I'd guess that computer science is very different from experimental physical science disciplines: In experimental, you have to have a research advisor with grant money if you want to have vacuum pumps/oscilloscopes/glassware/liquid nitrogen/etc., even if you plan to TA your way through grad school.
Of course, I have heard of R1 tenured professors who claim to spend 50% of their effort procuring funding (though I think 10% is a much more representative estimate).
And I'm not dissing them in any way - getting a grant requires the senior/experienced guys working on it. Getting the funding is a must-have (since without it researchers and students would be dismissed and the research would not exist); but spending your own time researching is a nice-to-have thing that you can do if there's any spare time left after teaching and bureaucracy.
Problem is, you can't take that route to get tenure in the first place, unless you're an absolute genius.
You're not going to be able to do a good grant proposal unless you have a good track record of research, with multiple papers. So in that sense, a grant proposal supersedes a single paper.
But, given that you've already done all those papers, a grant proposal is just writing, not new research, so it should actually be massively less work than doing all the original research that goes into a paper.
I'm not so sure that's a valid generalization.
I'm getting a PhD in computer science, but any corporate R&D department I know of that would hire a fresh graduate would only let you do R in service of their particular D needs, i.e., you have too little autonomy.
I definitely think I can find jobs that I love and that are lucrative (I have some in mind), but I won't be doing research as, say, a tenured professor would define it.