L40S vs A100: cloud rental cost compared
Short answer: pick the A100 when you're training or doing large-batch work that's bandwidth-bound. Pick the L40S when you're running inference at scale and the L40S's positioning and price fit better than the A100's training-grade spec.
You can rent both by the hour, no purchase required. A100 from $0.735/GPU/hr across 7 providers; L40S from $0.790/GPU/hr across 5 providers. The A100 is 7% cheaper at the entry rate.
Specs side by side
Price by provider
Which should you pick: L40S or A100?
The A100 has more than double the memory bandwidth of the L40S, 2,039 GB/s of HBM2e against 864 GB/s of GDDR6 (NVIDIA product pages, Aug 2026), which is what training and large-batch workloads are usually bottlenecked on, while the L40S is a data-center-AI-positioned card built more for inference density. Pick the A100 when you're training or doing large-batch work that's bandwidth-bound; pick the L40S when you're running inference at scale and the L40S's positioning and price fit better than the A100's training-grade spec. On price the A100 undercuts the L40S by 7% at the entry rate ($0.735 vs $0.790/GPU/hr). Worth weighing if the deciding factor above isn't a hard requirement for your job.
Decision criteria
Pick the A100 when
- You're training or fine-tuning and need the bandwidth
- Your job is large-batch and memory-bandwidth-bound
- Your pipeline is already built around Ampere/A100
Pick the L40S when
- You're serving inference at scale and want FP8 support
- You want the SKU NVIDIA positions for data-center AI inference
- L40S pricing or availability is better where you're deploying
Price-performance
At list rates, 1,000 GPU-hours costs $735 on the A100 against $790 on the L40S. Neither NVIDIA nor Aquanode publishes a workload-normalized $/token or $/epoch figure for this pair, so $/GPU-hour below is the only apples-to-apples number. A faster card can still cost less per finished job even at a higher hourly rate.
A100 vs L40S: common questions
Is the A100 or the L40S cheaper to rent?
On Aquanode's live marketplace the A100 starts at $0.735/GPU/hr (median $1.59/GPU/hr across 7 providers) and the L40S starts at $0.790/GPU/hr (median $1.09/GPU/hr across 5 providers). The A100 is the cheaper of the two at the entry rate, by 7%.
What is the difference between the A100 and the L40S?
The A100 has 80 GB of VRAM against the L40S's 48 GB; the A100 is an Ampere part and the L40S is Ada Lovelace (compute capability 8.0 vs 8.9); only the L40S supports FP8 compute. On price, the A100 lists from $0.735/GPU/hr and the L40S from $0.790/GPU/hr.
Which cloud providers offer the A100 and the L40S?
7 providers list the A100 (Vast.ai, RunPod, Verda, Massed Compute, HyperStack, Jarvislabs and Akash) and 5 list the L40S (RunPod, Vast.ai, Massed Compute, Verda and Nebius), across 7 and 5 regions respectively. Both are available from Vast.ai, RunPod, Verda and Massed Compute.
How much does 1,000 GPU-hours cost on the A100 vs the L40S?
At the lowest rates listed today, 1,000 GPU-hours costs $735 on the A100 and $790 on the L40S, a difference of $55 for the same runtime. Rates are per GPU per hour and update hourly.
Is the L40S a cheaper alternative to the A100 for AI work?
It depends what "AI work" means for you. The L40S has FP8 tensor cores the A100 lacks, which helps quantized inference, but the A100's 2,039 GB/s of bandwidth (vs the L40S's 864 GB/s) still wins for bandwidth-bound training. Today the A100 lists from $0.735/GPU/hr and the L40S from $0.790/GPU/hr on Aquanode.
How this comparison is calculated
Every price is normalized to a per-GPU hourly rate using the same pipeline as every other pricing surface on Aquanode, and only the cheapest qualifying offer per provider is shown. A dash means that provider does not currently list that GPU.
Architecture, compute capability and supported precisions come from each vendor's published datasheet for that generation, not from the marketplace feed. This page regenerates at most once per hour.
