Software's Top 10 2005 Trends: #3 Software As A Service

Software as a service has a bit of a bad reputation thanks largely to the Application Service Provider (ASP) debacle of the mid-1990s.  Back then, a huge amount of money was poured into ASPs such as US Internetworking, Corio, and Interliant in the belief that companies would like to buy enterprise applications as hosted services

As it turned out, not many companies were interested in buying mission critical software as a hosted service. Not only did hosted apps have significant security and performance issues, but they were difficult to customize and integrate into an enterprise’s other systems. On top of this, the economics of providing enterprise software as a service were terrible. ASPs were required to make huge upfront investments in hardware and software but they only got paid a monthly subscription (which could often be cancelled with little notice).  In addition, competition from other venture funded startups drove pricing down and drove customer churn and factor costs (such as hosting and bandwidth charges) up.  This turned out to be a recipe for disaster and many ASP’s ultimately ended up going bankrupt under crushing debt loads.

Fast forward to 2005 and things are now actually quite different.  First off, the main costs required to offer software as a service have declined dramatically.  As Ryan McIntyre outlines in his excellent post on data center economics, datacenter/hosting costs have plummeted in the last 10 years with bandwidth costs declining 88%, storage costs declining an amazing 99.7% and CPU costs declining an even more amazing 99.9%!  (When you think about this from a business economics standpoint it really is stunning.)  With lower costs, the upfront investment required to offer software as a service is actually now quite reasonable.

Outside of lower costs, three other developments have helped make software a service much more attractive.  First, developers have created new applications that have been engineered from the ground up to be offered as a hosted service and even many existing applications have been re-engineered to make them more “hosting-friendly”.  Second, the advent of XML and web services has made it easier for companies to integrate hosted applications and data into their own legacy systems.  From a technical perspective, this has removed one of the last major drawbacks of hosted software.  And finally, 10 years of exposure to the web has made many corporate managers much more comfortable with the idea of hosted-applications.  Even many IT managers, who at first resisted hosted applications as a potential threat to their jobs and influence have now warmed up to hosted-apps as a way to quickly meet business unit needs without adding significant costs to their own organization.  For many developers, selling a hosted software solution is now an easier and faster process than selling installable code.

On the strength of these developments, 2005 may very well turn out to be the year that software as a service goes from being an alternative means of delivering software to being the preferred means.

For VCs, the growing acceptance of software as a service combined with its improved economics raises the possibility that it is very risky, if not downright ill-advised, to fund any new enterprise application that is not designed primarily to be offered as a service.  It also raises the prospect that new applications architected from the ground up to be offered as a service may be able to displace well established client-server players in the ERP, database, and vertical application space much as those players did to mainframe and mini-computer vendors in 1980’s and 1990’s.  Indeed some categories of applications, such as inter-enterprise applications, are not even possible unless offered as a service.   All of this adds up to make enterprise applications a much more interesting investment space than it was just a few years ago.

For a complete list of Software's Top 10 2005 trends click here.