Introduction

Some time back, LinuxDevices.com invited me to write an article about how our company, Mind NV, has successfully become profitable while remaining a pure free and open source software based embedded development company (we primarily focus on Embedded Linux, eCos, and JVM development for R&D groups at large corporations). The financial results from 2Q-2002 are now in, and we do have a decent profit (EBIT 12%) and no more debts to the bank. I am proud of having achieved this result in the current times, after overcoming serious challenges in previous quarters.

For the further mass adoption of free and open source software in the embedded space, we need to find long term symbiosis of free software development and business profitability — not only in the support and customization business (where Mind is earning its revenue now), but also for funding development from scratch of frameworks and basic infrastructure. I suggest a model where the licensing scheme evolves along the life-cycle of the software, from enforced sharing of the initial start-up costs among the early commercial adopters to a truly free software model when the software reaches the early majority, mass adoption phase.

The past: the first 2.5 years

Mind started October 1999 with 2 founders, specialized in consulting and solutions for IT infrastructure and security. In February 2000, I joined with 3 more new shareholders and we executed a modest capital increase from our personal savings to a capital of 124,000 EURO (orders of magnitude less than some other Linux companies that got funded or went IPO around that time). During 2000 we continued to boom (on a relative scale), in parallel with most other start-up companies, including our customers. VC funded companies (in pharmaceutics, telematics, ASP's, ISP's, etc.) needed upgrading services from Mac, Windows, and UNIX boxes to Linux and required improved network security (in one case as part of the due diligence for an upcoming IPO).

In parallel, at a local tradeshow in Brussels (TMAB 2000, April 2000), we were contacted regarding the use of Linux in overhead projectors, financial transaction devices, VoIP boxes and the like. This sounded extremely interesting to our developers and shareholders, and it marked the start of our Embedded Development business unit. In this Embedded business unit, we could use the experience of our kernel programmers (on PowerPC, ARM, MIPS, IA32, etc.) to write device drivers, port Linux to new custom or standard platforms, and give in-depth architecture advice. As a solution to hard real-time requirements, we successfully added eCos alongside Embedded Linux, mostly using a dual processor approach (real-time tasks and local connectivity with eCos; LAN connectivity, GUI, and control tasks with Linux and Java); and we also use RTAI when it is advantageous to run the real-time and non real-time tasks on the same processor. However, with the advent of systems with multiple processor cores per system, even within a single FPGA, it has become increasingly attractive to spread tasks over different dedicated processors: some on Linux, and some on eCos.

Beyond technological differences, a major difference between our new Embedded business and the IT and Security business from which we had started was customer profile. Suddenly we found ourselves working for customers like Siemens-ATEA, Barco, Philips, Banksys, Newtec, Alcatel, etc., respectable companies on a local or international scale, that could hardly be called dot-com's.

Our combined business peaked towards the end 1Q-2001, but from April 2001 on, the business for the IT and Security business dropped drastically. The pharmaceutical company postponed its IPO plans, the telematics company started to cut budgets to live longer on the available VC funds, the ISP got into financial problems and cut down costs (by now all of them have found solutions through new funding or M&A). Around that time, we also faced a first natural management crisis and the number of managers was reduced in the small company we were.

I attribute the problem of not finding enough new customers in our IT and Security business unit mainly to the strong correlation between the adoption of Linux for IT and the type of customer company. Our customers were exactly the strong growth, VC driven companies that were open to new technologies, had no large installed base of UNIX or Windows, adopted Linux strategically and furiously and were prepared to work with our small start-up company to deliver the expertise. It were also these companies that were hit hardest by the implosion of the dot-com bubble. Another cause was probably the defocus that was created by operating a start-up in two completely different markets (as is taught clearly in all marketing literature on start-ups).

Since 2Q-2001, the revenue from the IT and Security business unit declined, while the revenue from the Embedded development group continued to grow. By 4Q-2001, the costs from our IT and Security business unit's operations far exceeded the revenue generated by that business unit, and more than absorbed the profits from the Embedded business unit. When this became clear in January 2002, because of our small capitalization, we needed to act quickly and a number of layoffs were unavoidable. This stopped the financial bleeding and after the one-time costs associated with the layoffs, we were at least break-even and the short-term existence of the company was secured. Looking back, we should have acted earlier; but hindsight is always 20/20.

