AWS has reduced the prices of some of its EC2 Nvidia GPU-accelerated instances to attract more AI workloads while competing with rivals, such as Microsoft and Google, as demand for GPUs and the cost of securing them continues to grow.
The instances that are seeing price cuts up to 45% include the P4 (P4d and P4de) and P5 (P5 and P5en) instance types on both On-Demand and Savings Plan options. Enterprises have the option of choosing two kinds of plans in the Savings Plan — Instance Savings Plan and Compute Savings Plan.
The pricing reduction took effect from June 1 for On-Demand purchases and June 4 for Savings Plan purchases.
“Price cuts on P4d, P4de, P5, and P5en GPU instances suggest a targeted price competition move. These instances, powered by Nvidia A100 and H100-class GPUs, are central to generative AI workloads and already in demand,” said Kaustubh K, practice director at Everest Group.
“The reductions can be considered all about removing cost friction for AI buyers and positioning AWS more aggressively against Microsoft Azure and Google Cloud in the high-performance compute space. This is designed to drive scale, increase stickiness, and secure long-term infrastructure loyalty among enterprise AI teams,” Kaustubh added.
Discounts across most plans
According to AWS, the price reduction is based on the instance type and the purchase plan opted for by customers.
For P4d instances, On-Demand purchases will see a decrease of 33% in costs. When the same instance is purchased via the Savings Plan, enterprises will see a 31% decrease in costs for both Savings Plan options for a one-year period.
For three-year periods, costs on P4d on the Instance Savings Plan will decrease by 25%. The Compute Savings plan is not available for the same period.
P4de instances get the same reduction as P4d instances across all plans.
For P5 and P5en instances, AWS has reduced prices by 44% and 25%, respectively, under the On-demand plan.
On the Instance Savings Plan, both instances are not seeing any reduction, at least for one-year packs under the Instance Savings Plan.
For three three-year periods under the Instance Savings Plan, P5 and P5en prices have been reduced by 45% and 26% respectively.
Under the Compute Savings Plan, enterprises will see a 44% and 25% reduction in the price of P5 and P5en, respectively, for one-year packs. The three-year Compute Savings pack for P5 has been discounted by 25%, AWS said, adding that P5en is not getting any discount on the same pack.
Further, the cloud services provider said that it was increasing accessibility to reduced pricing by making at-scale On-Demand capacity available for P4d, P4de, P5, and P5en to more of its cloud regions.
While the P4 instance is being made available in the Asia Pacific (Seoul), Asia Pacific (Sydney), Canada (Central), and Europe (London) Regions, the P4de instance is being made available in the US East (N. Virginia) Region.
AWS is extending the availability of the P5 and P5en instances in the Asia Pacific (Mumbai), Asia Pacific (Tokyo), Asia Pacific (Jakarta), and South America (São Paulo) Regions. AWS offers other EC2 instances for hyper-computing in the form of P2, P3, P6, G, and G5g series.