Kaspersky gives a sober assessment at ZDNet. Some applications make sense for the cloud and some don't. Expect mixed systems and mixed results. From my perspective, the scaling issues touted such as IT talent and the failure to determine if cloud computing suits the business model and particularly the business environment for a given business will be the reason for many cloud initiatives to fail.
The croc beneath the surface will be the ever rising cost of the services particularly those that rely on outsources for data such as mapping. Subscription-based data sources tend toward exponential increases in costs.
A cloud service is to the enterprise what an apartment is to newlyweds. Very pleasant and effortless until the size of the family exceeds the allotted space and the drive to drop the kids off to all of their social/educational activities consumes more time than the parents have. Eventually they buy a conveniently located house and learn to fix their own plumbing.
IOW, services don’t automagically expand per given resource over a given customer and this flattening removes the ability to compete at the mid to far end of the contract.
It has been illuminating to watch some try to cope with the limits of Facebook only to find themselves giving up in frustration and returning to their own self-managed resources. It has been illuminating to watch the costs of acquiring core data such as mapping double and triple if the buyer offers a public service in the face of other deals made by the seller (keiretsu can be evil). I don't expect an industry that considers the Webby Awards to be relevant and Streamys not to be to actually get this. I expect more lemming buys and a continued full-court press in the Beltway to sell these concepts.
But IBM's Bob Sutor put his finger on an important point: it is time to quit talking about racks of servers and talk about the applications to which I add in the context of their current market conditions. OTW, our heads are not just in the clouds.