Just don't pick things to specialize in that are guaranteed to date themselves. People who did program transformation masters theses in the mid-90's are just now coming into their own in the industry.
Specialization is not free; it decreases flexibility, and would seem that it would decrease cross-pollination of ideas from different areas of knowledge. It increases income, for sure, but also increases the attendant risk that people won't have need for an X specialist right at that moment, or at least no one you would want to work for.
I'll grant that generalization is not free either. Someone who really needs an X specialist will pay top dollar, because the supply is likely pretty low. And people may not trust a generalist to go and implement some specialized X performing application.
I think there are opportunities and attendant risks for both. I really don't think either one wins outright. But, I'll be happy to listen to words of wisdom you might have on this.
Personally, I am a database specialist. I do have skills in other areas (I do python coding as a hobby and I can write C# code that at least works), but my strong suit is definitely in databases, and even with databases I focus on MS SQL Server and Access with just a little experience in some others.
I personally chose to do that because databases appealed to me and by specializing I could gain a depth of knowledge that very few generalists can. In terms of careers this limits my options somewhat, but it means that within my particular domain I can out compete most others. When the company needs a generalist that can do a lot of things, I am probably not their best choice, but when they need someone with a deep understanding of SQL Server and relational theory then I am a good candidate.
Its all about trade offs really.