IMHO the cloud is going to be the biggest change in the way we use SQL Server since the original Sybase port. I think it is likely that by 2015 it will be hard for IT to justify the continued use of any application that is not in the cloud.

The main driver for moving to cloud is cost reduction. From what am seeing, cloud-based SLAs will be much more standardised that what we are used to, with the end-users seeing the cost implications of their choices. Standardisation of services means less staff needed to manage them, and possibly with less value placed on troubleshooting skills. The end result is far cheaper IT for those applications in the cloud compared to traditional methods of operation. IMHO IT will not be given the budget to keep things in-house when they can be put in the cloud.

Writing apps for the cloud uses an offset of the skills needed for in-house hosting. This will mean it will become harder to find staff who can maintain in-house apps, making them seem more expensive and less flexible than cloud apps. Those old enough to have experienced the move away from mainframes in spite of their superior facilities know how quickly things can move if the bean counters want it to, even if the people close to the technology see more problems than solutions in the move.