I like big storage, I work with virtualization in its various forms so that all the hype about "cloud computing" sounds like "well, duh!" to me. Still, I think it's wrong to look at it as the possible future of IT in general. We have worked hard to move away from centralized computing and I think we should not go back.
As I've written previously, I want just the opposite of the current "Cloud" fad - I want radical decentralization.
Instead of huge data centres owned by a few large corporations like Google, IBM, Dell, Fujitsu, Oracle, etc., I'd rather like a completely distributed p2p environment in which all data actually is in the cloud, but the "cloud" is taken literally - analogously to a coloidal mixture of all the particles of water vapour, a connected, non-centralized mesh of all devices, each providing storage and computing for all others.
Yes, I read a lot of SF, but nowadays it actually seems possible. My new mobile phone is at least in some tasks faster than this old (old! one year old!) netbook I'm writing this on, and it has better connectivity (3G HSDPA). We have a huge library of cryptographic algorithms that can ensure privacy of data (not to be confused with DRM, which is already proven impossible to do mostly because of physical laws of the universe), and we have started to discover many new forms of routing data reliably (p2p), securely (SSL) and, if necessary, anonymously (Tor).
What is missing is the initiative to put all these pieces together in a way that will not artificially introduce monopolies (like what happened with SSL CA-s and Telecom companies monopolies on wireless bandwidth). In essence, the original idea of the Internet which really was more P2P than C2S (see the definitions of early core protocols like SMTP, FTP, NNTP) should be continued.