I think Google's biggest risk is in not owning enough of the most important corpora of web content which are getting created today: social information. The Internet is being reinvented as a network of people instead of just a network of publications. Facebook and Twitter are leaders in owning this people-content, but not really in searching it effectively yet. Even though Twitter is viewable (i.e. crawlable) and open, they still have proprietary data which could provide the best overall understanding of the information and relationships which flow through the Twitter platform.

The technology of web search continues to get commoditized today. Google is quite good at it, but Bing/Yahoo still do alright. Whoever provides a differentiated and important corpus of content for some fraction of web searches (i.e. social search or something like "real-time" or news) has the potential to capture the search behavior across all queries. Users only want to type searches into a single search box on the web.