Well per my observations there are at least two cases when sticking with software company borders to matters of professional survival.
First case is if one is totally into coding - give me 80... 90... 100% time to code or I die. At software shops, this is almost a granted, as if everybody knows how to get there because, well, because everybody is doing just that. But outside, there is a really high risk to fail getting there. One may get as low as 50, 40, 30% (my personal coding load once dropped to 20% - no kidding, I measured in JIRA!) It's not because "they" don't want you to code - no they want but, but... they may just don't know how.
Second "deadly risk" is if someone has serious problems in communication. This may be troublesome even in software shops, right, but at least there are good chances to survive and live a nice productive life without disturbing interactions. :) At non-software companies though, such chances are much lower - quite an opposite, it's almost inevitable that one will eventually have to spend much effort to educate some outsider in IT basics just because otherwise it will be impossible to do the job.
Well except for two cases mentioned above I know no other strong reason to strictly lock self to software companies. Now, which side to prefer? as far as I can tell, this is more a matter of taste, of what kind of fun clicks more on you.
Both sides offer their own, distinct forms of getting fun. It's not easy to describe.
I'd say software companies are more fun to those aiming at "high marks" while outside ones give a thrill to those striving for "big difference". I think of it about like this (note numbers below are invented only to simplify making a point)...
- At software firm, one has a chance to deliver 100 features a year - a highest mark nobody yet achieved. It's gonna be hard, it's gonna be tough, it's gonna be top - making a cool 50% improvement over average 70 features a year. Great challenge, really.
- At the same time, in outside firm one has a chance to deliver 50 features a year - a highest boost nobody ever achieved. It's gonna be hard, it's gonna be tough, it's gonna be big - making a whopping 500% boost over average 10 features a year. Great challenge, trust me.
Note btw that chances to get 500% boost at software company are negligibly small in comparison - and respectively, chances to achieve 100 features are negligibly small outside.
Top marks at one side expand our understanding of professional limits, improving our knowledge of how to do things better. Big difference at other side makes a deep impact on the company culture, improving outsiders knowledge on how to to it right.
Now, if you have a clear preference for one or another, you know what side to take. Or, if you're indecisive, just feel free to swing between them as you wish. :)