Skip to main content
Container Apps

Serverless containers,
without the cluster.

Run microservices, APIs, jobs, agents, and GPU workloads on a fully managed, Kubernetes-grade platform. Scale to zero, burst to hyperscale, and never write a YAML manifest you didn't want to.

Scale to zero. Configure apps to scale down to zero replicas when idle and scale out again as demand returns.
Burst to hyperscale. Autoscale on HTTP traffic, events, CPU or memory, or any KEDA-supported scaler.
No cluster to manage. Kubernetes-grade platform, fully managed — no nodes, no control plane, no upgrades.
Any container, any registry. Run images from ACR, Docker Hub, or any public or private registry.
Dapr and KEDA built in. Microservice APIs and event-driven autoscaling on an open-source foundation.
Advanced networking. VNet integration, private endpoints, and fine-grained ingress and egress control.

Anatomy of a container app​

Azure Container Apps is a serverless platform for containerized applications. You don't manage servers, orchestration, or cluster infrastructure, so you can focus on code while the platform keeps your apps stable and secure. An HTTP scale rule adds one replica for every 10 concurrent requests, between the min and max replicas you set — drag the load below to watch it respond.

Min replicas
0
Max replicas
4
Concurrent requests
30 requests→ 3 active replicasceil(30 ÷ 10) = 3, clamped to [0, 4]
Container imageyour code, containerizedACR · Docker Hub
deploy
Container Apps environmentregionvnetlogs
my-apirunning
HTTPS ingress · min 0 · max 4 · +1 replica per 10 requests
replica-1
0.5 vCPU · 1 GiB
running
replica-2
0.5 vCPU · 1 GiB
running
replica-3
0.5 vCPU · 1 GiB
running
replica-4
0.5 vCPU · 1 GiB
idle
30 req
Customerssend requests30 concurrent

What you can build​

API endpoints. Public or internal HTTP and gRPC APIs with managed TLS and zero-downtime revisions.
Microservices. Service discovery, internal-only ingress, and Dapr service invocation between apps.
Jobs and event processing. Run work on demand, on a schedule, or per event with per-run isolation.
AI apps and agents. Host agents, MCP servers, and inference on serverless GPUs — scale with demand.
Background workers. Long-running or queue-driven processing that scales on KEDA triggers.
Many more… SaaS backends, web apps, CI runners — anywhere serverless containers fit.

Building blocks​

Container Apps or AKS?

AKS gives you the full Kubernetes API and node-level control — use it when you need custom operators, cluster add-ons, or direct control of the control plane. Container Apps is serverless: it manages the cluster, scales to zero, and stays out of your way. Choose Container Apps when you want to ship containers without operating Kubernetes yourself.

Get started​