There are a number of firms out there that promise "write once, publish to multiple platforms as native apps." Those companies include Appcelerator and Rhomobile (among others). Add Adobe to that list with the release of Adobe Air for mobile. The company's vision is that developers will use Air and create cross-platform "native" apps for multiple smartphone operating systems simultaneously.
By no means an original vision or idea, if this can be achieved by Adobe it will likely mean that the expected migration from apps to HTML5 (mobile Web apps) will not necessarily happen or, perhaps more accurately, happen as quickly. It now becomes a race between Adobe and HTML5 as the company struggles to make itself relevant in mobile. (There were suggestions that Adobe was trying to prevent the current HTML5 "spec" from being published.)
Apple has blocked and publicly disparaged Adobe Flash for mobile. Adobe's announced relationship with Android is a bit of an intentional poke in the eye to Apple. But back to the central question at hand.
This "Air for mobile" announcement and its promise is an example of why mobile is so hard to predict at the moment. People assume that apps won't survive but they could we under a true cross-platform publishing tool. Apps offer a generally better user experience and if they're easy to publish across the top platforms developers will do so for obvious reasons (they also can't make any money off mobile Web apps).
That doesn't mean that HTML5 or mobile Web apps wouldn't develop as well. Indeed, mobile websites (Web apps) will probably be developed in HTML5 going forward. It just means that the obituary for apps is perhaps been written too soon.