Samsung and Semiconductors
Good graphics at 1st link ...
>> Aiming For the Top
After reaching the chip industry's No. 2 spot, does Samsung have what it takes to overtake Intel?
Russ Arensman
Electronic Business
8/1/2004
reed-electronics.com
While most of the chip industry struggled through the rough seas of the recent recession, Samsung Electronics sailed through the storm with flying colors. Last year Samsung increased its chip sales by nearly $2 billion, to a record $10.5 billion, while its semiconductor business operating profit climbed 18 percent, to $3.7 billion. And for the first half of 2004, the company's chip unit has already achieved operating profit of $3.4 billion, nearly matching the total for all of 2003.
The Seoul-based company achieved this by dominating the memory market and targeting its most profitable segments. Since 1998 Samsung has more than doubled its chip sales and is now the second-largest chip maker in the world, behind Intel. The company became No. 1 in DRAM production in 1992, No. 1 in SRAM memory chips in 1996 and took the lead in overall flash memory sales from Intel just last year.
In 2002 Samsung surpassed Intel for the first time in terms of overall electronics revenue (which includes more than semiconductors), achieving $34 billion, compared to Intel's $27 billion. In 2003, Samsung's overall electronics revenue rose to $36.4 billion while Intel's grew to $30.1 billion. These revenues put Samsung at No. 11 and Intel at No. 14 on ELECTRONIC BUSINESS ' Top 300 (see "EB Top 300"):
reed-electronics.com
With operating profit margins up to 30 percent higher than those of its memory chip rivals, Samsung is building an insurmountable lead, says Bonjun Koo, semiconductor research director for Citigroup Smith Barney in Seoul. "The game is over in the memory market," he says. "It's going to be very tough for other memory makers to catch up."
Having secured domination of memory, Samsung is now gunning to dominate the entire chip market. Jon Kang, Samsung's senior vice president of technical marketing, is bluntly honest about the company's ambition. "The only thing left for us is to be No. 1 in semiconductors."
Samsung has a long way to go to surpass industry leader Intel, whose $30 billion in 2003 chip sales were nearly three times Samsung's $10.5 billion. But industry analysts and insiders aren't dismissing Samsung's chances. "It thinks long-term, and it thinks big," says Eli Harari, CEO of SanDisk, which uses Samsung flash chips in its memory cards. "It truly is one of the best companies out there, and it is willing to put its money where its mouth is. If it takes five or 10 years to enter a market and become a world player, that doesn't worry it a bit."
But to challenge Intel, Samsung must move beyond memory and into logic, an area in which it has struggled in the past. In memory Samsung has succeeded through a combination of aggressive capital spending, diversification of products and flexible manufacturing. Those tactics will help in logic, but Samsung will also need system and design expertise. So far, it has shown little inventiveness of its own in these areas. Rather than design its own chips, the Korean giant has so far relied largely on licensing others' technology, spending more than $1 billion last year alone on royalties.
Big Spender
Samsung sold $8.2 billion in memory chips last year, including almost $5 billion of DRAMs, making it the world's largest memory supplier for the 11th consecutive year. Much of its success is due to the company's continued investment in new technology and manufacturing capacity, even during industry contractions. Samsung has been one of the top four semiconductor companies in capital investment every year since 1999, according to IC Insights. No other DRAM maker was even close during that period. This year, IC Insights forecasts, Samsung will invest an astonishing $4.1 billion in its semiconductor business, more than any other chip maker, including Intel's $3.8 billion. Micron Technology, its closest memory competitor, plans $1.5 billion in capital investments this year.
And Samsung's spending shows little sign of slowing. In June Samsung Chairman Kun Hee Lee said that his company would invest a staggering $18.9 billion in its semiconductor business from 2004 through 2006. That would require an additional investment of more than $7 billion during each of the next two years. And that's not counting the $8.6 billion Samsung plans to spend on LCD screens during that time (see "Liquid crystal gold rush," below).
Samsung hasn't simply outspent its competitors. It also has pioneered new approaches to the memory business. In 1998, for instance, the Korean financial crisis and a devastating DRAM downturn left the company unable to afford a new factory to produce its next generation of 256-kilobit chips. Until then, the industry had always quadrupled the storage capacity of each new DRAM generation, requiring expensive new factories each time. When Samsung decided to "stretch" its existing 64-Kb fab to produce 128-Kb chips, its plan initially drew scorn.
