I think you can take that as a given. If you look at the GCHQ leak - they're basically just recording everything (including content) for 3 days, and keeping headers for 30 (shared with the NSA of course). That would give them most websites visited by an IP (which would take hardly any space to store, but are still really intrusive).

The only things preventing this from being a total capture of all information (to be sifted through later) are technical issues with storage, not moral or legal ones.