Not so long ago the common wisdom was that free and open source software would grow strongly in the market for infrastructure software - operating systems, databases and web servers that have a generic applicability across all markets - but that non-free software would continue to dominate in the realm of vertical markets - where applications fulfil a specific role in a specialised commercial environment.
This view, which is still held by many, was based on the assumption that the "hobbyist" programmers who initiated many of the more famous free software projects tended to pursue sexier and more immediate objectives, and that many areas of applications software development would be ignored as a result. A second assumption was that companies would not find a free software commercial model that worked for more specialist business applications, where there is a smaller market for support services.
But realities change very quickly. Open source allows individuals and organisations with a diversity of skills to come together to scratch an itch or solve a problem. Individuals gravitate towards an open source project because of a particular skill or interest. But organisations in many industries have recognised a similar benefit at the corporate level - that the value lies in the final product, not the enabling software - and that it pays for organisations to collaborate with each other to share skills and lower the barriers on research and development.
The incentives may be different, but the effect is the same. John Sarsgard of IBM stated the case quite clearly back in 2003: "Is it a charitable thing for IBM to have 250 engineers working on Linux? Long term, we are getting a cheaper operating system than we can by building our own. It's self serving. We can't build a Linux class operating system all by ourselves with only 250 people."
The computer hardware industry sees direct benefits in supporting free software. The advantages of a free and scalable operating system are efficiency and cost. Resources can be dedicated to other activities, and development costs can be shared across the industry. The driver is open standards, which give both manufacturers and users a better deal. This cooperation is now reaching far beyond the bottom layer of the software stack. Companies such as Novell and Sun Microsystems, for instance, have contributed large chunks of code and personnel to the development of OpenOffice, Gnome and Firefox - and this phenomenon is set to continue into other fields.
Rising from the horizontal
The same arguments can be applied to other industries, and other parts of the software stack. For most commercial enterprises software is a tool, not the final objective, and there is a common interest among competing companies to share the costs of developing software that "just works", but there is also a willingness to pay for the services around more sophisticated open source products higher up the stack.
There are many models for paying for software. The success of the JBoss J2EE application server has proved that a free software project, "free as in spirit, not as in beer", can generate substantial revenues without compromising its principles, and that subscription, installation, training, support, upgrades and maintenance can provide realistic opportunities for creating income streams for a product that is at least the equal of, and arguably superior to, many of its rivals. JBoss, now owned by Red Hat, has not only become a dominant player in its market, but also employs most of the key developers behind an expanding portfolio of Java middleware projects.