Making Money with Open Source

The question of whether a business can make money with open source software has most often been answered with an uninformed "no," at least until recently. As open source becomes mainstream and an integral part of the software industry, the common answer to this question is becoming, "yes, but I'm not exactly sure how." This was the topic that Joyce Chow (Apple Inc.) and I addressed in our presentation on Open Source Business Models at the Practicing Law Institutes seminar on Open Source Software in San Francisco earlier this month. Here is a brief summary of the highlights of our presentation.

To start, calling open source a "business model" is not accepted by everyone. Open source is clearly a development and license model, but it does not directly derive revenue. It is best viewed as a tool or strategy that can be used to generate revenue, much like any other business strategy. For the sake of simplicity, this post will treat use of open source in any meaningful way as a strategy that is an open source business model.

Open source business models are best considered as a spectrum of strategies using open source, from a 100% open source software model, to a pure proprietary model (though some might argue that "proprietary" is not the opposite of "open source" ... we will treat it here as the other end of the spectrum because it indicates that revenue is generated with no reliance on open source software). Along the spectrum are various combinations of development, licensing, services and other strategies with different triggers for generating revenue. Some of the most well know, starting from pure open source and moving to pure proprietary, are:

• 100% Pure Open Source - This is not a business model in the sense that entities in this category are typically non-profit entities that collect donations for the purpose of furthering the goals of their chosen open source project rather than profit-seeking. Entities like the Apache Foundation and Free Software Foundation fall into this category.
  

• Open Source as a Lure - Companies like Google and Sun Microsystems (my employer) employ this strategy. Google hosts its own open source code repository and open source applications (such as Android) optimized to its search results framework to generate higher web traffic and resulting ad revenue from developers and application users. Sun optimizes its hardware-dependent open source applications (such as Open Storage, xVM Ops Center, and MySQL database software) so they work extremely well on Sun's high-performance servers. In both cases, revenue is not derived from the open source software directly, but rather from the increased sales of related activities and materials.

• Aggregation and Services - These are really two separate models, but aggregation is not a viable business opportunity on its own and the two models are well matched. In consideration of seemingly infinite amount the open source code available on sites like Sourceforge or Launchpad, pulling related code together into a fully functioning application is a real value. Red Hat's ability to compose Linux from multiple sources with many copyright holders into an enterprise-friendly, reliable operating system is a perfect example. Providing services (including training, support, maintenance and professional consulting) is a very compelling revenue opportunity. Services are a natural addition to aggregation because of the expertise and knowledge gained in the process of aggregating code and ensuring it works properly. Red Hat also enhances its offering by providing a full spectrum of services to customers.

• Embedded Use - Open source software licensed as embedded components is another viable business model, but the revenue generating potential lies in the use of a dual-license model rather than the value of the open source software itself. An embedded component licensed under a permissive open source license (e.g., BSD, MIT, Apache, etc.) will not generate revenue because users have no incentive to pay for the freely available software with broad license rights. By contrast, an embedded component licensed under a restrictive or viral license (e.g., GPL) forces a potential OEM or system integrator to purchase a commercial license to the same software. Of course, this requires full ownership (or at least broad license rights) of all rights in the software. The MySQL database is an example of this model.

• Tiered Product - Tiered open source software offerings include two basic types of models. First, an open source offering could be an entry level version of a product, much like the evaluation or "light" version of a proprietary product, with a more fully-featured version available for a fee. However, the open source vendor has an advantage over the typical proprietary evaluation product because the source code is freely available for modification and customization. Funambol's open source and carrier grade editions illustrate this type of tiered product well. Second, an open source offering could be fully featured with no separate enterprise version of the product, and the open source vendor can compel purchase of a commercial license by offering additional tools or add-ons that make the open source software more valuable either by making it easier to use or enhancing its efficiency. In both cases, we come closer to a pure proprietary model in that revenue is generated by the value of the software itself either alone or in conjunction with other software or services.

• Incidental Open Source - Many of the most useful and reliable software tools, components and subroutines are available under open source licenses. As a result, virtually all software developed and distributed today contains open source software to save time and resources. However, the mere inclusion of open source components in a software product does not mean that a software vendor is engaged in the open source business. As a result, the incidental use of open source should not be seen as a means of generating revenue through the use of open source.

• Pure Proprietary - The name says it all. Distribution of software containing no open source components, and with no connection to other open source software is in no way an open source business model. This model was common 10 - 15 years ago with most software vendors such as Microsoft, Oracle and others, though all of these vendors not use open source materials at least on an incidental basis and are actively participating in other open source business models.

This is not an exhaustive list by any means. Each business entity must consider its goals, strengths, and the multitude of variables of each development, license, go-to-market element and revenue trigger for each open source strategy to ensure it finds the best point along the spectrum for it's unique needs. Moreover, these business models should be combined to achieve maximum effectiveness.