Even if Scihub will never add any new paper again, it will remain extremely useful, because it contains a large number of older papers that are very difficult to obtain from any other source.

For the new papers, a significant percentage can be found as preprints in arxiv or the like, or are published in open-access journals, so the difficulty in obtaining them is typically less than for older papers that are still important.

Moreover, even if I pay a non-negligible amount of money for the access at certain journals, I frequently prefer to search for their articles on Scihub, because I can get them so much faster, without wasting time with logins or with searching for articles in the wrong place, because they were published elsewhere.