My team has a case with Kinesis streams where multiple lambdas (and a firehose) processes the same data and performs different actions on them. This can't be done with SQS, and there's no "pausing" with SNS. If for some reason, a lambda has issues processing data, we have until the TRIM_HORIZON to fix the issue and resume processing.

I will say, though, the the KCL library is a PoS and the few containerize services that are reading from Kinesis exhibit this annoying behavior of causing the iterator age to spike to the TRIM_HORIZON when performing a blue/green deployment for the containers.