I agree with all the contributors above, but I'd say from our experience the overriding driver is perceived cost. We see a double-digit proportion of all migration inquiries saying that they want to explore migrating from an on-premise archive to an Exchange archive, either on premise or hosted. I am of the view that many are exploring this simply because they are driven to find ways to put cost pressure on their existing vendor and this is a way that they might do that. Conversely, if you've had trouble making your archive system meet your requirements and it has gained a bad reputation internally, then I imagine that the idea to hand that data back to Exchange where expertise and processes are more standardized, must be attractive.
Of the customer projects taking this route with TransVault Migrator so far there are only two of large size - both with about 30 TB of compressed archive and about 10,000 users, and both with on-premise archives that are end-of-life, not EV.
Three more that our partners are migrating are in the 10 TB range, but the majority come under 5 TBs. However the important thing to remember is that 5 TBs probably equals 15 TBs of data you now need to keep in Exchange Personal Archives or Online Archive - that's a large amount of data when you have no integrated eDiscovery, but I know in some cases customers have elected to keep a journal archive on premise.
Incidentally, a good reason use Migrator to do this with is that it has direct EWS ingestion built in so sending data to a Personal or Online Archive is a one-step migration rather than by going through a mailbox which generates lots of extra Exchange load and logs.
Final point: we just were asked about migrating many TBs from Personal Archives straight into EV. The answer was 'yes' and we expect that this could be a popular route too in the future!