Upper Limit of Data that TPOT Can Handle

This is less so an issue than a question. I haven't been able to really find populated discussion forums for this package, so I thought I'd ask here.

I'm currently working on datasets for practice that are around 500-1000 rows and 13 features. It takes a really long time to do work on this, which makes sense.

This got me thinking then for larger datasets. I came across another post here that mentioned it being very time consuming and improbably to train a set on 1m records or more.

What do you think is the acceptable dataset size TPOT should be used on typically?