The view from Japan: Chipmakers giving up on doing it all themselves*
04/01/2001
 |
Semiconductor makers are starting to question the value of their production divisions. They can't overwhelm the competition with their own in-house technology alone, and they face a worsening shortage of engineering talent. This means Japan's old do-it-all-in-house school of process engineering is breaking down, and technology is starting to flow more freely. Production technology will increasingly be transferred among players much like intellectual property is now in the design realm. The LSI makers who best figure out how to adapt to this new order, focusing with technology paranoia** on one particular production strength, will be the ones who come out on top.
Chipmakers can't afford to do all their own process development independently themselves anymore, now that silicon foundries can supply highly advanced technology so cheaply. "I prefer to use foundry technology instead of our own company's," says a designer at one of Japan's big LSI companies, voicing the opinion of more and more designers. The foundries use off-the-shelf solutions supplied by equipment makers like Applied Materials and focus on low-cost production.
There's also a shortage of engineering talent available. No one company has enough people to develop the staggering proliferation of new technologies it now needs, from shrinking geometries to new materials to flexible multitype production processes for systems on a chip.
And the engineers who are working on developing these processes want to get out. "Twenty percent of our process engineers have asked to transfer to the higher value-added system design division," says a manager at one big Japanese device maker, "and another 20% probably also want to."
 More transfer of technology will bring faster development at a lower cost; this will also allow chipmakers to increase value added. |
If companies aren't going to develop all their own process technology in-house anymore, they're going to have to find places to buy it outside. And those who have the technology will have more opportunities to sell it. A device maker can best capitalize on this new fluidity of technology by concentrating its development resources in a particular area of expertise, and buying the rest from other companies. Companies can get an edge by having the world's best technology in one area, and using the industry standard elsewhere. And by enabling their developers to do the world's leading work in one particular area, companies can stem their outflow of talent.
Technologically paranoid companies are already focusing their engineering efforts on being the world's best in one particular area, like Intel's effort to develop the world's fastest transistor, and Toshiba Semiconductor's aim to lead the world in low-cost production technology. Both these companies showed their lead over competitors in these areas was widening at IEEE's International Electron Devices Meeting in December, Intel with its work on reducing the gate length, and Toshiba with its work on cutting the number of mask steps.
There are not yet any separate providers of intellectual property in the process technology field as there are in the design business, though standard process technology is provided by some equipment makers. The flow of such standard process technology out to more users is starting to increase, thanks to alliances of equipment companies. To facilitate this trend, though, it will be vital to carefully verify the compatibility of different companies' process technology.
The distinctive leading-edge proprietary technology that the technologically paranoid develop in particular areas corresponds to star intellectual property in design. It flows out into the rest of the world only a very little at first, to partners and suppliers and the like, but in time it will spread further. A good example is IBM's copper interconnect and silicon-on-insulator technology. Eventually these advances, too, will spread to others and become standard technology, so it's vital that the developer be able to earn enough from selling the technology to fund development of the next round. These leading-edge proprietary technologies usually flow out to the world first through joint development agreements, like that of IBM
Infineon Technologies AG and United Microelectronics Corp. to develop 0.13µm CMOS processes. — Masahide Kimura, editor, Nikkei Microdevices
*This story was translated from the February 2001 issue of Nikkei Microdevices, our partner in Japan.
**Editor's note: The author's use of the word paranoia here and throughout the article refers to the well-known Intel statement that "only the paranoid survive" in the chip industry.