The first thing most people do when they turn on their computers is go online to check their email and browse their favorite web sites. Then, as they switch on their messenger service, they wait to hear the familiar melody announcing incoming messages from colleagues and friends.

To connect to the Internet, half the world’s computers rely on integrated chips emblazoned with the red crab logo of Taiwan ’s Realtek Semiconductor Corporation. Two-thirds of the world’s computers rely on Realtek ICs for their audio capabilities.

Realtek’s crimson crab has clamped down with a tight pinch on the world’s second largest IC design house, Broadcom, and its fifth largest, Marvell. Together, they form the triad of leaders at the top of the market for Ethernet controllers.

“Once Realtek’s products came out, its international competitors Broadcom and Marvell slashed prices themselves in order to compete,” says one chip fab executive with a broad swipe of his hand. In 2003, Broadcom set up an R&D center in Taiwan , taking advantage of lower personnel costs for product development in order to contend with Realtek.

Realtek’s NT$13 billion in revenues last year placed it sixth among Taiwan ’s IC design firms. Thanks to its solid foundation in technology, the crab has managed to flip itself upright and crawl out of its slump of five years ago. It has even secured the third position in the communications networks sector.

Built from Scratch

Realtek is an expert when it comes to technology. A few years ago, it slapped fellow Taiwanese IC design firm MStar with a lawsuit claiming that this competitor had infringed on its patents for LCD TV display chips. A professional analysis of the two companies’ patent profiles led to the amazing discovery that one of the names on the list happened to be that of an engineer who had been working at Realtek for almost two years. This bears witness to just how serious Realtek is about technology R&D.

In terms of technology, Goldman Sachs executive director Joey Cheng, who has long been in charge of sales on the front lines of the semiconductor industry, defines Realtek by asserting, “You could say Realtek is the MediaTek of communications networks.”

Realtek developed its first Ethernet controllers in 1993. At that time, the young company depended on the tireless work of one engineer who, though he was a graduate of the National Taiwan University Department of Physics and held a master’s degree from National Chiao Tung University’s Graduate School of Electrical Engineering, possessed absolutely no experience in the world of business. He would read two research articles in publications released by the Institute of Electrical and Electronics Engineers every week, and went over circuit diagrams endlessly, until he managed step-by-step to piece together from scratch the first Ethernet controller to be marked with Realtek’s red crab.

Lee Chao-cheng, Realtek’s chief technology officer, was a fellow student of Cheng’s during the time the Goldman Sachs executive studied in the NTU Department of Electrical Engineering. Cheng describes his classmate as a person of genius. People would sometimes feel odd when Lee spoke, because his brain was so sharp that his train of thought would outrun his mouth.

Sound technology has given Realtek the ability to clamp down on the leading international IC design firms. “Our strategy is to be a fast follower,” says Realtek president Alex Chiu, who registers a pair of dimples when he smiles.

According to Chiu, product cycles have sped up. In the past, Realtek could release the same product one year later than companies in the United States and still make a profit, but today the firm must catch up in half a year’s time. Realtek now must work constantly at upgrading its research and development capabilities and must achieve integration and innovation at many levels if it wants to turn a profit.

A decade ago, Realtek’s main 10/100 Ethernet controllers drew a price of US$8 per chip and brought a profit margin of 35 percent. These days, while one of these chips costs just one US dollar, Realtek claims nearly 50 percent of this as profit. It has been able to raise its profit margins by making its chips more energy efficient and smaller, or by adding more integrated functions.

Over the previous two years, Realtek’s strenuous efforts have begun to pay off, as their new communications network products – fast Ethernet controllers, wireless LAN ICs, network switch controllers and ADSL router controllers – have one after the other hit the market on the heels of similar chips from the big design houses.

Nevertheless, the take of most industry insiders is that Realtek as a firm is “strong on R&D, but not strong on management.” There was a period at Realtek when no one wanted to be president. The company then introduced a “revolving system” under which one person would assume the president’s position for a two-year term before giving it up to the next person in line.

This weakness on management has a great deal to do with the history of the company’s rise to fame. It starts back in 1987 with six engineers from the IC design division at United Microelectronics Corporation who, having worked for not even five years, were all still short of thirty. All graduates of the departments of electrical engineering at National Taiwan University , National Tsing Hua University and National Chiao Tung University , the six began calling up their friends and contacts and decided to set up their own company. Realtek was a whole decade ahead of most of the other companies, such as MediaTek and Novatek, that have also sprung from UMC’s IC design division.

Not many Taiwanese IC design companies existed in those days. Nor were there many competitors. These six young engineers believed that, by relying on their own technological knowledge, they had a chance to overrun the world like crabs on a beach. Hence, they singled out the red crab as the symbol of their new enterprise, Realtek Semiconductor Corporation.

A Nearly Fatal NT$10-billion Lesson

Looking back, Realtek vice president Chen Chin-hsing recalls how everyone at Realtek had immersed himself in research work in the early days. Some were so wrapped up in their work they forgot to return home on Chinese New Year’s Eve to dine with their families.

Then, beginning in 2002, Realtek’s “not strong on management” culture was forced to contend with a nearly fatal challenge.

Realtek had recently lifted its annual revenues past NT$10 billion for the first time and stood as the third largest IC design operation in Taiwan (behind VIA Technologies Incorporated and MediaTek). Then, caught unaware, it watched its average annual revenue growth rate stagnate at 5.2 percent for the next three years. Up to this point, its shareholders meetings had been regularly packed with nearly 200 shareholders. By 2004, the number of attendees had plummeted to just 50, leaving the company parking lot lying empty. On top of this, Realtek’s annual rate of resignations jumped from its previous five percent to ten percent.

“We were stuck defending a perilous NT$10-billion position,” remembers Chiu.

By 2001, Realtek had engaged its engineers in the development of a broad spectrum of product lines. Too broad, it turns out. This approach scattered the firm’s resources, leaving it accomplishing practically nothing at all. “Our ambition was too great,” Chiu concedes. Realtek had stretched its battle line wide and thin. It was working on DVDs, LCD television controllers, wireless LAN ICs, network switch controllers and ADSL router controllers, all at the same time. As a result, it failed to release a single product on time.

Although room remained for growth in each product line, the new products being added to each line multiplied exponentially. The breadth of management required to cover this extensive range of products stretched the company past its limits. Adding insult to injury, the major international players set about slashing their prices in an attempt to grab orders, and Realtek’s revenues stalled.

“ Deciding not do something was harder than deciding what to do ,” says Chiu. Five years down the road, what Realtek has learned is “position, focus, and self-restraint.”

Realtek abandoned its DVD efforts in 2005 and proceeded to get its feet back on the ground. In the end, it succeeded in the development of its fast Ethernet controllers, wireless LAN ICs, network switch controllers and ADSL router controllers, proving itself a formidable adversary, and a real headache, for such world leaders as Broadcom and Marvell.

“It’s fortunate that Realtek’s technology is strong. Its foundation remained intact, and that’s why they could hold on for three years,” declares Joey Cheng.

Having learned its NT$10-billion lesson, the tenacious Realtek is back on firm ground. This tough crustacean, hardened over the course of two decades, is intent on maintaining its tight clamp on the world’s leading IC design firms.

Translated from the Chinese by Stan Blewett

---

Chinese Version: 紅蟹緊咬　逼對方砍價求生