With memory chips accounting for more than 50 percent of Samsung Electronics’ profits last year, its masters in Seoul, South Korea, have decided it’s too dependent upon the highly cyclical DRAM business. So McDonald’s one-trick pony is racing into new lines of business and new technologies that will have a major impact on the U.S. market. By this summer, Samsung’s trickle of flat-panel displays will become a flood. Flash memory products will appear–in volume–later this year. The company expects to se ll nearly $1 billion worth of specialty memory chips, including SRAMs and EDO RAMs, in 1995. And last month’s agreement to make a $377 million investment in AST will change the PC market in the United States and Asia (see Close Up, right).
It’s not just Samsung Electronics that’s going through a transformation. The company’s parent, the $54 billion Samsung Group, is also diversifying and moving quickly to give foreign operating units more autonomy. Samsung Group Chairman Lee Kun Hee–invariably referred to as Chairman Lee–has decreed that Samsung will add automobiles, aerospace, transportation, and entertainment to its current repertoire of electronics, chemicals, and finance. Also on the capital budget: Samsung’s first fabs in the Unit ed States, Asia, and Europe. Although the facilities may be built under joint ventures, guess who’s expected to come up with as much as half of the capital? Samsung Electronics. “There is a lot of financial pressure,” admits McDonald, senior VP for sales and marketing of Samsung Semiconductor, a subsidiary of Samsung Electronics. “But we’ve been planning for three years. You’re not going to see us fumble.”
Deep Pockets Keeping all those balls up in the air is going to be tough. Samsung can’t afford to lose profits from its DRAM operation, but at the same time its long-term health depends upon a more diversified semiconductor strategy. Even a company with pockets as deep as Samsung’s can’t do everything at once, says Chuck Goto, a Tokyo-based analyst with Smith Barney. The key task of transferring process technology from Korea to the new U.S. fab, slated for production in 1997, could be difficult. “The pace o f expansion in Korea is too rapid to divert people who really know the process,” Goto says. And Samsung is very late to market with a number of key products, including TFTs and flash memory.
Sounds tricky. But Samsung has a lot going for it. Analysts, who had expected the DRAM boom to slow this year, now think sales and prices will stay strong for at least the rest of the year. “That should mean another good [earnings] year for Samsung; it will give them more time to diversify,” says Jae Ho Rhee, an analyst with S.G. Warburg, in Seoul. Perhaps more importantly, a company that once was known as a technology copycat–a description Samsung executives don’t dispute–has come into its own. Its process technology “may be ahead of Intel’s and is nearly neck-and- neck with IBM Microelectronics–the world’s best,” says Will Strauss, president of Forward Concepts, a market-research firm in Tempe, Ariz.
How will Samsung leverage its historic strengths into new lines of business? First of all, the company will use its older DRAM lines to produce new products, says Mark Ellsberry, Samsung’s vice president for memory products. “Utilizing those fabs is an absolute economic necessity,” he says. Because the lines are already depreciated, Samsung can afford to use 1M- and 4M-bit facilities to produce lower- volume parts like flash memory, ASICs, or chip sets. Samsung’s timing is certainly right. The transition to 16M-bit DRAMs is in full swing; by next year the denser part should predominate. And that means lots of otherwise-idle 4M-bit capacity. Look also for Samsung 4M-bit parts to find their way into printers as well as set-top boxes, a small market this year, but one that will grow rapidly by 1996.
At the same time, Samsung will unleash a strategy to enter new markets at the high end–as signaled by its measured push into the flash-memory market. Last year, Samsung sold just $2 million worth of flash products, not enough to register on many radar screens. This year, says marketing director Bob Eminian, Samsung expects to boost that to at least $40 million worldwide, about half in the United States. But don’t look for Samsung to produce low-density flash, a battleground dominated by Intel and AMD. Instead, Samsung is entering the flash market at the 16M-bit level, with an architecture, called NAND, that is designed to work best in mass-storage applications such as using flash cards to replace hard drives.
“Timely market move,” comments Dataquest analyst Ron Bohn. But 16M-bit flash applications are still rare–just 2 million units were sold last year, compared to 6 million 8M-bit units and 70 million 1M-bit. But Bohn says higher density is the direction the market will move as demand for cheap, portable storage soars. Demand for 16M-bit flash products will reach 10 million units this year and soar to 40 million in 1996. What’s more, Samsung is already sampling 32M-bit flash, the first in the industry, and is well along on development of a 64M-bit part. Other strategic moves include partnering with Toshiba and National Semiconductor to establish standards and making Samsung’s higher-density parts are pin-compatible with its lower ones to provide an easy upgrade path.
Some competitors scoff–“They’re always promising, never delivering on flash,” says one AMD executive. Others wonder if Samsung, like Toshiba, will find that NAND designs are costly and difficult to implement. You’ll see Samsung’s label on flash file (in densities up to 6M byte) and ATA-style cards (up to 40M byte) by midyear, Eminian promises. For now, Samsung will be making most of the components (excluding controllers) for the cards. Partners will do the actual manufacturing. The bottom line: Intel an d AMD–which collectively hold about 76 percent of the flash market–are out of Samsung’s league for now. But smaller players, like SunDisk Corp., who aim at the storage market, had better watch out.
McDonald jokes that “flash is just a hobby for us.” But flat-panel displays are no laughing matter. Samsung’s first display plant cost $375 million, not counting the enormous development expense. By the end of the year, the plant near Seoul is expected to be producing 20,000 substrates a month, giving Samsung a market share of about 12 percent. The company plans to spend an additional $1.25 billion on R&D and on construction of a second plant. When the second facility ramps up in 1997, it will doub le Samsung’s capacity. By the end of 1999, Samsung wants production of 120,000 substrates a month, which would make it the world’s third- largest producer.
Samsung’s entry into the TFT market will be all the more significant because of its high-end strategy: Samsung will concentrate on providing scarce 10.4 inch displays. “If you come late to market, you can not come with yesterday’s technology,” says Mike Crawley, Samsung’s director of TFT displays. Although most notebooks use 9.4-inch (or smaller) panels, Crawley notes that nearly all development is for the larger sizes. “Why should we go back and try and capture a market window that is nearly gone?” he says.
By next year, Samsung will be producing 11.3-inch displays. But just in case smaller turns out to better, the plant has a pilot line that is turning out small numbers of 9.4-inch displays, and a mass line that could be reconfigured fairly easily to produce smaller displays in volume. With other South Korean vendors entering the market and Japanese manufacturers ramping, worldwide production of color active-matrix panels will jump from 1.75 million units to about 2.9 million units, while prices for 10.4-inc h panels will drop by about 20 percent, says Dataquest analyst Jack Roberts.
Never Forget But Samsung hasn’t forgotten about its core business–DRAMs–which it didn’t enter until 1983. In the early years, the company relied on reverse engineering and other “indirect technology transfers.” Tourists used to be warned to avoid flying to Seoul from Tokyo on a Friday night. Why? Moonlighting Japanese engineers, intent on selling secrets to South Korean semiconductor companies, booked the flights long in advance, says Dan Hutcheson, president of VLSI Research. But Samsung doesn’t nee d the help now. It has scored a number of impressive technological wins. Samsung was among the first to sample 64M-bit DRAMs and last summer it announced the world’s first working model of the 256M- bit DRAM. This year Samsung Electronics will raise R&D spending 16.7 percent, to $875 million, including $500 million for semiconductor and multimedia products.
Samsung is also upping capital expenditures by 30 percent, to $3.8 billion. Among its improvement plans: shoring up its weak IT infrastructure. McDonald, who previously worked for Toshiba, says Samsung’s lack of good IT made perfect sense, in a twisted sort of way. “IT allows you to provide customer service, but neither Korean nor Japanese semiconductor companies have ever made this a priority,” he says. Samsung is going to start. By 1996, facilities in Asia, Europe, and North America will be linked by an order-fulfillment system that will allow customers to order–and check the status of orders–electronically. “We want to be rated among the top three semiconductor companies in service,” McDonald says.
Chairman Lee’s ambitious plans will reportedly cost the Samsung Group nearly $10 billion this year, and billions more in the future. That will tax every unit of the conglomerate, particularly the profit-rich semiconductor operation. How will you be able to tell if Samsung Electronics is living up to the Chairman’s expectations? Here’s one way. By the end of the year, McDonald is under orders to increase the non-DRAM component of Samsung Semiconductor’s sales to about 25 percent. If you only run into Samsung as a competitor in the old, familiar places, McDonald has dropped the ball.