AWS Reduces Prices for NVIDIA GPUs by 45%
On-demand pricing has dropped by 33% for A100s, 44% for H100s, and 25% for H200s.
Amazon Web Services (AWS) has cut prices for Nvidia H100 and A100 GPU instances, citing increased operational efficiencies.
The reductions, effective from June 1 for on-demand instances and June 4 for Savings Plans, apply to P4d (A100), P4de (A100), P5 (H100), and P5en (H200) instances.
On-demand pricing has dropped by 33% for A100s, 44% for H100s, and 25% for H200s.
Savings Plans now offer up to 45% reductions over a three-year commitment.
"This price reduction to On-Demand and Savings Plan pricing applies to all Regions where these instances are available. The pricing reduction applies to On-Demand purchases beginning June 1 and to Savings Plan purchases effective after June 4," AWS said in a blog post.
AWS’s EC2 Instance Savings Plan provides the steepest discounts for region-specific use, while the Compute Savings Plan offers greater flexibility. The newer B200 P6 GPUs are now also included in these plans.
AWS said such price cuts are part of its strategy to pass on scale-driven savings to customers. While smaller GPUaaS providers struggle with rapid price erosion, AWS’s vast infrastructure helps it absorb those shocks effectively.
AWS is now offering Amazon EC2 P6-B200 instances through Savings Plans to support large-scale deployments. These instances, initially launched on May 15, 2025, were previously available only via EC2 Capacity Blocks for ML.
"EC2 P6-B200 instances, powered by NVIDIA Blackwell GPUs, accelerate a broad range of GPU-enabled workloads but are especially well-suited for large-scale distributed AI training and inferencing," AWS added.
Comments ()