Several of the businesses I've talked to heard about the Application Service Provider (ASP) melt down in the early 2000s and are concerned on demand software (aka: Software as a Service) will follow the same path.  While no one can say for sure this will not happen, the benefits of on demand software and market acceptance makes this scenario highly unlikely. 

The ASP model failed because the economics didn't make any sense.  For those of you that aren't familiar with the ASP model, it involved hosting and managing dedicated servers for a customer's application (i.e. SAP, PeopleSoft, Oracle, etc.) in a third party data center.  From a cost perspective, there was very little savings since the number of servers required remained the same and software license costs were no different.  Unless a company did not have the skills in house to manage an application, this model merely moved the management application headache from the customer to a third party.  In some cases, the headaches remained the same or became migraines.

The on demand model is fundamentally different because the software is designed to securely support multiple customers on the same instance of the application.  As a result, the software company does not need to develop software for multiple environments (i.e. Microsoft, Linux, Unix, SQL, Oracle, etc.) and they do not need to support all those older versions.  Let's face it, upgrading software and hardware can be a nightmare so companies usually held out as long as possible.  With the on demand model, all customers receive small upgrades every couple of months eliminating the software provider's cost of supporting multiple versions.  While it's true the on demand model has costs that are not associated with "licensed software", these costs are relatively minor compared to the development and testing costs associated with supporting multiple environments/versions.  Most businesses love the low risk associated with on demand because there are typically no upfront costs and most on demand software has a trial period. 

The investment community is also a big fan of the on demand model because it generates a predictable revenue stream.  It's not unusual for a mature on-demand software company to have 75% of their annual revenue booked before the year even starts.  I can't think of a business owner that wouldn't like to be in that situation.  As a result, well over half of the software companies receiving funding are developing on-demand applications.  The number of applications will grow exponentially over the next several years giving businesses a wide selection of applications to choose from. 

The challenge for businesses will be selecting quality on-demand software with a strong management team.  Choice is a good thing if you have time or a trusted adviser to help you wade through your options.