Xilinx, Inc.

views updated May 23 2018

Xilinx, Inc.

2100 Logic Drive
San Jose, California 95124-3400
U.S.A.
Telephone: (408) 559-7778
Fax: (408) 559-7114
Web site: http://www.xilinx.com

Public Company
Incorporated: 1984 as Xilinx, Inc.
Employees: 3,295
Sales: $1.73 billion (2006)
Stock Exchanges: NASDAQ National Market
Ticker Symbol: XLNX
NAIC: 334413 Semiconductor and Related Device Manufacturing; 511210 Software Publishers

Xilinx, Inc., designs and develops advanced programmable logic devices and related software development systems. It is the world's leading supplier of programmable logic devices. In essence, the company's customized chips sequentially order the logic that tells sophisticated electronic gear the order in which its functions are to be performed. The company is credited with launching the "fabless" trend by outsourcing its manufacturing from the start.

Xilinx bolted to the forefront of its industry during the late 1980s and early 1990s through breakthrough product innovations. Following a dramatic slowdown in the telecommunications industry in the tech bust, the company has benefitted as its ever more powerful, cost-effective, and energy-efficient chips have found their way into new applications such as mobile electronics.

FOUNDED 1984

Xilinx was founded in 1984 by Ross Freeman and Bernard Vonderschmitt. Freeman and Vonderschmitt were both working as chip engineers at Zilog Corp. prior to joining in the Xilinx venture. Zilog, a subsidiary of oil behemoth Exxon Corp., was a developer of integrated circuits and related solid-state devices and had numerous technological innovations to its credit. It was there that Freeman came up with the idea that would soon make Xilinx larger than Zilog. He wanted to design a computer chip that effectively acted as a blank tape, allowing the user to program the chip himself rather than having to purchase a preprogrammed chip from the manufacturer.

Freeman was on the cutting edge of changes that were beginning to occur in the semiconductor industry. Prior to the mid-1980s, most computer chip manufacturers were interested almost solely in mass-market chips that could be produced in large volumes and sold for big profits. For several years that strategy was profitable for U.S. chipmakers, who dominated the global semiconductor market. When low-cost foreign manufacturers, particularly in Japan, began competing for market share, traditional manufacturers, including Zilog, suffered in what eventually became a commodity industry. At the same time, chip consumers began demanding increasingly specialized chips that could be used for specific applications.

Few big chip producers were excited about the prospects of chasing the market for application-specific circuits. Serving those customers meant designing and manufacturing many different chips, each of which would be sold to much smaller markets and at a lower total profit in comparison with mass-market chips. Because of the reluctance of chip makers to cater to their needs, consumers of application-specific circuits also were frustrated. Besides having to pay a relatively high price for customized chips, they were usually forced to endure costly problems related to defects in their chips. Specifically, if the semiconductor had a flaw or if the customer's chip requirements suddenly changed, the customer would have to wait for several weeks or even months for a new chip. Such a holdup could cost millions of dollars if an entire project was stalled while waiting for the new semiconductor.

Freeman realized that there might be a better way of meeting the need for application-specific circuits. His idea was to develop a sort of blank computer chip that could be programmed by the customer, thus minimizing risks associated with faulty chips and allowing much greater flexibility for companies designing equipment that incorporated the chips. The technology became known as "field programmable gate array," or FPGA. Freeman, who was a vice-president and general manager at Zilog at the time, approached his superiors and suggested that the development of FPGA devices could be a viable new avenue for Zilog. However, he was unable to convince executives at Exxon, who controlled more than $100 billion in assets, to chase a totally unexplored market that was worth perhaps only $100 million at the time.

Confident of the practicality of his concept, Freeman left his post at Zilog and began developing the first FPGA chip. He joined forces with another Zilog expatriate, Bernard Vonderschmitt. The 60-year-old Vonderschmitt also had been working as a vice-president and general manager at Zilog. Prior to that he had served 20 years at RCA, where he had headed the solid-state division. Their combined brainpower and management experience allowed them to attract several million dollars of venture capital, which they used to design the first commercially viable field programmable gate array.