The present: back to profitability

By April 2002, it became clear that we would not see a fast recovery of the Linux IT and Security business in the market segment we were aiming at, and that we would be better off focusing all our attention on the embedded development market. We struck a deal with the remaining employees in the IT and Security business unit to set-up a Management Buy Out by the manager of that unit. We now have a partnership with Digibel, a company some of them where involved in for many years before Mind existed. Using some different approaches, Digibel is successful at addressing the small and medium sized business with Linux and security. Nevertheless, while exhibiting at CeBIT 2002, we also felt increased interest from larger companies (the currently very successful gsm operators in Belgium, International concerns, etc.) for using Linux in their IT infrastructure, mainly for reasons of cost savings and security. In the coming years, I expect to see Linux move increasingly into these large accounts.

For Mind, the focus is now fully on Embedded Linux, eCos, and JVM development as infrastructure elements of embedded systems. As suggested in Crossing the Chasm by Geoffry Moore, we try to stick strictly to our niche of open source software embedded development. We try to satisfy our customers with support and customization services; however, I admit that we cannot always optimally deliver the full solution, for not all of the framework and basic infrastructure is yet available as free software (e.g. we miss a full-featured, free embedded development framework, a free JVM with large scale adoption, etc.).

The most differentiating part of our current business is localized development support. Since the software is essentially free, this localized support is the main aspect that will always remain of “billable” value. This is why we believe the following are the key factors to success in our service driven, open source development company: the employment of good developers, creating a team that shares information, continued promotion of open source solutions to corporate customers, and running a decent business in general. This approach results in repeat business and good references, which causes the self-referencing effect in our market that is described by Geoffry Moore. As a basis for the development, we often make use of technologies such as Debian GNU/Linux, eCos (created by Red Hat), eCosCentric, and Wonka (a Free JVM), and we also support a number of customers that use a commercial Embedded Linux distribution.

Other companies are also using these challenging times to focus on (or refocus) their core business, but sometimes in somewhat opposite directions. One relevant example that comes to mind is that Red Hat recently announced it has discontinued development of eCos, which seems a move that puts less emphasis on embedded development, while we at Mind see an opposite, increased interest for embedded development, including eCos.

An interesting side note is the in-depth discussion we have had with corporate customers about GPL licensing. In a number of recent cases, we were able to convince the customer to allow us to immediately contribute the improvements we made to GPL code to the community, although the GPL only requires this to occur when the product begins to be shipped to the market (typically, about one year later). Doing this improves the efficiency of the GPL development model.

The Future: a new licensing paradigm?

As a basis for continued income, we will continue our current strategy of high-quality, localized development and support. We also plan to shift more to fixed price projects, where the budget and deadline of the project is predefined at the time the contract is signed (there, we should benefit from our previous project experience).

However, next to this customization and support business on established software (e.g. Linux and eCos), there also needs to be increased development of new free embedded software, from scratch, and sometimes in competition with existing non-free offerings. This is where I believe a new licensing paradigm is required in certain cases.

In Part II of this two-part article (coming soon), I will describe our ideas about the proposed new licensing paradigm . . .

---

About the author: Peter Vandenabeele is CEO of Mind NV. Mind specializes in open source Software, including Embedded Linux, eCos and JVM development. Mind delivers services to various well known companies including Banksys, Newtec, Barco, Siemens, KPN Orange, Alcatel, Philips, Shell, and Acunia. Previously, Vandenabeele held a management position at a Belgian Internet Security start-up, and worked at 2 start-ups in Silicon Valley and at Alcatel-Bell research. He earned a Ph.D. at IMEC Leuven and a Masters Degree in Electronics at the K.U.Leuven.

---

 
This article was originally published on LinuxDevices.com and has been donated to the open source community by QuinStreet Inc. Please visit LinuxToday.com for up-to-date news and articles about Linux and open source.