I think siloed thinking rather than over-specialization is the biggest threat.  As people climb the greasy pole the Dunning Kruger affect comes into play.  Expertise in one area leads to over-confidence in another. Where are the next generation of good solution architects coming from?

I have made it my business to learn a little bit about a lot of different things outside of my core discipline.  Not because I want to be an infrastructure guy but because I want to understand what their stress points are.  For a small amount of effort and pragmatism on my part can one of their major headaches be avoided?  Ditto any other IT discipline.

Walk a mile in another man's shoes.  My experience is that the amount of effort to learn the basics of someone else's disciplines is greatly appreciated.  It fosters trust and with trust comes much more open communication.  Sometimes an external perspective leads to a radical shift in thinking.  I've certainly benefited from people with little data experience asking questions that people with data experience probably wouldn't ask.
I have found that taking the effort to be approachable  pays back far more than is invested into it.

It think over-specialization comes with its own risks.  It's the old joke about going to university and learning more and more about less and less until you know a hell of a lot about nothing.