I think you underestimate how difficult scaling is.
So, you want to offer 2GB of redundant storage over the internet to everyone. Well, first you're going to need 6-10GB for everyone to cover redundancy. You're also going to need an amazingly high amount of bandwidth to cover it - especially at the start because new users are the ones likely to decide "oh, I want to see how well this works so I'll just test upload download random things for a while" because of the novelty.
So, you assume that you can store 4TB per server at $1,000 per server assuming you're buying crappy white-box stuff - really scraping the efficiency barrel here. That means that for every 400 people, you have to spend a grand.
OK, so let's say that only 1 in 10 actually use the product, that's $1,000 for ever 4,000 people. Still not great. Most importantly though, your staff is only so large. You're not just allocating more space on S3 - you have to be installing and configuring these servers - oh, and let's not forget that rack space isn't cheap. Without bandwidth or power, rack space can go for $1,000 per month for around 40U.
So, let's say they just open it up and 100,000 people sign up for it - not unreasonable. It would require $25,000 to buy the hardware considering a 10% usage rate (people only actually storing 200MB rather than the 2GB limit). Above that, we're talking about installing 25,000 boxes into racks and configuring them all. Above that, they need to pay for electricity for all those boxen. Let's assume dirt cheap power at 4 cents per kWh and decently efficient servers drawing 200 watts (neither of which are going to be true in real life). That's $200/hr to keep those servers alive. Going monthly, we hit $144,000! Great way to blow through VC! Oh, bandwidth, how could we forget about that.
I'm going to stop beating a dead horse right now. It is really hard to provide what Dropbox is trying to do. It makes a TON of sense to start slowly in their case. This isn't like a social networking site where it matters how many of your friends are on nor does it have the luxury of storing very little data. Most importantly, unlike the Web 2.0 stuff out there, Dropbox's usefulness is harder to do. While many social apps are highly polished, Dropbox has a much higher technical barrier. Specifically, you need to buy a lot of iron to run it. This isn't something where you can use S3 (or a smal MogileFS cluster) to store/serve icons or some user photos. Do you think someone else is going to come along and compete with Dropbox? Sure, but not like they would with something like Twitter.