1) Games 2) Mobile apps 3) Video
#1 I agree that flash is better for making games, but how big of a market are games on the internet? It's a 'big' market, to be sure, but how big compared to the number of Flash websites a few years ago? Surely it's a low percentage (anyone have actual stats? I don't know of any). Is the amount of Flash games being made enough to keep Adobe interested in continuing to improve Flash? As the desktop market recedes and mobile devices become more popular with users (and I truly do believe that it will) what will happen to the Flash games?
As for #2 - mobile apps. This could be the answer to the above, but why use AIR when you can get better performance writing native code for whatever platform you are working on? I'm sure it will work out for companies that want to save a few dollars and be able to launch a simple game on many platforms while writing it once, but that kind of development has rarely succeeded in the world. Look at Java apps on windows/osx - they often feel klunky and look ugly and don't support the native UI elements of the host OS (does AIR do this on mobile phones? I'm admittedly not that familiar with it). Because of the history of this area of technology, I'm highly skeptical.
#3 I agree with, and covered this in a comment below. I believe Flash will be useful for desktop video delivery for a few years to come, but will eventually lose to native support.