Google has essentially decided to make “an extra copy” of the entire World Wide Web (or at least the HTML and images of it) and let you run it off their servers, because Google’s servers are always faster (Blogger and Orkut aside, of course).
Dig even deeper and the picture grows to reveal a whole nother world of possibilities. Google is really offering to replace the web, wanting everyone to use their copy instead of the public copy. While the World Wide Web is currently a decentralized network of nationwide servers, Google wants the whole web to run off its computers, in one of its anonymous, nondescript data centers.
Beyond the obvious concerns that by browsing the Web (or should I say “Google’s Web”) Google can monitor your activity, Google’s attempt to replace the Web with the cached version on their servers conflicts with the vary nature of the Internet and the Web itself. Instead of it being a distributed network of independent computers, the Web (as you might access it via GWA) now becomes merely pages of content housed on Google’s own private servers (which has its copyright issues as well).
More from Jeff Jarvis here, along with many links to other commentators.