Annual revenue growth at The MathWorks has become something of a routine. The privately held software firm reported $240 million in revenue in 2003, up about 9 percent from the previous year. Management expects the topline growth to continue. If it didn't, it might indicate that something in the world had gone terribly wrong.

Jack Little, CEO, The MathWorks

All the resources and energies of The MathWorks revolve around two core software products, MATLAB and Simulink, research and engineering tools that, in essence, assist in the design and testing of complex systems.

It's a business model that's apparently recession-proof. Jim Tung, The MathWorks fellow and chief strategist, explains that his company is simply more durable than others. It doesn't focus on a specific industry, but serves a multitude: aerospace, financial, biotechnology, education, semiconductors, communications, and defense, to name a few. Some of these industries run counter-cyclical to each other, meaning one sector's downturn could be another's boom. "It's a steady foundation," Tung says.

Even when markets go soft, Tung says, companies tend to invest in The MathWorks' tools because they help engineers develop those absolutely essential next-generation products.

The MathWorks software is for technical computing and model-based design. The flagship tool, MATLAB, is an "interactive environment for algorithm development, data visualization, data analysis, and numerical computation." It enables the user to solve technical computing problems faster than with traditional programming languages, such as C, C++, and Fortran.

MATLAB software can be thought of as a sandbox with a lot of pre-built things in it. A user can build on these existing building blocks or simply work in the sandbox to create things from scratch. Tung describes MATLAB as "a very powerful environment for trying out ideas and verifying them."

The software also allows ideas created in the MATLAB sandbox to be handed off to a test engineer or an analyst. Productivity gains come from speeding up the idea-creation process and the ability to pass ideas on to others. Chip-equipment suppliers, for example, use MATLAB in developing next-generation fabrication equipment.

Model-based design is a second category in which The MathWorks' tools have held a dominant presence. To understand the model-based design concept, Tung sketches the example of designing a car. Engineers used to design their specific systems independently of each other. For instance, the powertrain, the suspension, and the engine were all created separately.

But as cars grew in sophistication, with complex electronics controlling safety and performance, the cubicle approach began failing—the various systems began affecting each other's performance. Therefore the car had to be designed using a more holistic approach, with a master blueprint at the top level mapping out the interplay of the many separate parts.

The MathWorks' software tools fit in this space. They help a designer who is taking a system-level approach to connect and evaluate all the various subsystems that will work together in the final product.

After engineers devise the master blueprint, they build individual components for testing and verifying. "So you start with a high-level model, implement the strategy in hardware or software, then build a test bed that will confirm the functionality," Tung explains. The resulting design can then be passed on to others, such as materials suppliers.

In short, the software helps to accelerate innovation.

The founding of The MathWorks in 1984 resulted from a fortuitous merger of the theoretical and the practical. In the 1970s, Cleve Moler, an applied mathematician, had written MATLAB because he wanted to give researchers the ability to do matrix-based computations without having to program in Fortran.

Jack Little was in California in the early 1980s, working as an engineer in controls and signal processing, which involved matrix-based computation. Moler gave lectures at Stanford University, and news about the mathematician and his MATLAB software got back to Little.

"I was immediately impressed by the power and elegance of the MATLAB language," Little recalls. "I also knew that matrix computation was important across applications such as control-system design, signal processing, and statistical modeling. And I believed that the personal computer would become an important platform for technical computing, not just for office applications as it was used at the time."

Little teamed with Cleve to form The MathWorks and turn MATLAB into a commercial product. The combination of academia and business proved to be fruitful, and the two brought the software to the desktop PC.

Since then, The MathWorks has been parked in a highly desirable market position. As a dominant player in profitable markets spread across industries, it also has a lot of room to grow, says Daya Nadamuni, principal analyst in the design and engineering software sector at Gartner Dataquest.

Nadamuni estimates the total software-tools market at about $7 billion. The MathWorks holds a small percentage of that. "Their potential market size could be four times what their revenues are currently—or approximately $1 billion," she says.

In the semiconductor industry, The MathWorks software complements EDA products. Another role is analyzing the performance of chips. MATLAB core products for technical computing and data analysis connect to things such as Texas Instrument's Code Composer Studio for DSP designers, as well as tools from FPGA companies Xilinx and Altera.

The MathWorks is also big in the education sector. Dataquest's Nadamuni says that MATLAB has seeped into the majority of engineering programs in universities. When students enter the workforce, their familiarity with MATLAB makes the tools a popular preference.

CEO Little cites new interest in The MathWorks' tools from industries doing financial modeling, computational biology, and image processing. "New markets continually open up," he says.

The MathWorks has no plans to go public, and Little has characteristically remained out of the media, preferring to devote more attention to running his business than promoting it.

"The greatest advantage in being a private company is our ability to focus on the long term," Little says. Management can then make decisions based on the likelihood of long-term success, "and not have our strategies disrupted by short-term, quarterly results," he adds.