Moderate to move
Lambda on Vultr: what it becomes and what it costs.
On Vultr, Lambda becomes containers on Vultr Kubernetes Engine, or long-running services on Vultr Cloud Compute instances. Handlers port to containers; triggers from other AWS services need re-wiring.
Effort grade
Moderate
Same software, self-hosted: you patch, back up and fix it
Cost after the move, before engineer time
35% of today
$0.0000167 per GB-second is about $43 per always-on GB of memory; a CX33 (8 GB) is €8.49. 35% reflects typical request volumes where a 2-vCPU server replaces the function fleet, with the one-server floor applied.
Engineer time
6 h a month
To run the replacement: patching, backups, on-call. Plus about 5 engineer days for the move itself.
Other ways to replace it
The report lets you pick any of these per service; the figure and the grade follow.
Knative or OpenFaaS on Vultr Kubernetes Engine (VKE)
Keeps the function shape and scale-to-zero on your own cluster; triggers from other AWS services are still re-wired.
45% of today · Moderate
Cloudflare Workers
Stays serverless at a flat, cheap price; handlers are rewritten for the Workers runtime and AWS triggers are replaced by HTTP and queues.
50% of today · Hard
In three modelled AWS accounts
What Lambda costs in the modelled accounts and what the estimator makes of it on Vultr, with the default assumptions. Not customer data.
| Account | AWS /mo | Vultr /mo | Change | Priced by |
|---|---|---|---|---|
| Early-stage SaaS$2,981/mo in total | $66 | $23 | −$43 | ratio |
| Scale-up on Kubernetes$24,950/mo in total | $300 | $105 | −$195 | ratio |
Your own Lambda line, from your export
Drop one month of your AWS cost export and the report prices every service, Lambda included, on Vultr list prices of 18 Oct 2026, net of the engineer time to run it. In your browser; nothing is uploaded.