The Global Chip Payoffby
Texas Instruments' high-speed telecommunications chip may look like any other semiconductor. But it's the product of a world's worth of effort. Conceived with engineers from Ericsson Telephone Co. in Sweden, it was designed in Nice with software tools the company developed in Houston. Today, the TCM9055 chip rolls off production lines in Japan and Dallas, gets tested in Taiwan, and is wired into Ericsson line-cards that monitor phone systems in Sweden, the U.S., Mexico, and Australia.
Sound complicated? It is. Rather than build $1 billion factories around the world, most chip companies make most of their devices at home and ship them out by plane. Not Texas Instruments Inc. More than 40% of the employees in its $6.8 billion semiconductor group work outside the U.S. And it has cutting-edge fabrication plants--known as "fabs"--in more countries than anyone.
Some industry leaders question TI's globetrotting ways. With demand for chips soaring, most manufacturers are selling everything they can produce without jumping through hoops overseas. Rather than go offshore, Intel Corp., for one, is expanding fabs in New Mexico, Oregon, and Arizona. "We put our plants where they can be cost-competitive," says Intel Chief Operating Officer Craig R. Barrett.
But industry sentiment is swinging in TI's direction. Countries such as Taiwan and Germany are offering companies big incentives to build locally. A regional base also gives chipmakers a valuable hedge against currency swings. And a local presence can win contracts from customers who are clamoring for a new species of customized chip that requires close collaboration to design.
In short, TI's strategy is paying off. Some 65% of its chip sales take place outside the U.S. In contrast, rivals Intel and Motorola Inc. still sell half their chips in America, and NEC Corp. gets 60% of sales from its home market in Japan (chart, page 47). So, Dallas-based TI is poised to tap market growth wherever it occurs. "No one can beat TI's ability to design and manufacture semiconductors overseas," says G. Dan Hutcheson, president of market researcher VLSI Research Inc. in San Jose, Calif.
As a result, TI is getting ready to grab more than its share of the biggest bonanza in high-tech history. Despite the occasional glitch, the chip market is booming as never before: 1995 will be the third straight year of 30% growth--and possibly higher. And it's a decidedly global phenomenon. Dataquest Inc. expects worldwide chip sales to rocket from $110 billion this year to $273 billion in 2000. Some 61% of that growth will be outside the U.S.
BIG GAINS. Compared with the 1980s, when the U.S. seemed to have a near-monopoly on new, computerized products, a huge number of digital gizmos are being conceived and crafted around the world. Whether they are cell phones from Nokia in Scandinavia or Sony entertainment systems on seat backs in Boeing 777s, TI chips are designed-in from the beginning. "We're intent on using our geographic position as a competitive advantage," declares Thomas J. Engibous, president of TI's $6.8 billion semiconductor group.
In fact, after losing market share every year since it invented the integrated circuit in 1958, TI is the only other chipmaker besides Samsung Electronics Co. to gain global market share in each of the past three years, says Dataquest. And this is no case of profitless prosperity. In the first half of fiscal 1995, ended July 18, TI posted record earnings on record sales--$508 million and $6.1 billion, respectively. Wall Street has driven shares up more than 90%, to around 155, since Jan. 1. Intel and dozens of other chipmakers have also skyrocketed. But TI handily beat the industry average.
While there may be temporary setbacks, TI's long-term bet looks solid. Geographic balance and a huge portfolio of chip products--aimed at everything from computers and telecom to autos and audio gear--protect the company if there's a sudden slowdown in European phone sales or American PCs.
Mastering this complex formula didn't come easily. The low point came in the mid-1980s, when TI was almost forced out of the massive market for dynamic random-access memory (DRAM) chips by Asian rivals armed with low-interest loans. TI's market share sank from 11% to 5% over that decade. Most other U.S. DRAM makers just gave
up. But TI stuck it out by squeezing rivals for billions in royalty payments.
TI's next step seemed, at the time, like a complete break with common sense. In 1987, it began swapping research on next-generation DRAMs with one of its toughest rivals, giant Hitachi Ltd. Critics squawked about TI giving away the store. But the collaboration helped TI keep up with a technological moving target. Over the next five years, similar deals were struck between NEC and AT&T, Intel and Sharp, and the triumvirate of Toshiba, Siemens, and IBM.
