The Virginian-Pilot
                            THE VIRGINIAN-PILOT  
              Copyright (c) 1996, Landmark Communications, Inc.

DATE: Sunday, May 26, 1996                  TAG: 9605280205
SECTION: BUSINESS                PAGE: D1   EDITION: FINAL 
SOURCE: BY DAVE MAYFIELD, STAFF WRITER 
                                            LENGTH:  148 lines

VIRGINIA'S BIG DREAMS...REST ON A SMALL CHIP FOR THE INDUSTRY - AND THE STATE - SUCCESS IS A MATTER OF TIMING

The announcements, barely a month apart, seemed sharply contradictory.

In mid-April, Motorola Inc. said that because of slumping prices for computer chips, it would delay by a year the groundbreaking of a $3 billion plant in the Richmond area that could eventually employ 5,000 people.

Then last week, Motorola said it had formed a partnership with Germany's Siemens AG to build another chip plant near Richmond. And executives of the two companies said this factory, which would cost $1.5 billion and employ as many as 1,500, would be on a ``fast-track'' construction schedule.

What gives?

There's no simple answer, but largely it has to do with timing.

Developing and manufacturing semiconductors, those tiny chips that are ever more pervasive in our homes, autos and work tools, has become extraordinarily expensive.

So it's critical for those building ``fabs,'' as chip plants are known in the industry, to open them when prices are peaking for the products they make.

In the case of Motorola's $3 billion Goochland County plant, company executives reasoned it was better to wait another year. They hope better timing will translate into stronger demand and firmer prices for the plant's intended products: a new generation of microprocessors that function mainly as the brains of personal computers.

On the other hand, Motorola and Siemens are gambling that their best shot with the initial product of that joint-venture plant in Henrico County is to get to market fast, and hopefully, first. The Henrico plant, when it opens in mid-1998, will make a dramatically more powerful generation of another kind of chip used to store data.

Different products, different market conditions.

In either case, if the timing is wrong, Motorola stands to lose hundreds of millions of dollars in revenues to competitors in the global chip industry. And if the misstep is serious enough, the company's long-term competitive position could be weakened.

``It's very high-stakes and it's what keeps semiconductor executives up at night trying to make the right call on when to move forward,'' said Ken Phillips, chief spokesman for Motorola's Phoenix-based semiconductor products unit.

The goal in opening a new factory is to recover the initial investment within three years, said Angelos Angelou, a market strategist in Austin, Texas, who specializes in the semiconductor industry. Otherwise, he said, ``there is a good chance that the plant will not pay for itself.''

The reason is the industry's own remarkable record of improvement.

Every three or four years, companies like Motorola, Intel Corp. and Japan's NEC Corp. are churning out new generations of chips that are several times more powerful, yet the same size or smaller. The new generation commands a premium price, and its predecessors' prices fall precipitously.

Jeff Weir, a spokesman for the Semiconductor Industry Association, said heavy investment is necessary to keep up this pace. On average, the San Jose, Calif.-based trade group estimates its member companies plow back 31 percent of their gross revenues each year into research and development, and capital expansion - a rate Weir said is unmatched by any other industry.

Phillips uses Motorola's PowerPC chip, the microprocessor it plans to make in the Richmond area, as an example of the breakneck development rate. Two years ago, Motorola's Austin plant was cramming the transistors in the PowerPC chips to within 0.8 microns of each other. That measure, known as ``line width,'' translates into a gap equaling four-fifths of a millionth of a meter.

Incredible as that sounds, the Austin plant's chips now have line widths of 0.5 microns. And the Richmond-area plant, when it starts production at its new target date of early 1999 will be hitting line-width targets of half that or less, Phillips said.

With more and more transistors crowded into the same-sized chip, each new generation becomes increasingly more powerful.

Austin will go on making PowerPC chips, Phillips said. But if tradition holds, the Richmond plant's chip prices and profit margins will, at least initially, be higher.

What makes hitting the profit window for a new product like Motorola's particularly tough, however, is the semiconductor industry's cyclical nature.

Although global chip sales over the past 15 years have grown far faster than the world economy in general, there have been three slumps lasting from one to three years during that period.

The industry is tilting toward what seems like another slump, this one brought on by over-aggressive construction of manufacturing plants. That has caused prices for some chips to plummet by as much as 50 percent over the past six months.

The Semiconductor Industry Association last week revised sharply downward its 1996 forecast. It now predicts global sales of chips will rise about 7 percent this year, compared with a prediction last November of a 26 percent jump.

That's a dramatic slowdown from last year, when the association figured worldwide sales rose 42 percent to $144 billion.

Weir, the association spokesman, said industry executives generally believe the slump will pass quickly. Indeed, the association is predicting sales increases of 10 percent, 16 percent and 20 percent in 1997, 1998 and 1999, respectively.

But he acknowledged that the slowdown has shaken the industry and prompted some moves by individual players to ease their risks.

Plant delays are one result.

Motorola joined several big industry players in slowing construction plans. Micron Technology Inc. has even halted construction of a partially completed factory in Utah.

Another risk-averting move has been partnerships. They spread the cost of developing and building new chips, an increasing number of which are hitting the market to meet ever-more-specific applications.

Motorola-Siemens is just one partnership example. Another is IBM Corp. and Japan's Toshiba Corp., which are teaming to build a plant in Manassas. Motorola is also building chips in a joint venture with Toshiba in Japan. Japan's Sharp Corp. and U.S. leader Intel are involved in still another.

In spite of all the risks, Motorola's Phillips said there is no doubt his company will move ahead with both of its Richmond-area ventures, even the delayed Goochland County facility.

In a world hungering for more and more chips, the potential rewards from such investments are too great not to proceed, he said.

It's the timing that matters.

``You just want to get in on that far lefthand side of the demand and price curves,'' Phillips said. ``That's where the highest profits are.'' ILLUSTRATION: Graphics

THE FOUR LEADING TYPES OF CHIPS

A chip is not a chip. That's the first thing you need to know

about semiconductors. They come in many varieties. Here are the four

leading chips, in worldwide dollar sales, according to the

Semiconductor Industry Association.

DRAMs. Projected 1996 worldwide sales: $39.8 billion. DRAM stands

for dynamic random access memory. These chips are used to store

data, principally in personal computers and video-game machines -

but they lose their information once power is turned off. Motorola

Inc. and Germany's Siemens AG, at their plant in Henrico County near

Richmond, plan to make 64-megabit DRAMs, which will have four times

more data-storage capacity than what is currently available.

-Microprocessors. Projected '96 sales: $16.5 billion. These are

multipurpose devices whose main use is as the brain of personal

computers. Also used in laser printers, phone switches, among

others. These are the types of chips Motorola plans to make in

Goochland County near Richmond.

-Microcontrollers. Projected '96 sales: $12.1 billion. These

chips, the most ubiquitous of them all, control specific activities

like anti-lock braking systems in autos. They're also widely used in

appliances and consumer electronics.

-Microperipherals. Projected '96 sales: $10.2 billion. These

mainly control so-called ``peripheral'' elements of a personal

computer, such as modems and audio cards.

--Dave Mayfield.

RELAX, IT'S NOTHING NEW...

This year's projected slowdown in worldwide chip sales fits a

cyclical pattern in the industry since 1980. In that time period,

there were three previous down cycles followed in each case by

several years of robust growth.

SOURCE: Semiconductor Industry Assocation

[For complete graphic, please see microfilm] by CNB