AWS App Runner: current status, pricing and alternatives

AWS App Runner no longer accepts new customers, but existing services remain operational. AWS set the new-customer cutoff at 30 April 2026 and recommends Amazon ECS Express Mode for containerized applications. If you already use App Runner, assess your requirements and migration options before moving production traffic. AWS's notice.

This guide, published by Lizard, checks product status and linked pricing sources as of 9 September 2026. It distinguishes the policy change from a shutdown: the announcement does not say that existing services must stop on the cutoff date.

What App Runner does

App Runner provides a managed path from a supported source repository or container image to an HTTP service. It handles infrastructure tasks such as routing and service scaling. Your application still needs its database connections, secrets, health checks and release process.

For an existing user, its value can be the workflow already in place. Your AWS account may hold the registry, database, permissions and networking around the service. Replacing the compute layer alone does not replace those dependencies.

Understand the current support policy

AWS says it continues investing in security and availability but does not plan new App Runner features. New applications that require access as a new customer need another route. Existing users should check their account, requirements and AWS guidance before planning a migration.

Do not confuse this with Heroku's product-direction change. The policies are different, even though both can prompt a hosting review.

How App Runner billing works

App Runner distinguishes provisioned capacity from active capacity. A service can incur provisioned-memory charges while it waits for traffic. Active compute, source builds and other AWS resources can add charges. The exact rates depend on the region and configuration. App Runner pricing.

Use separate rows in your estimate:

Avoid multiplying every resource by a full month without checking its meter. Also avoid setting idle costs to zero just because the request count is zero. AWS's pricing examples are useful for checking which terms belong in the equation.

Which replacement should you evaluate?

ECS Express Mode reduces setup work, but you still need to understand the resources and costs in your account. Start with AWS's linked migration guidance and Fargate pricing rather than assuming an App Runner-sized bill.

For a broader choice, compare PaaS providers. For the difference between image packaging and runtime behaviour, see container as a service.

Keep the database decision separate

Moving an HTTP service does not require moving its database at the same time. You can test the new runtime first, provided the database connection remains secure and latency is acceptable. Cross-provider or cross-region traffic can change both latency and cost.

If you move the database too, test extensions, roles, connection limits and restore time. Verify application-level reads and writes, not just whether an import command completed. Decide how to handle writes during cutover and how to recover if the new app fails.

A migration sequence with clear checks

1. Record the image or source revision, runtime settings, IAM dependencies, secrets and connected services.
2. Deploy an isolated test copy on the candidate host. Set the expected port and health endpoint.
3. Connect test data and run the important flows: login, writes, file handling and any callbacks.
4. Load-test enough to observe latency, scaling and errors. Estimate the bill from the actual meters.
5. Prepare the final data and traffic switch. Include certificate and domain checks, monitoring and a rollback decision point.

On Lizard, the Git deployment guide covers source deployments and the configuration reference explains repeatable settings. If the app also has background work, configure it as a separate worker instead of forcing it into the web process.

FAQ

Is AWS App Runner shut down? No general shutdown is stated in the cited notice. AWS restricts new customers and keeps existing services operational, with security and availability work continuing.

What does AWS recommend instead? AWS names ECS Express Mode. Evaluate it against your app's networking, permissions and deployment requirements before switching.

Does an idle App Runner service cost nothing? Do not assume so. Provisioned memory can remain billable. Use the provisioned and active components from the pricing page for your region.

Can I move only the application? Yes, if it can securely reach its existing dependencies. Test the network path, latency and transfer charges before keeping that arrangement in production.

Is Lizard a drop-in replacement? It can run a conventional application service, but configuration, cloud-specific integrations and data access still need review. Test a copy before moving traffic.