by Lindsey Vereen
What do cotter pins and cross-compilers have in common? More than you might think. But before I elaborate, consider the long love-hate relationship between software and hardware companies. Software is expensive to develop and maintain and it doesn’t generate the revenues that silicon or systems do. It has been axiomatic that you use software to sell hardware. Systems companies have long been known to give software away to entice customers to buy hardware.
In the semiconductor industry, tools have been a means to sell silicon, but semiconductor vendors have never embraced tools enthusiastically as a part of their business. Often as not, they divest themselves of the tools business as soon as they can, figuring rightly that while they can turn silicon into gold, tools are as lucrative as lead. On the hardware design side, companies like Cadence Design and Avant! have thrived selling tools to chip companies, tools that 15 years ago many chip companies developed internally.
Software development tools have had a similar history. Chip companies introducing a new architecture have wanted the tools in place and have developed the tools themselves. But ultimately, they’d like it if third-party vendors took care of the tools. At the same time, tools are getting even more important, and they can play a competitive role in the semiconductor business.
Which brings us back to those cotter pins. Let’s say you wanted to start a new automobile company. You design your neo-Delorean, contract out to a lot of suppliers for parts, rent an assembly plant, and you’re in business. But the existing automobile companies don’t relish your intrusion into their market. So they corner the market for, say, cotter pins. You go ahead and build your automobiles, but with no cotter pins, you can’t keep the wheels from falling off. Ergo, you can’t move your product, in any sense of the word, and you’re out of business.
Think of tool companies as the cotter pins of the semiconductor industry. Without tools, chip companies can’t sell processors. So it’s probably no surprise that chip companies have gone on a binge lately of buying critical software companies. TI bought Tartan and Go DSP a couple of years ago. More recently they acquired Telogy. Not to be outdone, Motorola has announced it’s buying Metrowerks. If this acquisition is successful, do you think that Metrowerks will continue to support other architectures besides Motorola?
TI paid $435 million for Telogy, and Motorola is offering $95 million for Metrowerks. These are attractive numbers, to say the least. But even if semiconductor companies buy up the good tool companies and lock them in to their own architectures, that won’t be the end of the story. If someone buys up all the cotter pins, someone else is liable to open a cotter pin shop. The tool business has been nothing if not dynamic. It’s never long before entrepreneurs sniff out potential markets. And what greater incentive could there be than knowing that if they’re successful, some big chip company might snatch them up?