I agree with wrench- we do small games for corporate clients and as of the beginning of 2013 they are asking for HTML5 games, even with the performance problems and development difficulties vs. Flash. They just need their web games to run on mobile, period. I love AIR and still see it as a great technology, but the problem is it's geared toward standalone APPS, when it appears we are moving in the direction of pure HTML5 web apps as the standard:
http://www.businessinsider.com/html5-vs-apps-heres-why-the-debate-matters-and-who-will-win-2013-3
Adobe got stung once by Apple/Jobs on Flash and assumedly doesn't want to get creamed again betting on a losing horse. So it appears they are dumping everything they have into HTML5 web tools, while supporting AIR and Flash less and less. The big variable here is how long will it really take for HTML5 to catch up to Flash in terms of performance, and cross-platform ease of development. And who even knows if a massive, open-source, standards based collaboration like HTML5 can really ever match what Adobe has done with Flash/AIR?
Basically we are in a grey area right now and you have to be smart about picking the right technology for the job you are doing - consider the clients needs, and how future-proof it needs to be. As of now, Flash/AIR/Starling/Stage3D is still a very good fit for standalone APPS, especially games.