In the above titled article in Archives of General Psychiatry, the abstract concludes:

Conclusions  Advanced paternal age is a risk factor for BPD in the offspring. The results are consistent with the hypothesis that advancing paternal age increases the risk for de novo mutations in susceptibility genes for neurodevelopmental disorders.

Not so fast.

First, it is necessary to remind everyone that doctors-- let alone the layman and the media-- do not generally read these studies,  they read about them (like you're doing now.)  The few who do actually see the article will rarely go past the abstract.  That means that the abstract, and even more importantly the title, are the only information conveyed.

What becomes "known" in the field is a meme, a feeling, a gestalt, and in this case it goes something like this: "I have heard there is a lot of data on autism-- experts on autism seem convinced--  and now I see there is a study about bipolar, so I guess we are discovering that advanced age plays a role in psychopathology.  Probably the sperm is bad."

Awesome.  Suitable for happy hour conversation and Dateline NBC.

Back to the conclusion's last sentence.  This should properly read, "Advancing paternal age, or some factor it represents, is associated with an increased risk..." 

Worse, while this is an article about paternal age, it also slyly reinforces in your mind the authors' bias that bipolar is a "neurodevelopmental disorder" that may be caused by genetic mutations.  See? For them, the "hypothesis" is the part about the advancing paternal age.  The axiom is that bipolar is a genetic disorder.

I'm not saying it isn't a genetic disorder, I'm saying there isn't nearly enough evidence to be able to say that the advanced paternal age represents a genetic variable.  

Does advanced paternal age really increase risk for bipolar?

The data show that men over 55 had a 1.37 times greater risk of having a kid with bipolar than Dad's who were 20-24.  There was an even stronger association for those kids who had an early onset (before age 20)--  2.6 times greater risk.

The problem is that "advanced paternal age" could actually be a proxy for a lot of things: early maternal loss (children whose mother died before their fifth birthday had a 4.05 times increased risk of bipolar), or maybe even birth order.  Maybe there are reasons why the youngest born may be more "bipolar" than the oldest born.

There isn't enough/any evidence to determine if birth order is or is not relevant.  But neither does this study contribute any more than "X is associated with Y, but we don't really know what X represents or even what Y is."