Customers accepted the new parts, however, and Samsung's competitors eventually followed suit. "Clearly, being able to make that device in the same fab as the prior generation was a huge cost advantage," says Avo Kanadjian, an executive at Nantero who was then Samsung's U.S. vice president of memory marketing. Although born of necessity, Samsung's strategy created a new business model for the industry, and every generation of DRAMs since then has proceeded in increments of double, not quadruple, the previous generation's capacity.
Leading in Diversification
Samsung also leads the memory industry in product diversification. Several years ago, it began expanding its product range to lessen its dependency on commodity DRAMs. Today it produces more than 200 memory varieties, including DDR; DDR2; Rambus DRAM; synchronous and asynchronous DRAM; low-power mobile DRAM; graphics memory; pseudo-SRAM; and several hybrid devices that combine DRAM, SRAM and flash memory. In 2003, 64 percent of Samsung's memory sales came from consumer, graphic and mobile markets, rather than computers. "Contrary to other DRAM suppliers' one-size-fits-all approach, Samsung has been very open to developing specialty devices," says Kanadjian.
That diversity makes Samsung less reliant on commodity DRAM than its competitors and, the company claims, more profitable. It is more expensive to develop so many product variations, admits Tom Quinn, vice president of memory sales and marketing for Samsung's U.S. chip operation. But as long as the company picks its products carefully, he says, the expense is more than offset by the higher prices, and profits, available in niche markets. As a result, Samsung ships 26 percent of the industry's total memory capacity while capturing 33 percent of the total revenue, according to Quinn.
Samsung's leading-edge manufacturing also contributes to the company's success, says Nam Hyung Kim, an iSuppli memory analyst. Samsung began building 0.11-micron DRAMs and 90-nanometer flash chips well before competitors, he says, and its yields of working chips appear to be among the industry's best. "It clearly is the cost leader," he says. Samsung's production cycle for DRAMs is the industry's shortest, he says, allowing Samsung to complete a 64-Mb chip from start to finish in 30 days, nearly twice as fast as some competitors.
Yet despite its apparent cost advantages, Samsung hasn't relied on price cutting to win market share, at least not in recent years. "It is not a bargain house," says Semico Research DRAM analyst Sherry Garber. "In fact, it is known as the highest-priced vendor in the industry."
Another reason for Samsung's success is the company's ability to shift production to more-profitable markets when conditions change. Two years ago, DRAM sales to the PC industry were collapsing while soaring digital camera sales created a shortage of NAND flash-memory chips. Samsung seized the opportunity by quickly converting several of its older DRAM fabrication plants to build NAND chips. The product switch not only made Samsung the leading NAND supplier, with 2003 sales of $1.9 billion and a 45 percent market share, but it also swiftly erased the semiconductor division's losses, which in late 2001 had been running about $75 million a month.
Samsung's flash sales exploded from $268 million in 2001 to $2.3 billion in 2003, according to Semico. "Samsung went from having half a fab producing flash in 2002 to having three and a half fabs now," notes Semico analyst Jim Handy. And best of all for Samsung, its leading NAND rival, Toshiba, sold its DRAM business to Micron in early 2002, after suffering huge losses, preventing it from shifting excess fab capacity into flash production.
Samsung may soon have a harder time dominating the NAND flash business, thanks to Hynix, Infineon, STMicroelectronics and Micron, which are entering the market this year. Samsung has begun cutting its NAND chip prices in anticipation of the new competition, and Handy forecasts price declines of up to 60 percent by year's end. "We're entering an oversupply, and these new companies are just going to make that worse," he says. Handy expects supply and demand to slowly come back into balance by 2008, when, Semico forecasts, today's $4.1 billion NAND market will have grown more than six-fold, to $25 billion.
Predatory Strategy
Samsung was the most successful chip maker during the 2001-to-2003 downturn thanks to its "predatory" strategy of taking away competitors' key customers, says Derek Lidow, iSuppli's CEO. "A successful predatory strategy is not about bombing the market with cheap chips and therefore destroying your own margins," says Lidow. "It's about offering a good package of products and services to specific accounts that are important to your competitors." Samsung does that by figuring out customers' needs and adjusting its own product and manufacturing strategies to meet them, says Lidow. "It spends a lot of time understanding every socket and the needs of virtually every client and potential client," he says.
