Now I'm watching pictures of politically active Taiwanese, presumably members of the venerable Kuomintang party, rioting outside their president's headquarters. In the air are civil unrest, financial uncertainty, angry threats of military action by the People's Republic of China and the possibility that President Lee may find an excuse in these problems for forgetting to leave office at the end of his term.

None of this would be very germane to a column on semiconductors, except that a good portion of the world's CMOS capacity-or, a good portion of the capacity that is available to fabless companies-happens to be on this one small island. This fact was mentioned, in passing, when Taiwan was rattled by a severe earthquake a few months ago. But the fabs were mostly undamaged, and the temporary constriction in capacity was absorbed quietly by the industry. At least, no one wanted to admit that it had been a problem, lest they reveal a competitive weakness to their customers.

So we collectively decided not to discuss the implications. But what would happen if the infrastructure that supports the Taiwanese fabs were interrupted by major civil unrest? What would happen if the fabs themselves were damaged by armed conflict, or if they suddenly were inside a country with which we do not trade advanced technology?

Would the captive fabs, ASIC vendors and the few foundries in the rest of the world be able to absorb the volume? How long would it take to do the process migrations? What would the loss of ability to export, say, steppers to Taiwan do to the equipment industry? Do we really want to run the experiment?

The other interesting question is why we keep having to ask these questions. How many more times will we gamble on such artificial-and hence anticompetitive-concentrations of capacity in one geographic area? And what mechanism would we have, either through the free market or through global trade regulation, to avoid it? Maybe we should discuss that for a moment.