IoT QoS 1 Unacknowledged Publish Request Limit

AWS IoT Core will not wait to send a PUBACK, however there may be reasons for reaching that quota. As an example: In a situation where the downlink (i.e. Server -> Client) connectivity is slower then uplink (Client -> Server) connectivity it can cause a situation where this quota could be overrun. So yes, client is limited on how many "unacknowledged" QoS1 messages it can publish.

This should also be seen in conjunction with "Outbound publish requests per second per account". A broker sends the message to all subscribing clients and then replies with a PUBACK packet. If because of downlink connectivity or for some other reason, outbound queue is full, that will build a backlog of unacknowledged QoS1 messages.