I was talking to an IT 'consultant' the other day and I tried to start talking about my interest in SCA, BPEL, etc. and he quickly silenced me with blah, blah, blah, that's not important...It's all going to the cloud. Programmers, developers are obsolete...It's all going to the cloud.
Whoever told you that is an idiot. I don't know how long you have been doing this but I've been hearing about new technologies that will make developers 'obsolete' since my first year in out of school. It scared me the first time. It scared me a little bit less the second time. Now I just find it funny. If anything more developers will be needed. Unless the cloud is a sentient self-programming entity, the cloud doesn't make programming obsolete. There will be plenty of need for programmers for cloud based applications. If anything, IT programming in the future will be thought of less as a special task separate from the business function and more of a core part of that function.
Ok, fine, guess I'll get a new career. So, yeah, I agree, the hype is overwhelming. But, consider this. I wanted to try the some of the Oracle SOA Suite tutorials (don't flame me). That requires an Oracle database, an Oracle WLS cluster of at least one managed server, with an admin server, JDeveloper, and my browser at a minimum. Plus, I'm running an email client, all of my corporate security, corporate spyware, corporate IT required software, etc. The corporate laptop running XP just isn't up to the task. So AWS EC2 is my only hope. It's cheaper than having to go out, buy an expensive laptop for personal use just to try out a couple of new products as they come onto the market. I think you're making this point when you say that new startups should use EC2 rather than invest in a lot of expensive hardware. I think the same thing applies to individuals, like me, who just want to try something different, but don't have the resources available either personally or via corporate support. Not trying to hype the cloud, but it does offer an alternative to small fry with an idea and no place to try it out.
I absolutely think that in the future more an more of the backend of systems will be cloud-based. I'm not sure we are ready for this in terms of security or reliability but it's happening anyway. The pioneers will get burned and after the hype wears thin, more companies will go that direction.
What I am questioning is the assumption that users want all their client applications to run in the cloud. I think users want their applications to connected to the web and all it has to offer but I'm not convinced that people won't want programs that run most of their code on hardware in their hands or in their laps.
If people truly did want less computing power even as it got cheaper that would make computing power a inferior good or even a giffen good in economic terms. That's an odd sort of situation where you want less of something as it gets cheaper and your income goes up. Normally that only happens with things you have to buy (like food) because that's all you can afford.
Hosting applications on the web is good for the producers but not necessarily the consumers. I think it's important to not forget that as we listen to people so confidently proclaim the future.