Amazon AppStream 2.0 pricing

Amazon AppStream 2.0 is an AWS End User Computing (EUC) service optimized for Software-as-a-Service (SaaS) application streaming and converting desktops apps to SaaS.

With Amazon AppStream 2.0, you pay only for the Desktop-as-a-Service (DaaS) resources that you provision, public IPv4 address these resources consumes, plus a small monthly fee per end user depending on the operating system chosen. There is no upfront investment and no long-term commitment when using Amazon AppStream 2.0.

Streaming resources consist of AppStream 2.0 Always-On fleet instances, On-Demand fleet instances, Elastic fleet instances, image builder, and app block builder instances. Pricing below is per instance-hour consumed. Each partial instance-hour consumed will be billed per-second for Elastic fleet instances or instances that use the Red Hat Enterprise Linux, and Amazon Linux operating system. The pricing is based on the instance type, size, and operating system you select. The instance charge includes compute, storage, and any network traffic used by the streaming protocol. AppStream 2.0 monitors your fleet utilization and performs automatic adjustments to fleet capacity to meet your user demand at the lowest possible cost. The capacity adjustments are made based on scaling policies that you define based on the current utilization or based on a schedule. Auto scaling applies both to Always-On and On-Demand fleets. For more information, see Fleet Auto Scaling for Amazon AppStream 2.0 in the Amazon AppStream 2.0 Administration Guide.

User fees are charged per user per month in each AWS Region where users stream applications from AppStream 2.0 fleet instances that are based on Microsoft Windows Server operating systems. 

Using the US East (N. Virginia) Region, you launch a stream.standard.medium image builder to install applications and create an image, from which you provision an On-Demand fleet consisting of two stream.standard.medium fleet instances. During the month, one of your users launches streaming sessions from your On-Demand fleet using both the SAML 2.0 and Streaming URL authentication types. And, during the same month, another user launches streaming sessions from the On-Demand fleet using SAML 2.0. 

In the US East (N. Virginia) Region, the hourly price for Windows stream.standard.medium is $0.10 per hour. The stopped instance fee is $0.025 per hour. AppStream 2.0 also charges a user fee of $4.19 (Microsoft Remote Desktop Service Subscriber Access License or RDS SAL) for each authorized user that launches a streaming session from an AppStream 2.0 Windows fleet during the month. User fees are not incurred for administrators connecting to and using image builders to create images.

Assuming you have 100 users streaming On-Demand AppStream 2.0 instances for 5 hours a day, 5 days a week, 4 weeks a month (100 hours per user per month), and your instances are provisioned and stopped for 20 hours per user per month, your total monthly cost would be $1,469 ($14.69 per user). The total cost is calculated as (100 hours per user x 100 users x $0.10 per hour stream.standard.medium fleet instance) + (20 hours per user x 100 users x $0.025 per hour stopped instance fee) + (100 users x $4.19 RDS SAL user fee) = $1,469 per month. If an administrator uses the image builder for 5 hours per month, an additional cost of 5 hours x $0.10 per hour = $0.50 (no RDS SAL user fee applied). This brings the total cost to $1,469.50 per month ($14.70 per user per month).

Additionally, you will pay for public IPv4 addresses used by the provisioned instances and image builders.
Total IPv4 charges = (100 hours per user x 100 users x $0.005 per hour per IPv4 address for fleet instances) + (5 hours x $0.005 per hour per IPv4 address for image builders) = $50.025

In the US East (N. Virginia) Region, the hourly price for Linux stream.standard.medium is $0.084 per hour. The stopped instance fee is $0.025 per hour. Streaming from Linux instances does not incur user fee.

Assuming your image builder instance runs for 3.5 hours, your two On-Demand streaming instances each stream for 4.5 hours a day over one business week (45 hours), and your instances are stopped for the rest of the business week (195 hours), your total cost will be $8.85. The total cost is calculated as: (45 hours x $0.082 per hour stream.standard.medium fleet instance) + (195 hours x $0.025 per hour stopped instance fee) + (3.5 hours x $0.082 per hour stream.standard.medium image builder) = $8.85. There is no user fee for streaming from Amazon Linux 2 based fleet instances.

Additionally, you will pay for public IPv4 addresses used by the provisioned instances and image builders.
Total IPv4 charges = (45 hours per user x $0.005 per hour per IPv4 address for fleet instances) + (3.5 hours x $0.005 per hour per IPv4 address for image builders) = $0.2425

AppStream 2.0 also provides a simple pricing tool that you can use to estimate your total costs. For more information, see Amazon AppStream 2.0 releases a simple pricing tool on the AWS Desktop and Application Streaming blog.

Download the AppStream 2.0 Pricing Tool:  Microsoft Excel File  I  OpenOffice Calc File

* Hourly pricing fee charged for running instances only. For Image Builder and Always-On fleets, instances may be considered running if they are available for use, even if no user is connected. For On-Demand fleets, instances are considered running only if users are connected with an active streaming session.

Note: AppStream stream.graphics-pro.* instances will no longer be available after 10/31/2025 due to end of life of hardware supporting stream.graphics-pro instance types. Please consider using alternative graphics instances.

* Elastic fleets are billed for duration of the streaming session, in seconds, with a minimum of 15 minutes. Pricing is per instance-hour. An additional user fee may be assessed when users stream applications from streaming instances using the Microsoft Windows Server operating system. Pricing fee charged for running instances only. For app block builder, instances may be considered running if they are available for use, even if no user is connected.