When video camcorders reached consumer pricing in the 1970s everyone thought they could make videos. In the 1980s, the rage was that everyone could be a desktop publisher. In these 1990s I thought the rage would be that everyone could design a Web site, but I wasn't concentrating properly. No, the rage in electronics is that everyone thinks they can design their own ICs.

This last year has been interesting from a capital investment point of view. Foundries that were going to expand didn't; ones that were built lay empty of manufacturing equipment; plans were delayed by mostly everyone for any real growth in new foundries. And there was still massive over-capacity. And, in selling that capacity, foundries persuaded quite a few companies to go "fabless." (If the Beatles were the "Fab Four" what would the "Fabless Four" have been, one wonders.) Even well-established companies -- though some of them of the one-horse variety of product manufacturers -- closed down their foundries or mothballed them in favor of handing the manufacturing over to someone else.

Many of those companies, already suffering (?) from management changes, felt they had no control over plants in foreign (i.e. cheaper labor) lands. Fundamentally, however, the companies now have even less control over the production of their products: This strikes me as being a poor manufacturing decision. When business turns the corner (and for most it already has) the over-capacity is going to disappear at a fast rate, and even if your contracting foundry doesn't dump you it will exercise every price hike it can. So, you have even less control over costs, delivery and quality. When a small foundry shut its doors, overnight, in San Jose late last year it left one vendor in the lurch. In the next couple of years I predict we will see even more companies lose their production-capability to some other company.

The "Fabless Model" is a unique description mostly parlayed to VCs as a means of getting into the IC business at a reasonable cost. It is assumed by the investment community that there can never be another Analog Devices, or Burr-Brown, or Maxim, or Linear Technology because the costs would simply be too high. The analog semiconductor cell that was born and divided and sub-divided is supposedly now a mature life-form that can never grow out again.

To my mind that is a sad, closed-thought process. It might be difficult but with the right imagination and products it is certainly more than possible to build a new manufacturing analog semiconductor company today. Certainly in terms of taking real control over the whole operation it is the only way to go.

Nevertheless, the "Fabless Model" is in full vogue and if these companies' first products don't work correctly at first pass through their contracted foundries they will, for the most part, be in really bad financial shape. Did I hear partnerships? Yes, of course, get the fab to invest in you and they will be your "strategic partner" through the early days. And some of the larger companies do invest in 5-10% of fabless start-ups; it gives them an oversight of the directions being taken, brings some foundry business in and -- ultimately -- gives them almost total power to sway the future of the new company. Buy it out or kill it.

Whenever I am invited in to a fabless company I see genuine wishes to build its own pet products, and I see good people who I hope will get there. But I have also seen freaky things with people working in terrible buildings with poor equipment, no pay and, probably little or no future with the project. I have seen people throw their houses, their inheritances and their domestic lives into the abyss . . . like gamblers out of control.

What has really got to me is seeing even more companies getting into the "Fabless Model" who are already in business, mostly with a one-product line, who think they can integrate their product. They will hire an IC designer (probably a moonlighter, and how many good IC designers do you know who need to moonlight?) who will put it together for a fee -- and some never-to-happen royalties -- and then go back to his day job. Whether the product ever gets through a foundry is one question; then will it work to specification?

Then, how will you sell the products (There is absolutely no different in selling ICs compared to finished products, as we all know.) And, do you recognize the competition? In the last instance I have noted a company is going into the fabless design of Class-D amplifiers. That company will be competing with the likes of TI, Harris and Siemens who, themselves, have spent many years just trying to get past the problems involved.

Did you hear "splat?"

By: Paul McGoldrick
Sr. Technology Editor, EDTN

Analog Main | Product of the Week | Columns | Editorial | Tech Notes

		Click here to get your listing up.