Magazine

Japan's Chip Industry Is on the Comeback Trail


The past decade has been tough on the Japanese semiconductor industry. South Korean producers have dislodged Japan's chipmakers from their once-dominant position in memory chips. U.S. companies have continued to lead in high-margin chip design. And nimble Europeans have been grabbing business from once-stalwart customers such as Japan's Pioneer Corp. This triple assault has turned the likes of NEC (NIPNY), Hitachi (HIT), Fujitsu (FJTSY), and Mitsubishi Electric (MIELY), which together once dominated the chip business, into bit players. In 1988, Japan accounted for more than half of the world's semiconductor sales; by 2001, that figure had fallen to 26%.

Now, Japan's chip houses are fighting back. The leading semiconductor makers have spent the past year negotiating mergers or spinning off pieces of their once-mighty chip empires. They're pooling research and technology in joint ventures aimed at developing sophisticated chips for promising mobile-phone, consumer-electronics, and auto markets. And the government and leading universities are pitching in with projects to develop new semiconductor technologies. "Japanese industry has come through a period of maximum confusion and has figured out a comeback strategy," says analyst Stephen Usher of HSBC Securities Japan Ltd.

Leading the way is NEC, a titan in chip design that somewhere along the way forgot about making profits. On Nov. 1, NEC spun off its chip unit into a wholly owned subsidiary called NEC Electronics. The company hopes that a stand-alone chip operation will allow managers to focus their energies on building sales and profitability in the $6 billion division. Even before the company launched, NEC Electronics President Kaoru Tosaka shut a factory in Britain and drafted a new strategy to slim down product lines and focus on tailoring chips for the auto, electronics, and wireless industries. "The choice is simple," says the straight-talking Tosaka. "We change or go out of business."

Hitachi Ltd. and Mitsubishi Electric Corp. are planning a similar move. In April, the two will merge their non-memory chip operations into a new company called Renesas Technology Corp., which will focus on devices for cars and cell phones. The company will dominate the world market for microcontrollers--data-crunching processors that go into everything from microwave ovens to power plants--and inherit a hyperefficient factory from Hitachi. So far, though, the company hasn't made firm plans to close outdated plants or eliminate duplicate product lines. Renesas "has the technology and capacity" to succeed, says Yoshiharu Izumi, a senior analyst with J.P. Morgan Chase & Co. in Tokyo. "But it's not clear they'll install the right kind of management."

Good management is the key to survival for the new ventures. For evidence of that, they need look no farther than Elpida Memory, formed three years ago when NEC and Hitachi combined their money-losing operations in commodity memory chips. The two partners were unable to agree on direction, and miscalculated demand for faster PC memory chips. Since its launch, Elpida has racked up some $3 billion in losses. Next April, Mitsubishi Electric plans to join the group, but its arrival is unlikely to help the bottom line: Mitsubishi lost $390 million on memory operations for the year ending in March, 2002.

Now, Elpida has a new CEO and a new turnaround strategy. Yukio Sakamoto, a 27-year veteran of Texas Instruments, came on board in November. His plan is to woo Japanese clients more aggressively and outsource half of his production. Some 20% of Elpida's sales are to PC king Dell Computer Corp. (DELL) But to survive, Elpida needs customers at home. By developing chips better suited for consumer-electronics and mobile devices, Sakamoto thinks he can sell half of his production to Japanese consumer-electronics makers, many of which now purchase from Korea's Samsung Electronics Co. and Micron Technology Inc. (MU) of the U.S. His biggest challenge will be to scrape together $500 million to ramp up production next year at a new plant in Hiroshima.

The big problem with the chipmakers' strategies is that they're all heading in the same direction. With the exception of Elpida, the new players say they'll go after more profitable system chips--specialized microprocessors that provide the brains for everything from electronic toys to V-12 engines. That's the same segment targeted by Samsung, Intel (INTC), and Texas Instruments (TXN), all of which have excelled by concentrating on specific segments. Japanese companies, say industry experts, need to sharpen their focus and design chips for, say, Internet-ready cell phones, home appliances, or large flat-screen TVs. "The winners will be those that can dominate in their area of specialization," says Tsuyoshi Kawanishi, the former head of Toshiba's chip unit who is now a senior adviser to Shanghai-based Semiconductor Manufacturing International Corp.

To help the chipmakers get back on their feet, the Japanese government is pitching in. For the first time since the 1970s, the government is partnering with industry in a bid to boost competitiveness. Last spring, Tokyo earmarked $250 million for a test facility to make new, superdense chips where researchers from 11 chipmakers will work. And in June, the government opened two new clean rooms to start testing even denser system chips in cooperation with 25 companies.

The academic community is doing its part, too. At Tohoku University, scientists are cooperating with a dozen semiconductor companies to develop three-dimensional chip technology that could create even smaller chips and cut production costs by 90%. "Japanese industry may be in a slump, but there are many proposals that will give it a boost," says Tadahiro Ohmi, the head of the 3-D chip project. It's too early to say the future of Japan's chipmakers is bright, but at least they're finally moving in the right direction. By Irene M. Kunii in Tokyo


American Apparel's Future
LIMITED-TIME OFFER SUBSCRIBE NOW

(enter your email)
(enter up to 5 email addresses, separated by commas)

Max 250 characters

 
blog comments powered by Disqus