In February 1984 they incorporated the venture as Xilinx. It was combined with their research and development limited partnership several months later, and in April 1990 the company was converted to a Delaware corporation. Xilinx began selling its first product in November 1985.

Xilinx's FPGA was based on the company's patented Logic Cell Array technology. The company's system basically consisted of an off-the-shelf programmable chip and a software package that could be used to program and tailor the chip for specific needs. The technology was based on the arrangement of gates (the lowest level building block in a logic circuit) in complex formations called arrays; as the number of gates increased, the more complex were the functions that the semiconductor could perform. The advantage of Xilinx's system was that the software allowed the customer to program the gates and arrays, in a manner analogous to a connect-the-dots puzzle, to perform any number of different functions. Also integral to the success of the system was a small family of advanced standard semiconductors, which were manufactured for Xilinx under license by Seiko Epson in Japan.

Xilinx's FPGA systems ultimately lived up to Freeman's original vision, providing greater flexibility for equipment manufacturers and minimizing problems caused by traditional chip manufacturing methods. The company's first products offered less complexity (i.e., fewer gates) than non-field-programmable devices available at the time. By late 1987, after injections of venture capital amounting to more than $18 million, Xilinx was offering a new generation of FPGA chips that, with 9,000 gates, could compete technologically with all but the most advanced non-field-programmable products. The result was that in 1987, after marketing its products for little more than one year, Xilinx was generating revenues at an annualized rate of nearly $14 million.

COMPANY PERSPECTIVES

Xilinx is the world's leading innovator of programmable logic solutions. We supply customers with "off-the-shelf" logic devices that customers can program to perform specific functions using the development tools we provide. This programmability provides a revolutionary alternative to fixed or custom logic devices that typically require many months to design, test, and manufacture. Xilinx customers enjoy the benefit of faster time-to-market and increased product design flexibility as a result.

MARKET GROWTH AND COMPANY SUCCESS

As Xilinx was earning respect for its FPGA technology, the market for application-specific circuits continued to grow during the late 1980s and into the 1990s. The result was that the market for FPGA chips surged, contributing to rapid revenue and profit growth at Xilinx. Indeed, sales rose to nearly $30.5 million in 1988 (fiscal year ended March 30, 1989) before rising to $50 million in 1989. Xilinx posted its first surplus—a net income of $2.92 million—in 1988 and went on to generate profits of $6 million in 1989. Unfortunately, Freeman died in 1989. Vonderschmitt took the reins as president and chief executive.

Xilinx was aided during the late 1980s by a partnership with Monolithic Memories Inc. (MMI), which in 1987 signed a deal with Xilinx that gave MMI royalties and free patent rights to manufacture Xilinx's products. In return, MMI supplied capital to Xilinx. The arrangement provided needed funding to bring Xilinx's products to market and sustain its research and development initiatives. Soon after signing the deal, though, MMI was purchased by American Micro Devices, Inc. Xilinx became uncomfortable with the new arrangement, partly because American Micro was one of its competitors. So, in 1989 Vonderschmitt convinced American Micro to buy 20 percent of the company at a 10 percent premium and dissolve the original agreement. In search of new funding, Xilinx went public with a stock offering on the NASDAQ over-the-counter exchange.

With cash from the stock sale, Xilinx continued to grow. By late 1990 the company was selling its products throughout the United States, but was also shipping about 30 percent of its output overseas. More than half of its revenues were attributable to its popular XC3000 family of FPGA systems, but Xilinx had a stream of products in its development pipeline. The company sold nearly $100 million worth of its products in 1990, and its base of 3,500-plus customers grew to include big names like Apple Computer, IBM, Compaq Computer, Hewlett-Packard Co., Fujitsu, Sun Microsystems, and Northern Telecom. To keep pace with demand, Xilinx moved into a 144,000-square-foot plant in San Jose, California.

