I do tend to focus more on breadth, but I do go deeper on some subjects. However, I never give up breadth for the sake of going _Jon Skeet on C#_ level of depth. That, for me, is a waste of time[0]. I don't need to know a language or tool quite that deep. I can be very productive with a certain depth without touching bottom.
[0] Note: I'm not saying Skeet is wasting his time.
The other thing you need to be careful in with focusing solely on going deep on one topic is that that one topic can go by the wayside.
I've known people who were, in one example, arguably the world expert in performance on a long ago computer architecture. There came a point where no one cared any longer and I'm not sure to what degree he successfully moved on to other pursuits.
Another example is Y2K. A lot of consultants ended up defined as being Y2K guys and they didn't necessarily successfully transition to something new.
Not arguing that going deep is necessarily wrong but, if you do, you need to keep your eye on emerging areas that could benefit from your existing skills.
I do tend to focus more on breadth, but I do go deeper on some subjects. However, I never give up breadth for the sake of going _Jon Skeet on C#_ level of depth. That, for me, is a waste of time[0]. I don't need to know a language or tool quite that deep. I can be very productive with a certain depth without touching bottom.
[0] Note: I'm not saying Skeet is wasting his time.