Even though the amount of code in electronics systems continue to expand, the development challenge is being met by design teams that have not grown, and design cycles that may even be shrinking. If time to market is not increasing, then that means developers have to write more code in the same or less time, debug it, integrate it, test the whole thing, and then ship the product.
Is this scenario realistic? What's the secret to shipping more complex products on time? We do know that over the past several years the ratio of commercial to roll-your-own real-time operating systems has increased from about 35 percent to more than 50 percent. The implication is that companies are beginning to concentrate on their added value and are foregoing the luxury of supporting a proprietary RTOS.
At the same time, there is a business for software IP such as protocol stacks. But how much software do companies actually acquire from outside rather than develop internally?
If software development trends follow hardware trends, then I expect we will be seeing growth in the use of third party software IP. What is unclear is if companies will buy commercial software IP or acquire it freely from the public domain.
The embedded software business has always been something of an enigma, in that no one wants to pay very much for software or the tools to develop it. Managers tend to think of software as free and want to budget development costs on that basis. Tool developers struggle to price their tools at a point at which customers will buy them and at the same time generate enough revenue to fund further development.
As software becomes a more significant part of electronics systems, sooner or later the industry will have to recognize and address that reality.