Anyway one thing that really blew my mind the other day was Amazon's new S3 service. On the surface it seems really simple, a basic web service that provides cheap online storage (US$0.15 per Gigabyte for storage per month). What is exciting is the ramifications, if somebody (like you or I) want to store a lot of information for ourselves or others online there is no need to invest in big servers and fat Internet connections to serve that data. It also means that by design your web applications will scale effortlessly at least in the sense of the data storage mechanism assuming Amazon's server farm is up to the task. I have signed up for an account and read through the documentation and some of the features are pretty nice (access control lists, time sensitive url's and a lot more). There has been very good things said about it on TechCrunch and other places, plus some criticisim for not supporting the very simple XML-RPC protocol (which would have been very nice to have in simple applications).
What is needed now from someone like Google/MySQL is a distributed database platform that translates a similar business model across to the relational database market, say US$0.15 per thousand queries or database service in exchange for adverts on the site. It would be great to be able to harness to a massively scalable disk and database array with guaranteed uptime and almost unlimited scalability without having to spend thousands (or even millions) of dollars on server hardware and support. This type of setup would be really exciting and would benefit almost everyone involved (except for out of work server support staff). Whether or not something like this happens is uncertain but it certainly is something that Adam Bosworth from Google has obliquely hinted at in various speeches and blog posts.