This business model was actually rampant during the dot com bust. Company A thought it would provide an email service for free and once it has a million users it could cross-sell other services or add on some premium features. If even 10% users sign on that would still generate a good residual monthly income from a 100K users. But as soon as Company A goes premium another company would come out with the same model offering the same services for free and Company A's model collapses. Company A then might resort to increasing switching costs and locking people in (even though email has a high switching cost this didn't work but that's another discussion).
The reason I bring this up is that this is exactly what you could be facing down the line. You could make a strong case for sustaining for 5 years but if a company comes out 4 years from now and does a really good job of the same thing, you've just lost a lot of time and money but they got in cheap. Even if your service's switching costs are high and your users are locked in, it's entirely possible that the new company might offer a migration path to circumvent just this scenario. And you could then take measures to really hijack the user's data but that's not that easy and it would lead to negative publicity from frustrated users (they feel their data is theirs to take).
Since your question is quite general, I can't provide much feedback but unless the scenario is really compelling it's a hard road to tread upon. Even if Larry Page approached me with a search engine idea, I'd want to know the monetization strategy (but patents provide a safety net--see below).
There is also the question of employee morale. Employees don't get nice perks and bonuses because you're always strapped for cash and they always feel that the reward is too far out. You will end up giving out a lot of stock options and raising multiple rounds of investments which won't be easy.
Patents gives you a safety net in that when five years from now it becomes feasible, you can execute on your idea and still have exclusivity. And even if you fail to execute (or don't execute well), you may take comfort in the fact that someone else might be able to execute better, while you can get dividends.
Despite that, if you really want to get this thing out there, you might be able to throttle your costs by making it an invite-only beta so you won't end up mushrooming your costs. It will also provide you with a platform that you can mature over the years and as hardware becomes more affordable you can continue to issue more invites.