That said, I don't necessarily disagree that in the future MANY apps will be delivered using powerful browsers. The multiple platform argument works in that case -developing for a browser that works on many devices is much cheaper than developing multiple versions of software. Google is of course, the kind of company than can build the various browser versions needed for each platform, then leverage that commonality to deliver apps.
In the medium term, I think it will become apparent that this is the trend, but we will still need some kind of local, device specific access to store data or the app when the network is unavailable. A fail-soft capability will be important in those cases.
Considering that most applications are not mission critical, we all deal with dropped calls and network outages all the time, the success of this model will be tied closely to the quality of network services, both at fixed points, and of course in mobile applications.
In the long term, I agree that the majority of apps will be browser based, but they likely won't look anything like the browser we know today anymore than the modern PC looks like an Apple II. Same idea, quantum leap in capability.
Whether or not his vision is true will be the topic of hot debate for the next few years until it actually happens. Or doesn't. Should provide entertaining discussions for quite a few years.