Like most geeks I have far too much self-doubt to consider a start-up alone. Every time I have an idea I can pick it apart technically in minutes in my mind (Dunning-Kruger effect).

Really, who needs online file synchronization when a USB key and rsync are perfectly fine!

Try a couple of things that you'd normally discard and see if a bad idea doesn't lead to a good one once you get a bit of user feedback.

I have some IaaS ideas I've been kicking around but they are all capital intensive. The only way I can make it financially feasible is to build my own servers. And with no SLA or warranties I would basically have to live at the datacenter.