As somebody no longer attached to a university I rarely have the money to access all the papers I want to read. Sci-hub is the way that distribution should be. Accessible to all, free for all. Restricting access to published leading edge research just slows progress and hampers self education. I dream that some day the other aspects of journals can be served by software similar sci-hubs open access. The only people that would be upset about that would be the journal publishers themselves.
But, there are some key features needed to dethrone journals that aren't easy to accomplish. Sci-hub only satisfies the access requirement. To dethrone publishers also requires features for publishing, review and discovery. Some universities might still want printed journals but with all the on demand printers available that is something that could be satisfied easily. I think the only way this really works and doesn't just devolve into another pay to play journal is a federated system where paper producers run a server to publish and host their own papers and can subscribe to other producers' servers.
