GPU Economy

Amazon Web Services

EC2 on-demand, Linux, us-east-1 only; instance price stored per node with GPU count (p5 = 8x H100). Spot needs signed API calls - not recorded.

As of , Amazon Web Services lists 6 GPU models we track, from $1.86 per GPU-hour (NVIDIA L40S) to $14.24 (NVIDIA B200).

Websiteaws.amazon.com
Pricing pageaws.amazon.com/ec2/pricing/on-demand/
RegionsNot stated
Models tracked6

Compare Amazon Web Services with

DataCrunch (6 shared GPUs) · Modal (6 shared GPUs) · RunPod (6 shared GPUs) · CoreWeave (5 shared GPUs) · Google Cloud (5 shared GPUs) · Oracle Cloud (OCI) (5 shared GPUs) · Vast.ai (5 shared GPUs) · Hyperstack (4 shared GPUs) · Jarvislabs (4 shared GPUs) · Lambda (4 shared GPUs) · Microsoft Azure (4 shared GPUs) · Nebius (4 shared GPUs) · Cudo Compute (3 shared GPUs) · GMI Cloud (3 shared GPUs) · Together AI (3 shared GPUs)

All listed prices

Every Amazon Web Services rate we track, normalized to one GPU for one hour.
GPUUSD / GPU-hourTypeRegionMin. commitmentSource
NVIDIA L40S$1.86on-demandus-east-1Price page
NVIDIA A100 40GB$2.74
$21.96/node ÷ 8
on-demandus-east-1Price page
NVIDIA A100 80GB$3.43
$27.45/node ÷ 8
on-demandus-east-1Price page
NVIDIA H100 SXM$6.88
$55.04/node ÷ 8
on-demandus-east-1Price page
NVIDIA H200$7.91
$63.30/node ÷ 8
on-demandus-east-1Price page
NVIDIA B200$14.24
$113.93/node ÷ 8
on-demandus-east-1Price page

Data updated: