Every time the payload of a Mule message gets transformed we create a new message adapter. Which is intensive. There are some ways we can avoid making this an expensive operation which would reduce latency for some Mule scenarios.