By the way how much malice can we attribute to cloud vendors?

>Moreover, Kubernetes’s slow autoscaling meant I had to over-provision services to ensure availability, paying for unused resources rather than scaling based on demand.

A typical Linux instance on AWS starts up in about 8 seconds from the asking to start to command line, so lets double that. You could start up and shut down instances in 15 seconds.

Why the hell do you need to overprovision instances then, or leave empty ones running?

I don't see any other reasonable explanation than its in the best interest of cloud vendors to not have short lived instances for the purposes of load balancing, as well as make you consume as much CPU time as possible, even if you don't need it.

TL;DR someone didn't learn to plan capacity.