It depends on the outage. There was one a year or two ago (I think? They run together) that impacted EC2 such that as long as you weren’t trying to scale, or issue any commands, your service would continue to operate. The EKS clusters at my job at the time kept chugging along, but had Karptenter tried to schedule more nodes, we’d have had a bad time.
Meanwhile, AWS has always marketed itself as "elastic". Not being able to start new VMs in the morning to handle the daytime load will wreck many sites.