Up to this point, TI was playing catch-up with Japan. But within a year, it launched a series of cost-sharing agreements on the production side, bringing its cost of capital well below that of the Japanese. Battered by an unprecedented four-year chip recession, CEO Jerry R. Junkins broke with TI tradition and agreed to split the $1.2 billion cost for a new fab in Avezzano, Italy, with the Italian government. Finding no ill effects, Junkins completed similar deals with Taiwanese PC maker Acer, Japan's Kobe Steel, and a consortium in Singapore that included Canon, Hewlett-Packard, and the Singapore government. Most recently, it's started a joint-production company with Hitachi, called Twinstar Semiconductor Inc., whose first fab, near Dallas, will open next year.
In total, these alliances have saved TI more than $1 billion in plant investment and have left it with an envied network of cutting-edge fabs. All four of TI's production joint ventures are now doubling their capacity and delighting customers. "We produce our PCs a mile from our joint-venture fab with TI, so we can plan our production much more efficiently than before," says Acer America CEO Ronald Chwang. Acer recently O.K.'d a $1 billion expansion of the partnership's giant Taipei plant.
Even competitors give TI grudging praise. "They've done a good job of leveraging other people's money," admits Intel's Barrett. Adds Taiwan Semiconductor Manufacturing President Don Brooks: "TI brought fresh ideas about how to structure capacity deals."
Without TI's global push, the company might have missed the boom in the $23 billion DRAM market, where it makes some 30% of its chip profits, says Montgomery Securities. "Two years ago, no one wanted to be near the DRAM market," says Apple Computer CEO Michael H. Spindler. "Now, TI just sits back and says, `Thank you very much."'
TI has no monopoly on globalization, as rivals begin to imitate its formula. Japan's Mitsubishi Electric Corp. and Oki Electric Industry Co. have partnerships to build fabs in Taiwan. Other chip companies have pumped money into Taiwan's TSMC and Chartered Semiconductor in Singapore. Samsung and Hyundai are set to build their first ventures outside Korea.
But others will have to learn the same hard lessons TI struggled with. Its global strategy didn't begin to pay off until it changed the way it managed sales and manufacturing around the world. The company had long suffered from unacceptably high costs--the product of a corporate culture that rewarded regional performance rather than corporate health. Instead of joining forces to win worldwide business from multinational customers, salesmen would race past each other in the client's lobby, trying to snatch a contract. When parts were scarce, geographic unit chiefs fought one another over inventories.
Rather than spread knowledge accumulated across its 17 manufacturing sites, fab managers competed for top manufacturing honors each month. The result: Sales turned around, but profits and customer satisfaction didn't. "Just having people and functions around the world proved to be a disadvantage," says TI's Engibous.
So for the past four years, the 19-year veteran has been on a crusade to wipe out internal fiefdoms. For starters, he came up with the concept of the "virtual fab," in which high-tech factories would behave less like bickering siblings and more like a close-knit family. This meant insisting that all fabs use the same procedures and equipment to keep processes under tight control.
FEWER DELAYS. By upgrading its global information system, TI enabled engineers in Dallas to remotely operate giant testing machines at assembly plants in the Philippines that hunt for flaws deep within multimillion-circuit chips. Then, they can actually reprogram the computer-controlled production line by the time the morning shift arrives the next day. Special software lets operations managers distribute new orders to whichever fab needs the work at the moment. The result: TI can use its fabs more efficiently. With fewer delays in the factory, it has improved on-time delivery from 77% to 96% since 1990.
It's not all technical wizardry. TI stresses teamwork and communication. Country-by-country sales forces have been scrapped for a market-based approach, in which teams of marketers get global responsibility for a product's success. And teams made up of everyone from top executives to factory grunts with a flair for photolithography machines now routinely shuttle between continents to tackle problems as they crop up. When a sophisticated machine for depositing chemicals on silicon wafers kept getting contaminated by microscopic dirt at TI's fab in Avezzano, a team of "particle busters" from the U.S., Taiwan, and Japan convened to figure out why. Then they applied the findings at their home locations.
Indeed, one of TI's biggest advantages is a tight team of 10 plant builders. Over two decades of working together, they've refined the mind-boggling complexity of fab construction to an art. Today, they can sail through regulatory red tape around the globe, find suppliers of ultrapure water and chemicals, and create world-class cleanrooms an average of eight months faster than competitors, according to VLSI's Hutcheson.
The cumulative effect of these technological and management steps is huge efficiency gains. By tracking manufacturing performance of its 17 fabs as a single entity, Engibous says, TI freed up capacity equal to a brand-new, $500 million fab. "We are starting to see the kind of improvements that truly great companies make," says CEO Junkins, who is credited with shattering TI's top-down hierarchy. Says Motorola Semiconductor Group President Thomas D. George: "If TI is not doing a better job, they're hiding it well."
