It's a question most of the smaller chip manufacturers must face sooner or later: Is migration to processing chips on 300-mm-diameter wafers possible for such companies, or should they get off the escalator of ever-increasing economies of scale? Should they instead milk their 200-mm wafer fabs--run them into the amortized ground on which they stand--and contemplate a transition to a design-only existence?

If you are already a niche foundry riding your 200-mm wafer fabs hard, what do you do? This is particularly relevant in Europe, where companies like Austriamicrosystems AG are providing foundry services and where X-Fab Semiconductor Foundries AG and Tower Semiconductor Ltd. operate as pure-play foundries.

One possibility is that smaller foundries could be brought together to create a partnership or, more likely, a formal joint venture to build or acquire a 300-mm fab. Tower, for example, has land enough for a 300-mm facility next to its two existing fabs, but it lacks the financial wherewithal to build one on its own.

The simplest solution may be for smaller foundries to acquire older 200-mm wafer fabs and retrofit them for 300-mm wafer processing. But that is a cumbersome prospect involving much reengineering--something most prospective fab purchasers would rather avoid.

Ultimately, good old market forces may prevail, with one foundry buying up the others. Reports are circulating that X-Fab, which has grown through a series of low-cost fab acquisitions, is eager to acquire a 300-mm plant. The rub is that there are not too many on the market. It is not yet clear where an offer of a 300-mm wafer fab might come from, unless STMicroelectronics, which has a 300-mm shell in Sicily, is prepared to cut a deal or Qimonda could be persuaded to part with one of its Dresden plants.

The question may be difficult, but it's key. A smaller fab with access to 300-mm wafer processing and 90-nm processes faces far different prospects than one that has dead-ended at 200 mm.