September 2008's issue of Neuroinformatics, an Open Access online issue (one day this will be tautological) focused on the Neuroscience Information Framework, a wide assembly of online data and tools for neuroscience research.
Right now, besides maybe Google, scientific projects are what push the envelope of really large-scale, distributed, mathematically complex data-intensive processes. Most likely, business and governments will eventually follow suit, which will probably trigger another minor revolution equivalent to the introduction of the Internet.
Don't forget that the World Wide Web was first created at CERN (the people behind the Large Hadron Collider) precisely to deal with some of the problems of globally distributed research. In a way, the World Wide Web is but a collateral side effect of scientists' struggle to collect, analyze and distribute exploding amounts of increasingly sophisticated data.
Imagine the side effects of what scientists are building today...