Although Xilinx rapidly increased sales and profits in the late 1980s and early 1990s, it also ceded much of its market share. Indeed, after Xilinx had invented its niche and controlled 100 percent of the FPGA market during the mid-1990s, other companies began offering competing technology that rapidly eroded Xilinx's dominance. By the early 1990s, in fact, Xilinx was controlling only about 65 percent of the total FPGA market. It had succeeded, though, in pressuring many of its earliest competitors out of the business, despite the fact that some of them had access to much more funding. By 1993 only a few companies were seriously vying for market share. The largest was Actel, which had introduced its first product in 1988 and by 1993 was serving about 18 percent of the market. The distant third-place contender was Altera Corporation, which sold technology similar to FPGA systems.

Despite loss of market share, high demand allowed Xilinx to boost revenues to $135 million in 1991 (fiscal year ended March 30, 1992) and then to $178 million in 1992. Part of that growth resulted from Xilinx's jump into the market for EPLDs (EPROM technology-based complex Programmable Logic Devices), which effectively offered higher performance and higher density per chip, and were designed to complement FPGA devices. By 1993 Xilinx was capturing more than $250 million in annual revenue and generating net income of $41.3 million. In addition, market growth for FPGA chips was expected to intensify in the mid-1990s.

KEY DATES

1984:: Former engineers from Exxon's Zilog subsidiary form Xilinx to produce new programmable logic chips.
1987:: Annualized revenues reach $14 million.
1988:: Company makes first profit as sales exceed $30 million.
1989:: AMD buys a 20 percent holding in Xilinx, which goes public.
1996:: Revenues exceed $500 million in 1995/96 fiscal year; company has 1,000 employees.
1998:: Virtex series offers industry's first million-gate FGPA.
2000:: Revenues exceed $1 billion in 1999/2000 fiscal year.
2002:: Xilinx posts loss in tech slowdown; IBM contracts to manufacture some chips for hybrid processor.
2004:: Toshiba signs on as chip producer as IBM contract ends; Seiko and UMC continue as suppliers.
2005:: Sales exceed $1.5 billion; company has about 3,000 employees.

To take advantage of market expansion, Xilinx introduced a completely new line of FPGA products in 1994: the XC5000 family of FPGA chips and software. The XC5000 line was developed to cater to the market for low-end gate array products. Specifically, XC5000 chips were designed to offer a cost-effective alternative to high-volume non-field-programmable gate array products, thus giving Xilinx access to a new spectrum of the market for application-specific circuits. Xilinx followed the introduction of the XC5000 family with other new products, including the XC3100L and XC4000L. Both new families of FPGA chips were designed to complement low-power applications, which boosted Xil-inx's access to manufacturers of increasingly popular low-power devices like portable computers and related peripheral devices, portable and wireless communication gear, and digital cameras.

New products and healthy growth in demand for existing FPGA chips helped Xilinx to boost its sales to $335 million in 1994, a record $59.28 million of which was netted as income. Likewise, revenues rose to about $550 million in 1995 (fiscal year ended March 30, 1996). By that time, Xilinx was employing more than 1,000 workers in offices throughout North America, Asia, and Europe. The company was selling more than 40 varieties of programmable logic products and related software applications and continued to be firmly entrenched as the leader in its industry niche.

Early in 1996 Vonderschmitt stepped down as chief executive of the company he had co-founded. Vonderschmitt remained as chairman of the board for another seven years and passed away in 2004.

Control of the company went to Willem P. "Wim" Roelandts. The 51-year-old Roelandts had worked for 28 years at Hewlett-Packard Co., serving as senior vice-president and managing the company's Computer Systems Organization, among other posts. He was known as a seasoned high-tech industry veteran and was chosen to lead Xilinx into a "new era of growth," according to Vonderschmitt.

