Does Amazon EC2 Have a One-Minute Minimum Charge?

Amazon EC2 applies a one-minute minimum charge each time an On-Demand instance transitions into the running state for instances billed per second. However, not all Amazon EC2 instances use per-second billing. The minimum charge depends entirely on the operating system, licensing model, and instance family selected when launching the virtual machine. This article explains how Amazon EC2 billing granularity works, when the 60-second minimum applies, and which instances are subject to hourly or 24-hour minimums instead.

How Per-Second Billing Works in Amazon EC2

For eligible Amazon EC2 On-Demand instances, billing is calculated in one-second increments after an initial 60-second minimum period. Whenever you launch a new instance or start a previously stopped instance, the billing clock begins the moment the instance enters the running state and stops as soon as it transitions to the stopping or terminated state.

If an eligible instance runs for less than 60 seconds before being stopped or terminated, Amazon Web Services (AWS) rounds the billable duration up to a full 60 seconds. Once the instance exceeds one minute of runtime, you are billed strictly for the exact number of seconds the instance remains running.

Using the hourly On-Demand rate \(R\), the total compute cost \(C\) for a single launch and runtime session of \(t\) seconds (under per-second billing) is calculated as:

\[C = R \times \frac{\max(60, t)}{3600}\]

For example, if an instance runs for 35 seconds, \(\max(60, 35) = 60\), meaning you are charged for 60 seconds (or \(\frac{1}{60}\) of the hourly rate). If it runs for 125 seconds, you are charged for \(\frac{125}{3600}\) of the hourly rate.

Repeated Stops and Starts

The minimum charge applies each time an instance enters the running state, not just once per instance lifetime. If you stop an instance and start it again a few minutes later, that new start triggers a separate billing cycle with its own minimum charge requirement.

For instance, if you start a Linux instance, run it for 20 seconds, stop it, and repeat this process three times within an hour, each of the three runs incurs the 60-second minimum charge, resulting in 180 billable seconds total rather than 60 seconds.

Operating Systems and Instances Eligible for the One-Minute Minimum

Per-second billing with a 60-second minimum applies to On-Demand, Reserved, and Spot instances running the following operating systems:

• Open-Source Linux Distributions: Amazon Linux, Ubuntu, Debian, CentOS, and Fedora (when launched without paid AWS Marketplace subscriptions).
• Microsoft Windows Server: Windows Server instances that include the license from AWS (License Included), as well as Windows instances using Bring Your Own License (BYOL).
• Red Hat Enterprise Linux (RHEL) and SUSE Linux Enterprise Server (SLES): AWS transitioned On-Demand RHEL and SLES instances (including RHEL for SAP) from hourly to per-second billing with a one-minute minimum.

Exceptions to the One-Minute Minimum

While most modern workloads on EC2 benefit from per-second billing and the one-minute minimum, two major exceptions enforce longer minimum billing durations:

1. AWS Marketplace Instances (One-Hour Minimum)

Many commercial third-party software images offered through the AWS Marketplace are billed by the hour rather than by the second. If you launch an On-Demand instance using an Amazon Machine Image (AMI) that carries an hourly software licensing fee not eligible for per-second billing, the minimum charge is one full hour (3,600 seconds) each time the instance is started. Even if the instance is terminated after 45 seconds, a full instance-hour is billed.

2. Amazon EC2 Mac Instances (24-Hour Minimum)

Amazon EC2 Mac instances (mac1, mac2, and mac2-m2 families) are built on physical Apple Mac mini computers hosted as Amazon EC2 Dedicated Hosts. To comply with the Apple macOS Software License Agreement, which requires a minimum lease constraint on underlying hardware, EC2 Mac Dedicated Hosts enforce a strict 24-hour minimum allocation period.

Once you allocate a Mac Dedicated Host to run an EC2 Mac instance, you cannot release that host until a full 24 hours have passed. After the initial 24-hour period is met, the host transitions to per-second billing for as long as it remains allocated.

Summary of EC2 Minimum Billing Durations