HBase has a bulk loading feature that can be used by Sqoop to stage files and then bulk load them into HBase. This is preferable for large amounts of data as the normal CRUD based API is otherwise quickly overloaded. See the HBase suppied ImportTsv.java and its used of the "importtsv.bulk.output" command line option. It shows how to easily switch between direct API import and bulk file staging.
It might be necessary to add an additional step into Sqoop that allows to sample the data and presplit the table into the right amount of regions before the initial loading. This could be done here, or as another issue.