First movers don't always win.
http://staffweb.ncnu.edu.tw/clhung/MOT/Mohr02.ppt (slide 19 and on)
First movers have an advantage when they can raise entry barriers (economies of scale, network effects, switching costs).
However, fast followers or late movers have some important advantages, and if the early movers don't have significant barriers they can pull the rug from under them if they have a superior product or better business model.
IBM wasn't a startup -- it had massive distribution resources. When you have access to an enormous distribution channel and massive marketing budgets, the rules are obviously different.
> Microsoft didn't have the first operating system for x86 machines - Digital Research did.
Yes, but Microsoft piggybacked off of IBM's distribution channel. That's extremely rare, and 99.99% of startups can't swing that sort of deal.
> AOL once dominated social networking. Facebook moved into a quite mature space - Myspace, Geocities, and several others were big companies when Facebook started.
AOL, MySpace, and Geocities effectively committed suicide. It's nice when that happens, but you can't count on it.
There are no guarantees, it's all a balance of probabilities. The point isn't that the first mover always wins, it's that if you aren't the first mover the balance of probabilities is against you.