Bigger isn't always better. Smallest isn't always best. Standard isn't for everyone

11/01/2004

$3, $5, even $10B. These are the astronomical numbers being cited for the next generation of semiconductor megafabs. With each new generation scale reduction, the cost climbs exponentially higher, while uncertainty over the source of product demand needed to support this huge upfront investment also grows.

I have to wonder if this one-size-fits-all approach to the future of electronic device manufacturing is really the best, or only, way to go.

As we reach nanoscale semiconductor devices, some predict smaller footprint facilities, but this doesn't necessarily imply lower overall facility costs. The cost of manufacturing equipment for nanoscale devices, particularly metrology tools, will still be staggering. And, though produced in physically smaller manufacturing environments, the contamination control demands within these environments will reach never-before-seen levels.

For decades the nearly sole drivers for the semiconductor industry have been faster, smaller, cheaper—leading to increased production quantities, higher yields, and higher profits (during the good times). It has also lead to fewer players, fewer product choices, stifled innovation and overseas manufacturing. Though "many-of-the-same" may be cheaper to produce, they also place significant boundaries on the creativity of product designers, and the pace and diversity of overall new end-product development and introduction. And, this ultimately shrinks the market.

It seems to me there was a time when we envisioned true systems on silicon—not glorified ASICs or multiple standard components jammed into one or a few MCMs—but complete, functional end-products (or near-end products), designed and "fabricated" as one unit. Today, however, we're still largely stuck on making the individual pieces and leaving it to product designers to subsequently glue them together as best they can to meet some useful and/or saleable end.

This is true despite the fact that relatively inexpensive, software-based design tools have long been capable of creating such integrated designs; that fabrication materials are readily available that can accommodate multiple performance requirements; and that a host of more-than-modest-quantity, niche, improved and novel new products could make use of them if they could be built at a marketable (as opposed to head-to-head competitive) cost/price point. The problem, as I see it, is that there have been few, if any, facilities specifically designed, equipped and organized to efficiently provide this manufacturing approach.

Today, the technology needed to outfit a system-on-wafer fabrication facility to produce such small-to-medium size production runs is by no means leading edge. Rather, what is needed is flexibility, creativity, and the ability to think outside the big fab box. In fact, the same kind of approach could also be brought to bear to help standard device manufacturers develop, test and produce a variety of truly application-tuned products derived from their flagship products.

Innovation is always at the heart of advancement and progress. Given today's big-fab mindset, smaller, versatile production lines and facilities would both represent, as well as encourage, innovation.

John Haystead
Editor in Chief