A new line of high-end FPGAs called Virtex came out in 1998. Offering the industry's first million-gate FPGA, the series gave Xilinx a perceived technological edge over its rivals. Xilinx and Altera each held a little more than 30 percent of the PLD market in 1999, according to Electronic Business, although Altera was then a bit larger. Over the next several years, Xilinx would greatly expand its market share with its new products, while Altera's would remain static. Xilinx posted net income of $127 million on revenues of $614 million in the fiscal year ended March 28, 1998. It had about 1,400 employees.

TECH BUST AND RECOVERY

Xilinx's revenues just exceeded $1 billion in the fiscal year ended March 31, 2000. Unfortunately, the PLD (programmable logic devices) market would be nearly halved in the tech and telecom slowdown. Xilinx was burdened with millions in excess inventory. Though revenues were up to $1.7 billion in the 2000–01 fiscal year, net income slipped from $653 million to $36 million.

Eight years of legal wrangling came to an end in 2001 when Altera agreed to pay Xilinx $20 million to settle a patent infringement dispute. The two also agreed on a moratorium on intellectual property suits between the companies for the next five years.

At the time, both were reeling from the tech slowdown. Xilinx lost $113 million on revenues of $1 billion in 2002. However, new markets were opening for PLDs. Designers had typically used the flexible PLDs in prototypes, while switching to ASICs (application-specific integrated circuits) for mass production. The increasing complexity of smart phones and other multifunction electronic devices made them ideally suited for PLDs—once their price of these chips came down enough to make them practical for high-volume applications. Xilinx's new CoolRunner chips were compact and energy efficient, making them suited for portable systems.

Roelandts told Forbes that FPGAs would soon change the way computers were designed. Xilinx supplied the chips to a new web server from newly-formed Wincom that did away with the microprocessor, which was once the central feature of any computer system. FPGAs, which performed tasks in parallel, could perform some repetitive jobs more quickly than the traditional CPU. A number of third parties were designing add-on products with Xilinx's FPGAs to make traditional processor-based computers run many times faster.

Xilinx worked with IBM Microelectronics to build a hybrid system that teamed the Virtex-II FPGA with IBM's PowerPC microprocessor. In 2002, IBM contracted to manufacture the Virtex chips itself, the first time it had entered a volume production agreement for an outside party. Japan's Toshiba Corp. took over this manufacturing in 2004 after the arrangement with IBM lapsed. Xilinx chips continued to be produced by Seiko in Japan as well as Taiwain's United Microelectronics Corp. (UMC).

Xilinx commanded the PLD market with a 50 percent share, according to Electronic Business. Its revenues were $1.7 billion in the fiscal year ended April 1, 2006, while net income had climbed to $354 million. Altera was still its only close rival. Xilinx had grown to employ about 3,000 people.

Not all of Xilinx's growth had been organic. Acquisitions included that of San Jose programming company LavaLogic, bought from Canada's TSI Telsys Corp. in 2000. Santa Clara's Hier Design, Inc., which made software for designing FPGAs, was purchased in 2004. Next, Xilinx bought AccelChip, Inc., a producer of software for designing digital signal processing systems, in January 2006 for $19.6 million.

Dave Mote

Updated, Frederick C. Ingram

PRINCIPAL SUBSIDIARIES

Xilinx Asia Pacific Ptd. Ltd. (Singapore); Xilinx Hong Kong Ltd.; Xilinx Ireland; Xilinx Limited (United Kingdom); Xilinx K.K. (Japan).

PRINCIPAL COMPETITORS

Altera Corporation; Lattice Semiconductor Corporation.

Xilinx, Inc.

views updated May 21 2018

Xilinx, Inc.

2100 Logic Drive
San Jose, California 95124
U.S.A.
(408) 559-7778
Fax: (408) 559-7114

Public Company
Incorporated: 1984
Employees: 1,144
Sales: $550 million (1996)
Stock Exchanges: NASDAQ
SICs: 3674 Semiconductors & Related Devices; 7372 Prepackaged Software

Xilinx, Inc. designs, develops, and sells advanced programmable logic devices and related software development systems. It is the leading global supplier of programmable logic. In essence, the company’s customized chips sequentially order the logic that tells sophisticated electronic gear the order in which its functions are to be performed. Xilinx bolted to the forefront of its industry during the late 1980s and early 1990s through breakthrough product innovations.

