International business finance

The future is not what it used to be.

The good thing about conferences is that they demonstrate the infinite variations of reality. One person's view of reality can be very different from another's, and conferences are where differing realities rub up against each other and are tested against each other. But even more various than views of reality are expectations about the future.

The recent International Electronics 2002 conference in Morocco was no exception. One the one hand you got Siroyan CTO Adrian Wise telling us that process technology is becoming standardized, that "in house processes are dying out," and that the microprocessor sector is turning into an IP business.

On the other hand you get Sony CTO Tsugio Makimoto asserting that an in-house advantage in process technology is vital to having a successful games microprocessor, that an edge in proprietary microprocessors is vital to success in the games machine business and that games microprocessors are becoming so technically advanced that they could replace DRAMs as the industry's technology driver. Is Makimoto wise, or is Wise?

Another luminary gainsaying the "IP is the future of the industry" view was Aart de Geus, CEO of Synopsys. "I don't think there will be many IP suppliers because it's very difficult to do reusable IP," said de Geus, "so I believe the center of gravity will be with the semiconductor producers. They will attract a few partnerships with IP companies, which they will legislate very closely. There are a few successful IP companies -- ARM, MIPS and, in some ways, Rambus--plus some DSP companies, but all the rest are small stuff."

Still other opinions diverged. Whereas STMicroelectronics CEO Pasquale Pistorio reckoned, "Manufacturing competence is vital for a company such as STMicro;" Hitachi Semiconductors CEO Satoru Ito said that Japanese chipcompanies "have to considermodifying the IDM (integrated device manufacturer) model by going outside for foundry capacity." And, NEC Semiconductors Executive General Manager Hiro Hashimoto asked, "Is process technology a differentiating factor for SOC?" replying indecisively: "Yes, but how much so?" He argued for desiliconized SOC design and talked about collaboration to standardize the silicon process.

With design-rich Europeans talking about the importance of manufacturing process and with the Japanese masters of manufacturing technology talking about defocusing on manufacturing, the future looks to be quite different. "The industry is not going back to pre-2001," said Ulrich Schumacher, CEO of Infineon, predicting even greater turbulence, uncertainty and risk.

Customers are the reason. The main customers for semiconductors will no longer be the PC people; it will be the communications and consumer companies, argued Sony's Makimoto. And consumers are notoriously unpredictable, adding to market volatility while, in predicting the communications market, the equipment companies "don't know any better than you do," Schumacher said.

The result, as both Schumacher and Hitachi's Ito agreed, is that semiconductor companies have to talk to their customers' customers to understand what the end-market wants and to show what semiconductor technology can bring to the party. "Telecommunications operators are getting closer to semiconductor manufacturers to understand better what the possibilities for innovation are," Schumacher said. "We need a network where everyone talks to everyone--not just to their immediate customer."

So the simple semiconductor industry formula of "make it faster, less power-hungry and more powerful, and it will sell" is not working any more. When the computer industry dominated the semiconductor market, its requirement was always for bigger, better, faster. But now that the consumer and communications industries are dominating the market, the requirement is not just for pure technology improvement--the requirement is for a value proposition.

Value propositions for introducing new technology are difficult to do because you can't do market research on something with which people are unfamiliar. Ask the average Joe if he'd like a Mbit/sec. down his phone line and he probably won't have a considered view. Ask a telecom operator if they think their customers should have a Mbit/sec. down their phone lines and they'll shudder like Scrooge. They'll reply, "What will people use it for?" Intel co-founder Gordon Moore recalls asking the same question when someone proposed to him in the 1970s that Intel should build a PC.

No one can know what people will use new technology for, but if everyone takes the view that the application has to be self-evident before new technology is implemented, then progress will be slow. Reality does not evolve linearly or predictably but empirically, and sometimes accidentally.

David Manners is an editor for Electronics Weekly, a sister publication of Electronic News.