Motorola Bets on Its Chips

What do you do when the world as you've always known it simply ceases to be? When your $8 billion business shrivels to less than $5 billion in less than a year? When new competitors and new technologies explode the industry economics? When everything that worked before won't work -- and can't work ever again?

Most big companies first respond to such cataclysm by tinkering at the edges. As the crisis deepens, they attack costs with reengineering and layoffs. And when the bloodletting fails, they throw up their hands and sell the doomed unit.

A couple of years ago, Motorola's semiconductor products sector faced just such a watershed -- and came up with a vastly more innovative response. It's not yet clear that it chose the right path. But grant Motorola this: Faced with an untenable business model, it neither tinkered nor gave up. Instead, it created a radical new business model.

How radical? Within Motorola, it exploded decades of tradition. While the old Motorola had to own all of its manufacturing assets, the new Motorola outsources or shares much production. Nearly two years ago, the company's intellectual property was considered proprietary; today, key telephone-chip technology is licensed out. Motorola used to make and sell every sort of chip, because it could; now it's focusing on just a few markets that it can dominate.

Here's why it had to. In 2001, the semiconductor industry suffered a cyclical contraction of epic proportions. Cyclicality was nothing new: Every four years for decades, the chip trade has met rising demand by overbuilding manufacturing capacity -- and then inevitably has struggled with falling prices. But this time, in the wake of the dotcom and telecom busts, industry sales dropped 30% from the record year before. At Motorola, with its heavy dependence on cell-phone and networking customers, semiconductor revenues fell by 37%; the unit lost $1 billion.

As the market was collapsing, two other developments altered industry fundamentals. The first had to do with production economies. Every three to four years, engineers have figured out how to manufacture bigger silicon wafers that will yield more chips. The bigger the wafer, the lower the per-chip cost. Today, the leading edge in semiconductor manufacturing is a 12-inch wafer.

The problem: A 12-inch wafer "fab" is an enormous factory costing at least $2.5 billion to build -- about double the investment required for its smaller predecessor. And once it's built, that fab will spit out between $5 billion and $8 billion worth of silicon per year. To justify the investment, the fab must operate at at least 70% of capacity, says Will Strauss, president of Forward Concepts Co., a research firm in Tempe, Arizona.

At the same time, state-supported companies in China, Singapore, and Taiwan began building big-wafer fabs of their own. Given lower labor costs and big government incentives, these manufacturers enjoy an economic edge. Just as important, they're disrupting the old order by their willingness to produce chips for anyone with a design, which significantly lowers barriers to entry for small semiconductor players.

"All that looked pretty scary," says Ray Burgess, the gregarious Scot who runs the semiconductor group's strategy, marketing, and communications. Motorola's output wasn't big enough for the company to justify its own 12-inch fab. Nor was Motorola strong enough to compete with every upstart in every niche. "Carrying on as we had didn't make a lot of sense. We had to remake the whole rule book," says Burgess.

Rewriting The Rules New rule number one: If we don't have to own it, let's not own it. (And if we do have to own it, let's reduce the risk by sharing it.) In 2000, Motorola owned 18 wafer fabs. By this month, it will have just 8. At present, 30% of its chip revenue comes from products made, tested, or assembled by Taiwan Semiconductor Manufacturing Co. and other contractors. And Motorola has created a joint venture with Philips and STMicroelectronics that will invest $1.4 billion in new semiconductor designs and processes by 2005.

New rule number two: If we can't dominate a category, let's get out. "We could, in theory, make something equivalent to anything that's sold today," says Burgess, who explains that there are hundreds of niches that Motorola could attack. But given the fragmentation of design expertise and the entry of so many capable competitors, "the question has become not, How can I make everything? but, What is it that I want to make? And if someone else can make it better, why not just exit?" In the mid-1990s, Motorola sold 70,000 different semiconductor products; today, that number is down to around 10,000.