I'd argue the whole point is actually the fact that you can lease CPU/Memory/space as you need it, and capacity constraints now become simple cash constraints. You don't need to shell out millions of dollars on a specialist enormous hardware just to be able to use it for an hour.

Lots of big companies that operate extensively in AWS/Azure/GCP don't go anywhere near the managed services they offer, because they end up being a horror show in terms of scalability, functionality and troubleshootability. Depending on your risk appetite, running Kafka on Fargate/EC2 is a lot more attractive than using Kinesis (for example).