"In a sense, hardware is becoming much more like software, up to the point where you actually fabricate an object. That's why you're starting to see open source techniques in hardware. Design is largely going to shift out from manufacturers to the communities."
The study of open source software (OSS) development and business strategies has become the subject of significant research interest. However, there is little research on business models associated with the development of open source assets other than software such as hardware and content. This article provides an overview of current business models for creating revenue for both open source software and hardware. It then summarizes our research of market offers to identify the ways companies use open source hardware (OSH) projects to make money. The research results provide insights about the dimensions of open source business models in an area other than typical software development.
There is no consensus on the definition of OSH. With OpenSparc, Sun Microsystems defines OSH from a development perspective as "a collaborative process around developing new chips." Jeffrey Rowe defines OSH as "hardware for which all the design information is made available to the general public". OpenCollector points out that in addition to design information, information on how to use the hardware, including documentation about its interfaces and the tools used to create the design, must be freely available. Jamil Khatib of Opencores.org suggests that: "All design files should be available for free. This includes schematic, Hardware Description Language (HDL) code, and layout files. Software and firmware interfaces such as drivers, compilers, instruction set, and registers interfaces should be available and open source. All information and documentation, like application notes and interfacing information, should be also openly available".
Gregory Pomerantz provides another OSH definition with more emphasis on license characteristics. He points out that OSH licenses must: i) grant permission to freely distribute the source code and any hardware device design based on it; and ii) grant permission to create derivative works based on the source code and distribute them under the same license. Phillip Torrone explains that OSH must use a license that permits users to study, change, and improve the design, and to redistribute it in modified or unmodified form. Torrone organizes electronic hardware into six layers and explains that the license terms and the information that is shared in each of the layers is different. The six layers are:
- Hardware diagrams including the physical dimensions for enclosures and mechanical subsystems. Designs in this layer are normally described by vector graphic files.
- Schematics and circuit diagrams including images in PDF, BMP, GIF or other formats.
- Parts list of needed components.
- Layout diagrams including electronic circuit designs, copper prints of circuit boards and drill files.
- Core and firmware including the source code that runs on an Integrated Circuit (IC). It could also refer to the design of the IC itself described in Hardware Description Language (HDL) files.
- Software or Application Programming Interface (API) including the source code that is used in a computer to communicate with the OSH.
Most business model definitions are related to the way companies create value for customers and how a company takes business opportunities to generate profits. In Do Some Business Models Perform Better than Others? Malone et al. developed a general business model classification based on each company's revenue stream by means of two variables. The first variable is the type of user rights sold to customers. The four defined types include the:
- creator who sells the right of ownership by significantly transforming the asset
- distributor who sells the right of ownership without significantly transforming the asset
- landlord who sells the right to use the asset, such as companies that license the use of software
- broker who sells the right to be matched with potential buyers or sellers of another asset
The second variable is the type of asset used to create the market offer and includes four generic asset types:
- physical: durable and non-durable items, such as food and hardware
- financial: monetary assets such as cash, stocks, bonds and insurance policies
- intangible: non-physical assets such as protected intellectual property (IP), knowledge, goodwill, brand image, and software
- human: people's time, effort, knowledge and skills
The combination of the two variables results in 16 detailed business model archetypes that could be used as a tool for business model analysis.
OSS Business Models
OSS profitability and business models are still poorly understood and there is no single framework that would explain their potential determinants. The most critical issue for an OSS business is that the licensing terms used allow the free redistribution of the licensed software. Therefore, it is usually not feasible to base revenue on licensing fees.
Rajala et al. identify the following essential elements in any business model for software companies:
- product development
- source of revenue and pricing
- sales channel options
- servicing and implementation approach
Jussi Nissila identifies the key elements of any software business model in slightly different terms:
- value creation and revenue logic
- market offerings and positioning
- product development, implementation and servicing
He also argues that, in the case of open source business models, the above key elements must be complemented by the: i) extent of community development and review; ii) style of development method as more open or more closed; iii) license type as more restricted or more liberal; and iv) importance of the OSS in the end product. Importance ranges from pure OSS where no proprietary components are added, OSS driven where the core is open source with proprietary component added, to proprietary software driven where the core is proprietary with some open source components added.
Dual-licensing seems to be one of the most popular ways of making money. In addition to the revenues that comes from selling the fee license, this model may also provide complementary revuenues through technical support and services. Dual-licensing differs from a purely free model. First, the OSS community does not have the development power to start competing products because the control of the core is held by the original developer. Second, users have the possibility of buying a proprietary license. There are two fundamental legal requirements for a commercially successful dual-licensing model: the i) need of a license with a strong copyleft clause; and ii) possession of undisputed rights over the software.
OSH Business Models
There is little research on the types of business models specifically related to OSH. With OSH, the open asset is different from the ultimate market offer, the manufactured hardware device itself. OSH is not about hardware, it is about the intangible assets related to hardware design information. This makes existing OSS business model classifications not directly applicable to companies involved in OSH projects.
Salem & Khatib suggest the following models:
- Design distribution: enabling companies to sell distributions containing a set of designs.
