AWS Lambda introduces MicroVMs for isolated sandboxes with full lifecycle control
AWS launches AWS Lambda MicroVMs, providing VM-level isolated sandboxes with rapid launch, state preservation, and lifecycle control without infrastructure management.
AWS Lambda MicroVMs is a new serverless compute primitive within AWS Lambda that allows running user- or AI-generated code in isolated, stateful execution environments. Powered by Firecracker, it offers virtual machine level isolation with no shared kernel or resources between users, near-instant launch and resume from snapshots, and full lifecycle control including suspension and resumption with memory and disk state intact. This is designed for multi-tenant applications like AI coding assistants, interactive code environments, and vulnerability scanners that require dedicated isolated environments for untrusted code. Developers create MicroVM Images by packaging code and Dockerfiles into zip files uploaded to Amazon S3. Lambda builds the image by running the Dockerfile, initializing the application, and taking a Firecracker snapshot of the running environment's memory and disk state. Subsequent MicroVMs launch from this pre-initialized snapshot, enabling rapid startup. MicroVMs support up to 8 hours of runtime and can be suspended automatically after configurable idle periods to reduce costs while preserving state. The service is available in US East (N. Virginia, Ohio), US West (Oregon), Europe (Ireland), and Asia Pacific (Tokyo) regions on ARM64 architecture, offering up to 16 vCPUs, 32 GB memory, and 32 GB disk per MicroVM. Lambda MicroVMs complement Lambda Functions by providing isolated environments for long-running interactive sessions, while Lambda Functions remain suited for event-driven workloads. Pricing and documentation are available on the AWS website.