I have read a while back that one of the advantages of normalization is to save disk space (since normalization minimizes redundancy), but is saving disk space still considered an advantage of normalization in current time where disk storage is very cheap?
Normalization was never about saving disk space. In his 1971 normalization paper E.F.Codd wrote: "It is important to remember that we are not making a case for or against any physical storage structures in this paper" (emphasis in the original).
Normalization is all about the best logical representation for data and by no means implies that a normalized database will take up less space than a denormalized one.
Given the range of advances in database storage techniques since then (e.g. compressed formats, join indexes, columnar storage) there is probably even less correlation between normalization and storage size today than there was 50 years ago.
Answers originally left as comments on the question:
One of the most important benefits of reducing redundancy is the fact that you can reduce considerably (or eliminate) the risk of inconsistent updates: that this the possibility of updating a data in only a subset of the places where it is repeated. – Renzo
I would say that it never was the main purpose of normalization, but more of a nice side effect. Like Renzo says, the main purpose is to avoid possible update anomalies. Despite the rapidly increasing amount of data stored, I don't see much discussion regarding normalization and disk space. On the contrary, new architectures often involve multiple copies of distributed information. Regarding disk and storage, quite a bit of effort is invested in compression of data and techniques to evaluate queries against compressed data. – Lennart
As others mentioned it was not really about space savings. And if it was, it is less disk space but more page cache and bandwidth savings. – eckes