is specified per GPU. Multiple GPUs have separate memory: using a larger model across them requires compatible software and model partitioning. NVLink does not make them one GPU.
The L40S form factor is still being confirmed. Specs from NVIDIA’s official materials, checked 26 September 2026.
The L40S by workload
- Inference: LLMs up to 32B in FP8 on one GPU
- Fine-tuning: QLoRA up to 32B, LoRA up to 14B on one GPU
- Training from scratch: models up to 1B on one GPU
- Rendering: scenes up to 32 GB
- FP64 computing: not suited to this GPU
An estimate from open formulas, not a benchmark or an invoice.
Servers with the NVIDIA L40S
Server configuration (CPU, RAM, disk, network) is being confirmed.
Sizing assumption: up to 8 GPUs per server. We’ll confirm the actual configuration in our reply.
More than 8 GPUs? Discuss a clusterSources
Similar GPUs
- NVIDIA RTX 6000 Ada Generation
48 GB GDDR6 · 0.96 TB/s · PCIe 4
RTX 6000 Ada leads on 1 of 5 metrics. RTX 6000 Ada has more memory bandwidth (0.96 vs 0.86 TB/s). Equal: memory size. Nearly equal (under 1% apart): FP32, FP16/BF16, FP8.
- NVIDIA L4
24 GB GDDR6 · 0.3 TB/s · PCIe 4
L40S leads on all 5 metrics. L40S has more FP8 (733 vs 242 TFLOPS) and FP32 (91.6 vs 30.3 TFLOPS).
We’ll reply with terms for your workload. Nothing to pay before the contract.