Customers and partners agree. Samsung is especially good at targeting large OEM companies such as Dell, Nokia and Hewlett-Packard, notes SanDisk's Harari. "It does a phenomenal job of serving those guys," he says. Bruce Goldberg, CEO of All American Semiconductor, which has distributed Samsung products since the 1980s, adds, "It has a great sense of knowing what the market wants, as opposed to telling the market what it needs."
Samsung is also known for its fast decisions, which are crucial to correctly timing its huge capital investments. "Its decision-making process is really, really quick compared to that of Japanese companies," says iSuppli's Kim. He contends that Japan's chip makers have traditionally relied on consensus decisions, whereas Korea's Samsung uses a faster, top-down approach.
LSI Disappointment
Although Samsung has been nearly unstoppable in memory, its attempts to diversify into other chip markets have been disappointing. In 2001 Samsung officials were predicting that the company's non-memory chip sales would grow from about $2 billion to $5 billion by 2005. As of last year, however, the non-memory "system LSI" business was stalled at about $2.2 billion, according to Gartner Dataquest, and about half of those sales came from LCD driver chips, primarily for Samsung's own booming display business.
Samsung's system LSI business uses Qualcomm technology to make CDMA cell phone chip sets but is allowed to sell them only to its own handset business. Other non-memory sales are divided between ASICs (a $340 million business last year, according to iSuppli), smart cards, consumer electronics chips, image sensors and processors, and chips for mobile devices. Among its notable design wins, Samsung has supplied ARM-based applications processors for some of HP's iPAQ pocket computers.
Yet despite years of investment, Samsung's chip division has produced few hit products besides memory. Samsung's Kang concedes that system LSI has lagged the company's memory business but adds, "One of Samsung's biggest strengths is diligence—don't be surprised if we make progress." The company's recent decision to build its first fabrication plant dedicated to non-memory chips is a key inflection point for the system LSI business, adds Seok Hyun Hong, sales and marketing director for the U.S. system LSI business. The company expects the $2 billion fab to begin producing 12-inch wafers by late 2005.
In March Samsung agreed to co-develop several generations of advanced chip technology with IBM, Infineon and Chartered Semiconductor. In a little-publicized part of that deal, Samsung licensed IBM's 90-nanometer process technology for the new LSI fab and plans to use some of the fab's capacity to produce chips for outside customers. "In a year and a half, we're going to have probably one of the best manufacturing capabilities, with the leading process technology, in the world," says Hong.
The new fab should provide a good test of whether Samsung's aggressive capital investment strategy can extend beyond memory into other chip markets. Samsung paid close to $100 million plus an exchange of patent rights for IBM's technology, estimates Richard Doherty, research director for the Envisioneering Group. But even if it enhances Samsung's LSI manufacturing, it does little to address the company's heavy dependence on outside intellectual property (IP). "Samsung is very good in capital-intensive businesses, but in products that need IP, it is pretty weak," says iSuppli's Kim. The billion dollars the company spent last year on technology royalties covered, among other things, Rambus memory, SanDisk and Toshiba flash, Texas Instruments DRAM and Qualcomm CDMA.
Trying too Hard?
Meanwhile, some wonder if Samsung is trying to be something it's not. Citigroup's Koo, for instance, worries that Samsung may overspend on its LSI business, with little to show for its investment. "I don't want to see Samsung invest a lot of money on system IC where it doesn't have strength," he says. "It still has a lot of room to make more money from memory by squeezing weaker competitors."
Others think that Samsung has little choice but to continue trying to diversify. "To keep growing as it has, it has to focus on higher levels of integration," says Nantero's Kanadjian. Non-commodity markets, although perhaps more risky, "can give you a better relationship with customers, to make you a more strategic supplier," he explains.
Lidow recently issued a report warning South Korean electronics companies, including Samsung, that their technology leadership is threatened by a lack of IP ownership. "Korea has a very large deficit in terms of IP payments," he says. His solution? Either learn how to develop more-innovative products and services or acquire them.
The cheaper option is to acquire technology, especially in today's chip industry, in which some 450 companies already compete for a market whose long-term growth is slowing. "Buying is a very attractive option now. But it has to be done properly," says Lidow. "If it's done heavy-handedly and with cultural insensitivity, then, of course, it won't work."