- Design technical support: enabling companies to offer technical support related to open designs.
- Design implementation: enabling companies to implement and sell the designs by paying royalties to the original developers.
- Releasing: enabling companies to release their designs under a GPL-compatible license, whenever the implementation is considered commercially available.
Additionally, companies working under contract for a chip manufacturer could open source a design to enable the participation of other chip manufacturers and develop multi-purpose and customizable designs. Companies could also open source IP that is not intended to be sold, but rather to be integrated into a product or service. This IP will have the benefits of an open source development process without risking revenues.
Clive Thompson identifies two main economic models for OSH-based market offers. The first is to concentrate not on selling hardware but instead on selling expertise as the inventor. The second is to sell OSH devices while trying to keep ahead of the competition. The heart of this second economic model is higher competitiveness based on complementary assets.
In OSS, "free" may be confused with "gratis" because it often costs nothing to make your own copy. In OSH the situation is different. People can download free hardware designs, but they either have to pay someone to manufacture the hardware or buy the components and tools and manufacture the hardware themselves. In most cases, it is very costly to manufacture the hardware. The costs are related to the replication of the physical hardware, not with the replication of the design itself. However, the benefits of an OSH collaborative process are not reduced by the slowness or difficulty of the replication process of the physical hardware. Companies may still choose to capitalize on the commercial benefits of the OSH collaborative process by participating in collaborative development of hardware that is required by all but is a differentiator for none.
The costs related to designing, verifying and understanding OSH are also high. This requires appropriate EDA tools which are very expensive. In addition, hardware testing and verification requires expensive external hardware equipment such as oscilloscopes, analyzers and wafer probes. Currently, there are open communities developing open source EDA tools that will eventually improve to the point where they will be competitive with commercial EDA tools. A major obstacle is the fact that some commercial EDA tools are designed to work with commercial FPGAs that are protected by commercial secrecy. Open source EDA tools could not be adapted to interface with those FPGAs without facing legal problems. One of the suggested solutions is the development of open source FPGAs whose interfaces would be open enough to allow the use of any open source EDA tool.
In some pieces of hardware, the cost of the IP, which includes the cost of the design, is much lower than the cost associated with manufacturing and integration. For example, in the case of microprocessors, designs built on OSH IP cores alone are not likely to be commercially successful. This means the cost of some commercial IP cores must be added to the final cost of the hardware product. Bugs in hardware designs can unexpectedly increase manufacturing costs by causing physical damages to the chip or other parts of the system. This challenge is a major difference between OSH and OSS. It is additionally aggravated by the increasing technological complexity of state-of-the-art microchip designs and manufacturing requirements.
It is difficult for OSH developers to design products without infringing existing patents. Spooner has pointed out that established companies, such as IBM and Intel, might not be directly challenged by the OSH movement because of the patents they use to keep their products safe. In many cases, existing hardware design patents will seriously complicate the deployment of OSH designs.
None of the existing OSS licenses are designed to cover all aspects of hardware design. At the same time, hardware has become more like software. HDL code is considered by the Free Software Foundation (FSF) to be software which can be legally distributed with OSS licenses. However, according to Richard Stallman, although IC definitions written in HDL can be copyleft, ICs themselves cannot be copyleft, because they cannot be copyright. A copyright can protect a design from being copied and modified, but it cannot regulate the manufacturing, distribution and use of products based on the design. Patents are used in such cases, but patents are expensive and not as flexible as copyright. The final result is that, for a design that is copyleft by a GPL-like license, "any person can legally draw the same circuit topology in a different-looking way, or write a different HDL definition which produces the same circuit", thus making the benefits of copyleft in hardware somewhat limited.
Since 2007, the Tucson Amateur Packet Radio (TAPR) organization has been promoting the TAPR Open Hardware License (OHL) as a legal framework for OSH projects. TAPR argues that the license can be used for any kind of physical tangible product. As a share-alike license, modified designs can be redistributed only by using a license with the same rights as those granted by the license of the original design. They point out that "OHL is not primarily a copyright license" and, although it does not prohibit a company from patenting an invention related to an open hardware design and then enforcing its patent rights, it states that parties who receive any benefits from an open hardware design "may not bring lawsuits claiming that design infringes their patents or other intellectual property".
Another challenge arises as hardware is not as modular and compartmentalized as software. Modularity is a critically favourable characteristic for OSS production. For example, modularity was important in the case of the Apache software allowing developers to work in particular areas without affecting other modules. Netscape faced significant difficulties when releasing the Mozilla browser as OSS due to the insufficient level of modularity in its software architecture and a major restructuring of the program was needed to address that problem.
Another obstacle facing OSH is what OSS faced more than a decade ago: lack of credibility. It is expected that the OSH community will eventually convince users that the OSH model works and high quality OSH designs can be achieved.
Identification of Different Ways of Monetization
Our examination of 4 companies, 88 market offers and 93 OSH projects resulted in the identification of seven different ways of making money with OSH:
- Services over owned or third party OSH: this category includes companies that offer customization and consulting services over hardware designs developed by the OSH projects they are participating in.
