In the traditional business, I suppose the answer would probably be yes (although I haven't really seen anything that supports this). Working in tech however, is much more of a crapshoot. Most of the time, you can't really research a market you're creating. This alone makes the whole process a giant guessing game.
If forming a business plan is just a series of guesses, you'd be better off moving straight into the part where you're actively testing those guesses with a workable product.
What if your new product, that has never been done before, will costs tens of millions? Surely those products have existed, and yet are still somehow able to do the prudent 'guesswork' to convince venture capitalists.
Bluntly, my opinion here would be that this isn't a viable option for a startup, unless they have access to easy angel capital or can self fund to that amount.
Developing an idea with that price tag is an is a risk orders of magnitudes higher than a low cost startup. Having a hit is hard enough, but you'd need a run-away smashing super hit to even break even.
Even if you choose that path, there are still aspects of the idea that can be harnessed and build at much lower costs to begin income generation to help fund the remainder of the idea.
Investing money of that amount into an idea that has yet to prove itself is not my idea of a good move, really.
EDIT - Curious, what problem space are you thinking about/working in that requires that level of funding?
If you look into startups in the medical, pharmaceutical, chemical and energy fields, you will see very high startup costs while sharing the high risk attached to software startups.
Even in software though, there are companies working on solving difficult problems that would require the work of scientists, engineers and administrators for 1-2 years. This would place you into that multi-million dollar category pretty quickly.
But, how feasible is it to start companies this way? Sure it may have worked well for them and it may even work incredibly well for your business. As hackers though, we know that just because something worked once, twice or 500 times doesn't mean it's the truth or will always work.
It's best to be prudent wherever possible and do some research into your market and pricing strategies. I'm not saying go off and create 5 year business plans, I'm just saying it definitely deserves some of your time cycles.