If you do some research and then want to publish it in a journal, there is a large scope of work on the publishing side which
includes hosting. Scihub just redistributes what journals have already spent the money to vet, edit, proof and publish. The "cost of access" is paying for all of the aforementioned work. So there is a lot more to do, and pay for, that is not being captured by people looking from the sidelines going "why is nobody giving me free cake?"
If all you want to do is say "we have pre-pubs on a free FTP server", sure, that's cheap as hell. But it doesn't replace the journals, it doesn't address research funding models, it doesn't move scientific progress forward.