Imagine a web "page" where it was just as easy to embed audio or video as it is now to insert an image.
Pages are more structured with "sections", "articles", "headers", and "footers". They can include a "nav" section.
Embedded data is possible with native "datagrids" and "datalists" andRSS feeds embedded right in web pages.
More interactive elements with server-side "events", "progress", and "output".
Elements like <center>, <font>, and even <tt> will finally be removed (instead of just depreciated) – even <frame>s finally meet their worthy demise.
All kinds of new APIs, including persistence/database, audio/video, drawing, and more.
There is a high-level document describing the differences at the W3C site – still just a draft but quite a look at what might be to come.