There are multiple components in this. How fast is BQ able to deliver data - you could for example test how long it takes to do the same extraction within BQ and store the result in a file (or actually it will be multiple files with that amount of data) in a google storage bucket? That's an indication of how long the BQ processing time is.
There's also the fact that it takes time to transfer data between servers. Network capacity between BQ and the Qlik server is a factor. If you have a 1Gbit/s connection that is fully used, 1TB takes about 16 minutes to transfer. But this speed depends on som many different things.
There's probably not much you can do about the actual data transfer time.
Transforming this amount of data also takes time, of course. Sometimes it is possible to optimize transformations but it really depends on what you are doing in the transformation, and if there are any alternate ways that would speed up the process (for example: applymap() is usually faster than joining tables, but is only approporiate in certain situations).
