Often, when I talk to people about my line of work – once I have explained exactly what embedded systems are – they will comment that it must be a good line of business to be in. I would tend to agree. However, I would actually be agreeing that it is an interesting business, whereas my interlocutor is more likely to be thinking of the financial benefits.
Of course, as embedded devices are so ubiquitous, there overall size of that business is enormous – I could not really put a figure on it. However, I am not in the business of making embedded systems; our role in the business is providing tools to embedded software developers. It would seem logical that this enabling technology must also be quite a substantial market in its own right …
The size of the business supporting embedded system development is hard to determine for several reasons. First, the definition of “embedded” is far from precise, so interpretations vary. Second, there are almost no substantial exclusive vendors of embedded development support products that are public companies. The firms are either private companies that do not reveal their financial details or public companies, like Mentor Graphics, that also operate in other areas. However, perception is everything and there are plenty of companies that want to join the party. Two interesting participants are Microsoft and Intel.
Over the years, whilst dominating the desktop software market, Microsoft has made several attempts to enter the embedded software business. Most of their earlier shots at it tended to flounder, because their approach would constrict the designers’ freedom to innovate. Somehow a Windows powered photocopier never seemed to make sense. I would say that, nowadays, they seem successful in their participation, which has been achieved by looking at the market in a novel way. Instead of trying to provide a universal solution, they are targeting different types of embedded systems in different ways. There are the Windows Embedded products, which enable systems to be built around PCs. Windows CE facilitates building smaller systems, with graphical UIs and good connectivity. For cell phones, Windows Phone, in its latest incarnation, seems to be getting a good reception. And, of course, they have a direct involvement in supplying a particular class of embedded system with Xbox and Kinect. I am quite sure that at least one Microsoft employee will put me right on my perceptions.
Intel are different. They almost single-handedly started the embedded systems business, as the 4004 was the first real microprocessor. In due course, the 8080 became very popular. The 8051, despite its rather ugly architecture found a large following and remains popular today. The 186 family more or less defined the 16-bit microcontroller market. Somehow Intel seemed to lose their way with embedded – I assume that the PC market distracted them. The Z80 grabbed lots of 8-bit device users; Motorola [now Freescale] snatched the 32-bit market. Intel have made a number of attempts to get traction in embedded. I recall their 3-pronged attack: x86 for PCs; the 860 family for graphical workstations [I never understood how these were distinct from PCs]; the 960 family for embedded. Both the 860 and 960 are history. There have been a few attempts to make viable embedded x86 parts, but none of those have really set the world on fire. Lately, Intel seem to be trying a couple of strategies. They recognize the need for an ecosystem and acquired Wind River. And they seem to have focused a lot of chip design effort on low power in order to combat ARM. The announcement of an Intel-power cell phone at CES is an interesting move.
As the Chinese saying goes, we live in interesting times. Happy New Year.