Datacenter · Ada Lovelace · launched 2022 · prices checked 2026-09-03
Rent NVIDIA L40 — 48 GB, $0.234/hr on-demand
- VRAM 48 GBGDDR6
- FP16 tensor 181TFLOPS
- PowerScore 127RTX 3090 = 100
- Configs 1–4×PCIe 4.0
- Online now 6 2 regions
Billed per second, price locked at deploy. Storage $0.08/GB/mo · bandwidth $0.01/GB — the whole fee schedule. Next weekly market re-check: 2026-09-10.
Datacenter · Ada Lovelace architecture
The L40 is the datacenter Ada card built for visualisation and inference: 48 GB GDDR6, 18,176 CUDA cores, passively cooled at 300 W. It shares silicon with the L40S at slightly lower clocks, making it a cheaper 48 GB option for render farms and SDXL serving.
A server-class card built for sustained 24/7 load: passive cooling in proper chassis, ECC memory, and drivers validated for compute. The sweet spot for inference fleets and fine-tuning jobs that need stability more than headline FLOPS.
NVIDIA L40 specs: VRAM, TFLOPS, bandwidth
| GPU model | NVIDIA L40 | Architecture | Ada Lovelace (2022) |
|---|---|---|---|
| VRAM | 48 GB GDDR6 | Memory bandwidth | 864 GB/s |
| FP16 tensor perf. | 181 TFLOPS | FP32 perf. | 90.5 TFLOPS |
| CUDA cores | 18,176 | TDP | 300 W |
| PowerScore (RTX 3090 = 100) | 127 | PCIe generation | Gen 4.0 |
| Multi-GPU | 1× – 4× | Max instance storage | 4,000 GB NVMe |
| Network up to | 10,000 Mbps | CUDA | 12.4 – 13.0 |
Bandwidth, CUDA cores, TDP and FP32 are public NVIDIA figures; FP16 tensor is the dense (non-sparsity) number. Machine-level values come from live inventory.
L40 price per hour: on-demand, interruptible, reserved
One public rule sets every price on this page: the marketplace median for the L40 ($0.34/hr, snapshot 2026-09-03) × 0.70, rounded down — so on-demand is $0.234, 30% below market. Interruptible halves it; a 3-month reservation takes another 35% off.
| Mode | Per GPU-hour | Per day (24 h) | Per month (730 h) | What you get |
|---|---|---|---|---|
| On-demand | $0.234 | $5.62 | $171 | Guaranteed capacity, price locked at deploy, stop anytime |
| Interruptible | $0.117 | $2.81 | $85 | Flat −50%; may pause under capacity pressure, disk kept, auto-requeue |
| Reserved (3 months) | $0.152 | $3.65 | $111 | −35% on on-demand, rate locked for the term, capacity held |
Per GPU: an 4× machine costs exactly 4× — no multi-GPU premium. Estimate a full month with storage and bandwidth in the GPU cost calculator.
What you can run on a L40 (48 GB VRAM)
With 48 GB of GDDR6, a single card holds a ~14B-parameter LLM in FP16 or up to ~72B parameters quantized to 4-bit, with room for KV-cache at practical context lengths. Scale to 4× GPUs on one machine for bigger models or bigger batches — the per-GPU price stays $0.234.
- One-click template: vLLM on a L40
- One-click template: PyTorch on a L40
- One-click template: ComfyUI on a L40
- Sizing help: LLM VRAM requirements guide
L40 availability by region
6 × L40 across 2 machines, live from inventory:
Montréal
Paris
L40 vs alternatives: price per TFLOP
| GPU | VRAM | FP16 | On-demand | $ / TFLOP-hr |
|---|---|---|---|---|
| L40 this card | 48 GB | 181 | $0.234 | $1.29‰ |
| L4 | 24 GB | 121 | $0.225 | $1.86‰ |
| A10 | 24 GB | 125 | $0.168 | $1.34‰ |
| Tesla T4 | 16 GB | 65 | $0.104 | $1.60‰ |
| Tesla V100 | 16 GB | 125 | $0.089 | $0.71‰ |
‰ = dollars per 1,000 TFLOP-hours of FP16 — a rough value-for-compute yardstick across cards.
Renting a L40: frequently asked questions
How much does it cost to rent an NVIDIA L40 per hour?
$0.234 per GPU-hour on-demand — a fixed price set at least 30% below the current market median of $0.34. Interruptible capacity costs $0.117/hr and a 3-month reservation $0.152/hr. Around $171/month if you keep one running non-stop, billed per second.
What can a L40 with 48 GB VRAM run?
In LLM terms, roughly a 14B-parameter model in FP16 or up to ~72B parameters 4-bit quantized on a single card, with context headroom. Multi-GPU instances (up to 4× on current inventory) multiply that; diffusion and rendering workloads fit comfortably at this VRAM class.
Is the L40 available to rent right now?
Yes — 6 GPUs across 2 machines in 2 regions are listed as we render this page. Configurations go from 1× to 4×. Deploy from the console and it is running in about 30 seconds.
How do I deploy a L40?
Create an account (email + password, no card, no KYC), top up in crypto, open the console, filter by L40, pick a machine and a template such as PyTorch, vLLM or ComfyUI. The same deploy is one command with the CLI: powergpu launch --gpu l40 --template pytorch.
Why is the L40 cheaper here than on GPU marketplaces?
We price from the public marketplace median and fix our on-demand rate at least 30% below it, rounded down. The price is re-checked weekly (next check 2026-09-10) and published — no auctions, no per-host roulette, no bidding.