SPEEDY. The reason customers are pleased with TI's achievement is that it helps them move more quickly in designing and introducing new products. It used to take Nokia Mobile Phones Ltd. as much as six months to forge annual contracts that set terms for delivery, payment, and such. "If you made an agreement with a local [TI] organization, say in Europe, you had tremendous problems getting guys in the U.S. and Asia to agree to the same conditions," says Nokia Senior Purchasing Manager Kari Taimi. "But since the organization has been restructured, [the contract] is implemented immediately."
The gains haven't been consistent around the world. Two years after launching a massive reorganization that exchanged a slew of country sales groups for five continent-wide product groups, the European operation is just starting to gain ground on competitors, after five years of shrinking revenues. And compared with telecom pros Motorola and AT&T, says another longtime telecommunications customer, "TI is still weak."
But TI's efforts are paying big dividends in other regions. With more fabs in Asia than any company from outside the area, TI Asia has already passed TI Europe in size and could overtake the U.S. unit by 2000, says TI Asia President Keh-Shew Lu.
That's not surprising, given TI's technology strategy, which is centered on speedy chips called digital signal processors (DSPs). These are near-cousins to Intel's famed microprocessors. But they excel at manipulating real-world signals, such as phone calls or frames of video. That makes them ideal for consumer products, from multimedia PCs to futuristic gizmos such as pen-size communicators. In DSPs, "TI is the team to beat," says Will Strauss, president of market researcher Forward Concepts, in Tempe, Ariz. He figures the DSP market will hit $16 billion by 2000. "TI is the dominant player, and it's still continually increasing market share."
Instead of just supplying chips, TI also wants to help launch whole product categories, such as a high-definition projection TV that uses an exotic assembly of semiconductor micromirrors. TI invented these unique chips and spent 10 years developing them. Then, it was able to tap long-time partner Sony for help with the first prototypes. When it comes time to build this kind of product, says Kenji Hori, head of Sony's U.S. research labs, "Japanese engineers are the best choice."
Efforts like this highlight TI's desire to draw customers into the chip design process--and to be physically near them during product development. That's especially important in the dawning age of so-called systems-on-a-chip, for which TI has been a driving force. The idea is to etch all the electronic functions of a new product directly onto a single slice of silicon. When it works, it's cheaper than assembling a board full of stand-alone chips. But without local hand-holding, this approach becomes unwieldy. "Customers don't want to be down for 24 hours or more every time they find a bug," says Senior Vice-President Richard K. Templeton.
DANGERS AHEAD. Competition in such single chip-based systems is intense. TI is up against dynamos such as LSI Logic Corp. and Cirrus Logic Inc. NEC, Fujitsu, and Toshiba are hot on the trail as well, as are consumer-products giants Sony and Matsushita. Sony hired LSI Logic to design the guts of its hot game machine, called the Play Station. But for the next generation, "Sony will do the work in-house," says research president Hori. Sony's Atsugi (Japan) chip division is booming. And it has groomed two U.S. labs--in San Jose and San Diego--to make systems-on-chips for new digital products of exactly the sort that TI wants to pioneer.
There are other competitive dangers. TI could take a hit on its huge DRAM business if newcomers keep crowding in. Motorola's Thomas George doesn't want these memories to be more than 5% of Moto's chip sales. He thinks DRAM prices will stay high for a few years, but sooner or later, the old roller-coaster cycles will return. "It could be even worse, because there are more big players in the market than ever before," he warns.
TI insists that if it produces a wide enough variety of chip products in the right places, it can ride right over the rough spots as the industry expands. "This is not the industry it was in the 1970s and 1980s," Junkins says. "The engine that's pulling us along is bigger than ever. Where else can you find a $100 billion market with $100 billion in growth opportunities?" That tells him that chip companies must shelve their perennial worries. "You make a big mistake if you look in the rearview mirror," he says. Today, if he does glance back, he'll have the gratifying view of competitors around the world eating a lot of TI's dust.
HOW TI'S SYSTEM WORKS
-- In Sweden, Ericsson engineers brainstorm with TI experts to improve the quality on Ericsson's digital phone equipment.
-- On the sunny Riviera, a team of TI designers turn Ericsson's vision into electronic blueprints.
-- To make prototypes, TI taps Japan's manufacturing expertise through its 31-year-old Japanese subsidiary.
-- Production fans out to include Dallas, so TI can maintain a stable supply and hedge against currency swings.
-- Like most chipmakers, TI packages the finished chips in Southeast Asia. Its global information system lets engineers fix bugs remotely by tweaking automated assembly lines in Taiwan.
-- TI's chips arrive on time and are embedded in Ericsson phone switches
in the U.S., Mexico, Europe, and Australia.