1980s Founding

Xilinx was founded in 1984 by Ross Freeman and Bernard Vonderschmitt. Freeman and Vonderschmitt were both working as chip engineers at Zilog Corp. prior to joining in the Xilinx venture. Zilog, a subsidiary of oil behemoth Exxon Corp., was a developer of integrated circuits and related solid-state devices and had numerous technological innovations to its credit. It was at Zilog that Freeman came up with the idea that would soon make Xilinx larger than Zilog. He wanted to design a computer chip that effectively acted as a blank tape, allowing the user to program the chip himself rather than having to purchase a preprogrammed chip from the manufacturer.

Freeman was on the cutting edge of changes that were beginning to occur in the semiconductor industry. Prior to the mid-1980s, most computer chip manufacturers were interested almost solely in mass-market chips that could be produced in large volumes and sold for big profits. For several years that strategy was profitable for U.S. chipmakers, who dominated the global semiconductor market. But when low-cost foreign manufacturers, particularly in Japan, began competing for market share, traditional manufacturers, including Zilog, suffered in what eventually became a commodity industry. At the same time, chip consumers began demanding increasingly specialized chips that could be used for specific applications.

Few big chip producers were excited about the prospects of chasing the market for application-specific circuits. Serving those customers meant designing and manufacturing many different chips, each of which would be sold to much smaller markets and at a lower total profit in comparison with mass-market chips. Because of the reluctance of chip makers to cater to their needs, consumers of application-specific circuits also were frustrated. Besides having to pay a relatively high price for customized chips, they were usually forced to endure costly problems related to defects in their chips. Specifically, if the semiconductor had a flaw or if the customer’s chip requirements suddenly changed, the customer would have to wait up to several weeks or even months for a new chip. Such a holdup could cost millions of dollars if an entire project was stalled while waiting for the new semiconductor.

Freeman realized that there might be a better way of meeting the need for application-specific circuits. His idea was to develop a sort of blank computer chip that could be programmed by the customer, thus minimizing risks associated with faulty chips and allowing much greater flexibility for companies designing equipment that incorporated the chips. The technology became known as “field programmable gate array,” or FPGA. Freeman, who was a vice-president and general manager at Zilog at the time, approached his superiors and suggested that the development of FPGA devices could be a viable new avenue for Zilog. But Freeman was unable to convince executives at Exxon, who controlled more than $100 billion in assets, to chase a totally unexplored market that was worth perhaps only $100 million at the time.

Xilinx’s FPGA was based on the company’s patented Logic Cell Array technology. The company’s system basically consisted of an off-the-shelf programmable chip and a software package that could be used to program and tailor the chip for specific needs. The technology was based on the arrangement of gates (the lowest level building block in a logic circuit) in complex formations called arrays; as the number of gates increased, the more complex were the functions that the semiconductor could perform. The advantage of Xilinx’s system was that the software allowed the customer to program the gates and arrays, in a manner analogous to a connect-the-dots puzzle, to perform any number of different functions. Also integral to the success of the system was a small family of advanced standard semiconductors, which were manufactured for Xilinx under license by Seiko Epson in Japan.

Xilinx’s FPGA systems ultimately lived up to Freeman’s original vision, providing greater flexibility for equipment manufacturers and minimizing problems caused by traditional chip manufacturing methods. The company’s first products offered less complexity (i.e., fewer gates) than non-field-programmable devices available at the time. But by late 1987, after injections of venture capital amounting to more than $18 million, Xilinx was offering a new generation of FPGA chips that, with 9,000 gates, could compete technologically with all but the most advanced non-field-programmable products. The result was that in 1987, after marketing its products for little more than one year, Xilinx was generating revenues at an annualized rate of nearly $14 million.

