Updated October 25, 2026
· ~2 min read
Lambda made "serverless" mean no servers at all: you pay per request and per millisecond, $0.20 per million requests plus compute by GB-second, and AWS runs everything underneath, scaling from zero to thousands of concurrent executions. Knative and OpenFaaS bring the same developer experience (deploy a container or function, scale to zero, scale on requests) to a Kubernetes cluster you run.
The difference is what "serverless" leaves you responsible for.
Scale to zero on Kubernetes scales your pods to zero, not your bill: the nodes underneath still run unless the cluster autoscaler removes them, and the control plane always does. Lambda's economics come from AWS pooling capacity across every customer. If a workload is idle most of the time, nothing self-hosted competes with that; if it is busy all the time, a container on Fargate or a node you already pay for usually beats both.
Question 1 of 1
Do you already run Kubernetes that this must live on?
On AWS, use Lambda for event-driven and spiky work — it is cheaper than anything you can run at low and moderate volume, needs no cluster, and integrates natively with SQS, EventBridge, S3, API Gateway and DynamoDB. Use ECS on Fargate rather than Knative when you outgrow Lambda's limits but want to stay managed. Choose Knative only when you already run Kubernetes and want scale-to-zero services that run the same way on any cluster; OpenFaaS for a simpler functions platform on Kubernetes or at the edge.
We work with startups and growing companies on AWS architecture decisions. Tell us about your situation and we will recommend the right approach.
As of October 25, 2026. Provided as is, for general information, and accurate to the best of our knowledge on the date shown. Not legal, compliance, tax, financial or other professional advice. Prices, limits and vendor services change; check current vendor documentation before relying on anything here. Terms of use.