We stand on the precipice of major change.
Cloud based applications will eventually decouple applications from any particular operating system and nullify OS lock-in. It is easy to debunk the utility of cloud based applications as we base it on a model that is only apparent today, failing to see it evolve out of necessity to address consumer concerns regarding privacy and ownership. As soon as one vendor solves this problem others will follow to remain competitive (and for the most part it has already been solved).
I believe the future of the cloud, for businesses and the consumer, is in hybrid applications, that are both native and available over the web, that store data locally as well as synchronise when on-line, with strong encryption and security and practically unlimited online storage(for a price).
Do not make the mistake of imagining that access and bandwidth in the future will be as limited as it is today. Bandwidth is increasing exponentially, and the amount of data points, at a minimum, around major cities will increase. You may lament you have limited access or speed right now, but this is unlikely to be the case in 10 years time. Today's cable and ADSL connections will feel like 56k baud modems popular in the 90s.
Google, AND MANY OTHERS, have developed and released technology that addresses most of your concerns. Many companies already aim to provide native compilation of apps and offline databases as well as browser based graphics acceleration. I've personally written apps using my own hybrid technology that run locally as well as on the web.
I believe app stores and vendor lock-in are a different matter entirely to cloud computing.
But let me take my view of the future much further. With web applications becoming increasingly ubiquitous, we will find many new Web based desktops taking advantage of native code compilation and graphics acceleration. These will relegate the OS to a layer that merely runs applications launched by URLs that in turn open browser based windows. To all intents and purposes web desktops/Window managers, customised to your own usage, will appear to perform as efficiently as regular desktops do today, on any system anywhere you are in the world. The era of the native desktop as a necessity or differentiating factor may end some time in this decade.
The major driver for Cloud computing is business and their desire to provide SaaS(Software As A Service) so they can pick up regular licensing fees and curb piracy. Personally, I see nothing wrong with this model which continues to keep the IT industry afloat, not to mention keep me and many, many developers in a job.