Market Growth and Company Success

Xilinx was aided during the late 1980s by a partnership with Monolithic Memories Inc. (MMI), which in 1987 signed a deal with Xilinx that gave MMI royalties and free patent rights to manufacture Xilinx’s products. In return, MMI supplied capital to Xilinx. The arrangement provided needed funding to bring Xilinx’s products to market and sustain its research and development initiatives. Soon after signing the deal, though, MMI was purchased by American Micro Devices, Inc. Xilinx became uncomfortable with the new arrangement, partly because American Micro was one of its competitors. So, in 1989 Vonderschmitt convinced American Micro to buy 20 percent of the company at a ten percent premium and dissolve the original agreement. In search of new funding, Xilinx went public with a stock offering on the NASDAQ over-the-counter exchange.

Although Xilinx rapidly increased sales and profits in the late 1980s and early 1990s, it also ceded much of its market share. Indeed, after Xilinx had invented its niche and controlled 100 percent of the FPGA market during the mid-1990s, other companies began offering competing technology that rapidly eroded Xilinx’s dominance. By the early 1990s, in fact, Xilinx was controlling only about 65 percent of the total FPGA market. It had succeeded, though, in pressuring many of its earliest competitors out of the business, despite the fact that some of them had access to much more funding. By 1993 only a few companies were seriously vying for market share. The largest was Actel, which had introduced its first product in 1988 and by 1993 was serving about 18 percent of the market. The distant third-place contender was Altera, which sold technology similar to FPGA systems.

Company Perspectives

Xilinx’s success can be attributed to one fundamental reason: we help our customers get to volume production faster. In today’s competitive technology marketplace, it is not enough to bring products to market rapidly. Time-to-volume production has become the new imperative for technology leaders.

Despite loss of market share, high demand allowed Xilinx to boost revenues to $135 million in 1991 (fiscal year ended March 30, 1992) and then to $178 million in 1992. Part of that growth resulted from Xilinx’s jump into the market for EPLDs (EPROM technology-based complex Programmable Logic Devices), which effectively offered higher performance and higher density per chip, and were designed to complement FPGA devices. By 1993 Xilinx was capturing more than $250 million in annual revenue and generating net income of $41.3 million. In addition, market growth for FPGA chips was expected to intensify in the mid-1990s.

To take advantage of market expansion, Xilinx introduced a completely new line of FPGA products in 1994: the XC5000 family of FPGA chips and software. The XC5000 line was developed to cater to the market for low-end gate array products. Specifically, XC5000 chips were designed to offer a cost-effective alternative to high-volume non-field-programmable gate array products, thus giving Xilinx access to a new spectrum of the market for application-specific circuits. Xilinx followed the introduction of the XC5000 family with other new products, including the XC3100L and XC4000L. Both new families of FPGA chips were designed to complement low-power applications, which boosted Xilinx’s access to manufacturers of increasingly popular low-power devices like portable computers and related peripheral devices, portable and wireless communication gear, and digital cameras.

New products and healthy demand growth for existing FPGA chips helped Xilinx to boost its sales to $335 million in 1994, a record $59.28 million of which was netted as income. Likewise, revenues rose to about $550 million in 1995 (fiscal year ended March 30, 1996). By that time, Xilinx was employing more than 1,000 workers in offices throughout North America, Asia, and Europe. The company was selling more than 40 varieties of programmable logic products and related software applications and continued to be firmly entrenched as the leader in its industry niche.

Early in 1996 Vonderschmitt stepped down as chief executive of the company he had cofounded 11 years earlier. Vonderschmitt remained as chairman of the board, but handed control of the company to Willem P. “Wim” Roelandts. The 51-year-old Roelandts had worked for 28 years at Hewlett-Packard Co., serving as senior vice-president and managing the company’s Computer Systems Organization, among other posts. He was known as a seasoned high-tech industry veteran and was chosen to lead Xilinx into a “new era of growth,” according to Vonderschmitt.