I think this is precisely the point: traditional, browser-based Web (text-based, asymmetrical...) is designed to work perfectly with low-bandwidth connections. With the advent of broadband, it becomes obsolete and is to be superseded by modern technologies which better utilize this type of connections.
Of course, Web is not dead by any means; but Wired is in the business of selling tech-related sensations.
I know many people are excluded because they don't have iOS running on the machine they use for the web, but not everyone had the means to surf the HTML/Flash-based web on Opera's web browser a while ago. Were they not part of the web?
All that is no news, and the whole concept was very hip back in the days when XML and Web services were all the rage, some ten years ago. In my opinion, it's not about who will take over -- we will simply have different forms of content transfer over the Internet, and will decide on a case-by-case basis which is better in each particular situation.
Here's how Ajax works: A whole page is downloaded once from the server, then the page's Ajax functions start running — and then when part of the page needs to be updated, the Ajax toolkit only has to reload that specific data and render it on the page.
The problem I see with the graph is it's based on percentage of total traffic but it doesn't take into account a huge increase in bandwidth and internet users. Put it in perspective and you realize the web's "peak" according to this graph was when most people were using 56k modems (2000 which was about a year after V.90 modems started hitting the street).
There's also an efficiency factor. Ajax type technologies have dramatically shrunk the size of the data chunks being passed back and forth. Back in 2000 most were still resending the whole page on every refresh.