- Manufacturing of owned or third party OSH without additional proprietary hardware components: this category includes companies manufacturing and selling pieces of hardware based entirely on the OSH designs developed by the OSH projects they are participating in.
- Manufacturing of proprietary hardware based on OSH: this category contains companies that sell the right of ownership of pieces of hardware based on OSH assets developed in projects they participate in. The designs of those pieces of hardware are not entirely open source, but contain some proprietary modules and components. The final design of the market offer is a modified and improved version of the original OSH design.
- Dual-licensing: this category is similar to the dual-license model found in OSS. It includes companies owning OSH designs that are offered for use under either a GPL license or a proprietary license. The design does not contain any proprietary module.
- Proprietary hardware designs based on OSH: this category is similar to the previous one, but the hardware design that is offered for use under a proprietary license contains proprietary components or modules.
- Hardware tools for OSH: this category includes companies selling the right of ownership of development boards for the testing and verification of hardware devices manufactured on the basis of the OSH assets developed by the OSH projects the companies participate in. The designs of these boards are entirely proprietary.
- Software tools for OSH: this way of making money with OSH includes companies selling the license for software tools for the development of OSH assets. The software tools are entirely proprietary.
To examine the distinguishing characteristics of the seven ways of making money we used three dimensions:
- Type of user rights sold to customers. Two types of user rights were found to be relevant to our research: i) creator, which is based on selling the right of ownership of the market offer by significantly transforming the OSH asset; and ii) landlord, which is based on selling the right to use the market offer asset.
- Type of market offer: this includes three relevant types of assets: i) physical, including the hardware device manufactured on the basis of an OSH asset or the software and hardware tools used for the development, test and verification of the OSH assets or hardware devices; ii) intangible, including the OSH asset itself or other intangible assets related to the market offer; and iii) human, including people's skills, expertise, time and effort to offer services such as custom designs, consulting or training.
- Degree of integration of the OSH asset into the market offer: includes four options: i) pure open source, when the market offer is the same as the OSH asset and there are no proprietary components added; ii) open source driven, when the core of the market offer is open source but there are some proprietary component added; iii) proprietary driven, when the core of the market offer is proprietary but there are some open source components added; and iv) proprietary, when the market offer is not based on the OSH asset and includes only proprietary components.
The results of our research can be summarized as follows:
- Companies turn OSH challenges into business opportunities. We identified two types of market offers that were unique to OSH: i) manufacturing OSH devices; and ii) developing proprietary tools for test and verification of OSH designs and OSH manufactured devices. The development of proprietary OSH test and verification tools represents a clear opportunity for companies exploring new ways of making money based on OSH. Both types of market offers address two of the main challenges associated with OSH. First, most of the product costs related to OSH are associated with manufacturing. Second, the tools for the test and verification of OSH designs and manufactured OSH devices are very expensive. Therefore, by exploiting these opportunities, companies turn OSH challenges into business opportunities.
- OSH stack development is not exploited as a way of making money. While some companies make money by selling stacks of software comprised of OSS only, a similar way of making money with OSH was not found. This dissimilarity could be related to the lower degree of modularity of hardware designs. Hardware stacks are more difficult to design because of the higher interdependency of hardware modules.
- Subscription services are not exploited as a way of making money. While OSS companies can make money by selling subscription services, many OSH projects do not require further improvements in their designs because they already fulfill the function needed by the end users. The OSH designs found in our research sample do not seem to require a continuous development process and were based on a single release of the source code. This is especially true for OSH projects related to small IP cores. In contrast, most OSS projects require a continuous development process and frequent releases. It is difficult for a company working on OSH to make money by selling subscription services because most of the OSH projects are developed "on-time" or have few or very infrequent releases.
- OSH dual-licensing follows the same logic as in OSS as it also requires the use of highly-restrictive licenses and undisputed rights.
- OSH can be used to create standards and commoditize a particular technology layer. It is worth mentioning that our research found one company owning an OSH asset without any associated market offer: Sun Microsystems and its OpenSPARC project. Sun Microsystems publicly articulates its strong interest in creating a healthy open source community around OpenSPARC to: i) commoditize a particular technology layer and move profitability to adjacent layers where market offers are optimized and have greater value; and ii) create standards or eliminate competitors who could potentially capture value from that layer.
- The classification of the ways of making money depends on the integration of the OSH asset into the market offer. It was found that in order to deal with the distinction between the OSH asset and the market offer, we needed to introduce the degree of integration of the OSH asset into the market offer. The analysis of OSH-related business models should take into account the specific ways of integration of the open source assets into the final offering.
The objective of our research was to identify the ways companies use OSH projects to make money. The manufacturing of hardware devices based on OSH designs was found to be the most popular commercialization method. Services over existing OSH assets, dual-licensing and the development of OSH test and verifications tools were identified as additional viable options. The study shows that the classification of the different ways of making money depends on the degree of the integration of the OSH asset into the market offer and can not be based only on the type of market offer and the type of user rights sold to customers.