AWS lambda documentation change

Service: lambda · 2026-06-25 · Documentation low

File: lambda/latest/dg/gettingstarted-limits.md

Summary

Added new 'Lambda MicroVMs' section documenting quotas and limits for the MicroVM feature

Security assessment

The change introduces operational quotas for a new compute option without any security context, vulnerabilities, or security features. It purely documents service limits without security implications.

Diff

diff --git a/lambda/latest/dg/gettingstarted-limits.md b/lambda/latest/dg/gettingstarted-limits.md
index 399adff35..ae1ae827a 100644
--- a//lambda/latest/dg/gettingstarted-limits.md
+++ b//lambda/latest/dg/gettingstarted-limits.md
@@ -7 +7 @@
-Compute and storageFunction configuration, deployment, and executionLambda API requestsOther services
+Compute and storageFunction configuration, deployment, and executionLambda API requestsMicroVMsOther services
@@ -32,0 +33,2 @@ The following sections list default quotas and limits in Lambda by category.
+  * Lambda MicroVMs
+
@@ -125,0 +128,47 @@ Remainder of the control plane API requests (excludes invocation, GetFunction, a
+## Lambda MicroVMs
+
+Lambda MicroVMs sets quotas for compute, storage, and API requests. Quotas marked as adjustable can be increased through the Service Quotas console.
+
+###### Note
+
+Lambda MicroVMs support the ARM64 (AWS Graviton) architecture.
+
+### Compute and storage
+
+Resource | Default quota | Adjustable  
+---|---|---  
+Memory allocated across all MicroVMs (per account, per Region) |  400 GB 1,024 GB in US East (N. Virginia), US West (Oregon), US East (Ohio), and Asia Pacific (Tokyo) Burstable up to four times this quota. | Yes  
+Maximum execution duration per MicroVM | 8 hours (28,800 seconds) | No  
+Disk space per vCPU | 4 GB | No  
+  
+### Images and versions
+
+Resource | Default quota | Adjustable  
+---|---|---  
+MicroVM images per account per Region | 100 | Yes  
+Versions per MicroVM image | 50 | Yes  
+Concurrent image builds (per account, per Region) |  5 10 in US East (N. Virginia), US West (Oregon), US East (Ohio), and Asia Pacific (Tokyo) | Yes  
+  
+### Per-MicroVM throughput
+
+Resource | Limit | Adjustable  
+---|---|---  
+Concurrent connections per MicroVM | 8 (1 vCPU), 16 (2 vCPU), 32 (4 vCPU), 64 (8 vCPU), 128 (16 vCPU) | No  
+Requests per second per MicroVM | 40 (4 vCPU / 8 GB), 160 (16 vCPU / 32 GB) | No  
+  
+### API rate limits
+
+API operation | Rate (TPS) | Burst | Adjustable  
+---|---|---|---  
+`RunMicrovm` | 5 | 5 | No  
+`ResumeMicrovm` | 5 | 5 | No  
+`SuspendMicrovm` | 2 | 2 | No  
+`TerminateMicrovm` | 10 | 10 | No  
+`GetMicrovm` | 100 | 100 | No  
+`CreateMicrovmAuthToken` | 50 | 50 | No  
+`CreateMicrovmShellAuthToken` | 5 | 5 | No  
+  
+###### Note
+
+TPS = transactions per second. These rate limits are per account, per Region, and are fixed. Retry throttled requests using exponential backoff with jitter.
+