So the problem with this kind of reasoning is that you deny what happens to companies after they are acquired. Look at Keyhole, which became Google earth. This was a small company in Maryland with a small audience of paid, mostly government and enterprise, users. They were purchased and in a remarkably short amount of time, ported their imagine serving to googles infrasctructed, expanded coverage and deployed the free Google Earth to tens of millions of people worldwide. This would not have been possible without the use of Googles operations and data center people and hardware, or the storage technologies that uses them.

Similarly, look at Android. Android was purchased very early in their development cycle, and literally grew to a team of hundreds at Google, where the bulk of the work that went into that took place.

This is not to discount that we buy companies because they are doing something interesting/cool/new/smart, but to discount what happens after denies a very large part of how large businesses work.

Maybe what I'm trying to say is if Google can take a legitimate hit when an acquisition goes poorly (dodgeball comes to mind) then maybe it is worth considering what happens when acquisitions come off.

To actually answer your question, gmail, pr, gfs, bigtable, orkut (As a stealthy side project kind of thing), google news, finance, google talk, the igoogle homepage and the gadget spec therein which lead to opensocial, streetview, containerized and large scale data center operations, etc, etc,etc..

But that also discounts the value we've seen from our adoption of open source technologies and the day to day work of thousands of engineers.

Maybe what I'm trying to say is that (ugh, I hate this word) innovation is wherever you find it, and a decent company tries to find it everywhere.