Samsung has a poor, and rather limited, track record with foreign acquisitions. Its best-known deal ended disastrously, after it paid $854 million in 1995 for U.S. PC maker AST Research and then lost millions more on the troubled company before closing it four years later. Recently, though, Samsung has invested in Linux developer MontaVista Software and in power-management chip maker Advanced Analogic Technologies. The company plans to make further investments in semiconductors, displays and other strategic technology through its $300 million Samsung Ventures America fund.
Clearly, Samsung has a long way to go and much to learn if it hopes to catch up to Intel. "Samsung is world class, but it's got chinks in its armor," says Lidow. "But, that said, Samsung is a competitor that everybody needs to be very aware of and very respectful of." <<
Patience, Persistence And Progress
Samsung’s chip business took more than a decade to reach profitability
Russ Arensman
Electronic Business
8/1/2004
reed-electronics.com
Samsung Group, which started in 1938 as a small trading company in Taegu, South Korea, expanded gradually into trucking, real estate, and commodities before setting up Samsung Electronics in 1969 to make parts and home appliances for Japan’s Sanyo Electric. It entered the semiconductor business in 1975 by buying the country’s first chip maker, Korea Semiconductor, which initially made chips for digital watches and later expanded into transistors for TVs and other consumer products.
In 1983 Samsung Group founder Byung-chull Lee concluded that the chip operation, which had never been profitable, needed to build higher-volume products to stanch its losses. Samsung recruited a group of experienced Korean engineers from U.S. chip companies and completed its first 64-Kb DRAM design that year. Although it managed in subsequent years to develop a series of increasingly larger-capacity DRAM chips, it was handicapped by starting out several years behind the industry leaders. At the time, U.S. and Japanese chip makers were waging a fierce battle for market control, and from 1985 to 1987 Samsung’s chip division accumulated losses of $140 million.
In 1988, however, a DRAM shortage developed after a brutal price war forced all but two U.S. DRAM makers out of the market and even Japanese makers reduced their investment and manufacturing capacity. The price of 256-Kb chips soared from $1 to $12, and suddenly Samsung was able to sell all the chips it could make.
Former Samsung Chairman Jin-ku Kang writes in his autobiography that the company’s memory chips earned $227 million in net profit that year, “which wiped out our accumulated losses and still left us with a considerable profit.” Kang admits that the company used profits from its communications division to offset its chip losses through its painful early years. But he also gives credit to Samsung’s founder and to the engineers who for years worked nearly around the clock to close the gap with their foreign rivals. Samsung’s founder Lee died in 1987, without ever seeing the chip business earn a profit.
Samsung introduced the world’s first fully functional 256-Mb DRAM in July 1994, the same year it became the world’s largest overall memory supplier and earned a profit of nearly $1.25 billion. Since then it has scarcely looked back, steadily increasing its DRAM sales and market share, which by 2002 reached nearly $5 billion—one third of the total market.—R.A.
Russ Arensman is a Glenwood Springs, Colorado-based business and technology writer and editor. <<
Top Semiconductor Companies 2003 (sales, in billions of $)
·
Rank Company Sales Share
·
1 Intel $27.10b 15.3%
2 Samsung Electronics $10.50b 5.9%
3 Renesas Technology $7.94b 4.5%
4 Texas Instruments $7.41b 4.2%
5 Toshiba $7.36b 4.1%
6 STMicroelectronics $7.18b 4.0%
7 Infineon Technologies $6.86b 3.9%
8 NEC Electronics $6.31b 3.6%
9 Freescale (Motorola) $4.63b 2.6%
10 Philips Semiconductor $4.51b 2.5%
Others $87.65b 49.4%
======= =====
Total market $177.45b 100.0%
·
Source: Gartner Dataquest
·
Samsung Electronics Co. Ltd. At-A-Glance
·
Corporate Headquarters: Seoul, South Korea
Founded: 1969 (as a subsidiary of Seoul-based Samsung Group)
Web site: samsung.com
Chairman: Kun Hee Lee
CEO and Vice Chairman: Jong-Yong Yun
Exchange/Ticker Symbol: Korea Stock Exchange/05930
Employees : 75,000
·
In billions of $ 2001 2002 2003
===== ===== =====
Sales 26.98 33.26 36.41
Profit 4.98 5.89 2.46
Return on assets 15.2 20.5 10.6
Return on equity 20.3 28.1 15.1
·
Source: Samsung Electronics
- Eric - |
