> EC2 instance that performs the same computations at a fraction of the cost

And typically with much lower latency/request times.

GCP functions, AWS Lambdas, etc... my anecdotal experience is that they are way slower for request times vs a $20 VPS running the exact same workload.

But managed K8s is so easy to setup and makes life and the development cycle so much easier you're basically robbing yourself if you don't.

I assume you just dropped a shit ton of features on your move to that EC2 instance unless of course you wasted more time then it would take to get everything working better on Lambda/SQS.

The actual compute is decoupled from the CI/CD pipeline. Moving to an EC2 instance does not preclude us from leveraging Lambda and other technologies.