AWS Lambda adds 90-minute function timeout on Managed Instances

AWS raised the Lambda function timeout to 90 minutes for asynchronous and event-source-mapping invocations on Lambda Managed Instances, a six-fold increase from the long-standing 15-minute ceiling. The change targets workloads that previously had to be split across services or moved to containers/ECS: long-running data processing, media transcoding, financial calculations, AI inference, and batch jobs.

Mechanically, the higher timeout only applies on Managed Instances (not standard Lambda), and it pairs with a separate announcement that Managed Instances now support Graviton5-powered C9g/C9gd/M9g/M9gd instances delivering up to 25% better compute performance versus Graviton4. Together they push Lambda further into the territory of long-lived, compute-heavy tasks while preserving Lambda's serverless operational model.

For AI builders specifically, a 90-minute window makes on-Lambda inference and multi-step agent runs far more viable without re-architecting onto always-on infrastructure — a meaningful convenience as AWS also builds out Bedrock AgentCore (agents running up to 14 days on dedicated compute) and Agent Registry governance. The caveat is cost and cold-start behavior at longer durations, and the Managed-Instances-only scope limits who benefits immediately. Watch adoption among teams currently bridging Lambda-to-containers just to escape the 15-minute limit, and whether the batch/AI-inference use cases AWS cites materialize in practice.