"My Enemy's Enemy" — The Rise of Korea

On the evening of February 8, 1983, Lee Byung-chul picked up the telephone in his suite at the Hotel Okura in Tokyo and placed a call across the East Sea to Seoul. The Okura sat across the road from the American embassy in Toranomon, a sober mid-rise hotel that Japanese executives favored because the lobby was quiet and the meeting rooms were reliably anonymous. Lee, then seventy-two and recovering from cancer surgery the previous year, had spent the week walking the corridors of MITI’s preferred meeting rooms and the lobbies of NEC, Hitachi, and Toshiba, watching how a Japanese chip industry organized itself. He had asked to see fabrication lines and been politely refused. He had asked for technical briefings and been given marketing brochures. The visit had been, on the surface, a courtesy tour. Lee had concluded, somewhere between the third and fourth Japanese factory the engineers had declined to show him, that Samsung had to make memory chips, that it had to do so before the Japanese became unbeatable, and that his board of directors would either follow him into the decision or watch him make it alone.

The call was to Hong Jin-ki, the chairman of Samsung’s affiliated newspaper, the JoongAng Ilbo. In Korean business culture of 1983, the JoongAng Ilbo was not a neutral organ; it was the family’s mouthpiece. Lee told Hong, in language his biographers later transcribed almost identically, that no matter what anyone else thought, Samsung was going to enter the semiconductor business, and that Hong should report it to the public the next morning. The line ran to one paragraph. Samsung’s executives in Seoul were not yet consulted. The board had not voted. The capital had not been allocated. By the time the morning edition reached Korean breakfast tables on February 9, the announcement that would later be called the Tokyo Declaration, known in Korean shorthand as the Two-Eight, after the date, had become the kind of fact that its target audience could not unsee.

Lee was committing Samsung Group to a business in which it had failed once already, and which most of his American and Japanese counterparts considered too capital-intensive for a Korean conglomerate to enter. He would later write, in language that has been quoted almost as often as the original announcement, that he had thought more than half the group might be blown away if he got the decision wrong, and that no one but Samsung had the nerve to take the risk. The note of bravado concealed a more careful calculation. Lee had concluded that the Japanese had won the consumer-electronics era and would not yield it, and that the chip itself, the small piece of doped silicon that sat inside everything Japan was making, was the next era’s leverage. The decisive piece of his reasoning was the geopolitics. He had calculated that the Americans, consumed with their anger at Tokyo, would tolerate a Korean chip producer in a way they were no longer willing to tolerate a Japanese one. The first two judgments would prove correct on technical and economic grounds. The third would shape the next decade in ways even Lee did not foresee.

To understand how he had arrived at the suite at the Okura, one had to go back nine years, to a poor Bucheon factory south of Seoul where a small company called Korea Semiconductor was preparing to file for bankruptcy. Korea Semiconductor was a subsidiary of an American venture called ICII, set up in 1974 by Kang Ki-dong, an ethnic-Korean engineer who had spent the late 1960s at Motorola in Phoenix and then tried to build a watch-chip business out of Sunnyvale. The Bucheon plant had been bankrolled to make the cheap CMOS integrated circuits that powered the LED wristwatches that were briefly everywhere in the early 1970s, and the wristwatch fad had collapsed faster than the factory could produce. By late 1974, Korea Semiconductor was hemorrhaging cash and Kang was casting around the chaebol families for a buyer.

Lee Byung-chul was not interested. The Samsung patriarch had built his fortune on sugar refining, then textiles, then department stores and insurance, and he regarded the chip business as a financial sinkhole that even Western firms struggled to make money in. His third son, Lee Kun-hee, then thirty-two and serving as a director of the family’s Tongyang Broadcasting Company, disagreed. The younger Lee had spent part of his graduate education at George Washington University in Washington, D.C., and at Tokyo’s Waseda before that, and he had absorbed the conviction, common among Korean technocrats of the postwar generation, that Korea would remain what he called a colony in technology unless it learned to make its own. In a move that Korean business journalism would replay for decades, the son went around the father. Lee Kun-hee personally bought a fifty-percent stake in Korea Semiconductor with his own money, a sum most often cited as fifty thousand U.S. dollars drawn from his personal funds rather than the group’s treasury, and put the failing watch-chip plant in Bucheon under Samsung Group’s umbrella, daring his father to either bless the move or repudiate his son in public. The patriarch, characteristically, did neither, and let the company drift on its own books for the rest of the decade. Samsung would not buy out the remaining fifty percent of Korea Semiconductor until 1980.

The Bucheon plant did not become a chip business in the years that followed. It made the kind of products that small Korean factories made in the late 1970s: linear integrated circuits for televisions, custom logic for white goods, mostly under license from Japanese partners who treated their Korean buyers the way Detroit had once treated its Mexican subsidiaries. What it did was give Samsung a building, a few hundred trained workers, and the experience of running a fab in a country that had never before run one. None of those assets, individually, was decisive. Together they were the difference between a company that had to start a chip business from scratch in 1983 and one that had a ten-year-old running line to retrofit.

The other condition that mattered had been put in place even earlier, by Park Chung-hee, the general who had seized power in May 1961 and ruled South Korea through assassination in 1979. Park’s regime is remembered, fairly, as authoritarian and at times brutal, but its industrial policy is the other half of the legacy. In 1966 Park established the Korea Institute of Science and Technology, KIST, modeled on Battelle Memorial in Ohio and staffed by the first wave of Korean PhDs that the regime had begun aggressively recruiting back from American universities. In 1971 he authorized the Korea Advanced Institute of Science, later renamed KAIST, on a blueprint commissioned from Frederick Terman, the Stanford dean who had helped invent Silicon Valley. In January 1973, Park announced what he called the Heavy and Chemical Industry drive, an explicit national push into steel, shipbuilding, machinery, autos, and electronics, with the chaebol, the family conglomerates Park had alternately threatened and rewarded since his coup, as the chosen instruments. Most of the chaebol Park empowered are still recognizable: Hyundai, Daewoo, Lucky-Goldstar, Sunkyong, Samsung. The conglomerates were given preferential bank credit, foreign-exchange allocations, and effective state guarantees against bankruptcy in exchange for executing the regime’s industrial priorities. It was the same patient-capital model Japan had run a generation earlier, transposed into a country one-third Japan’s size and with a fraction of its technical base. Korean economists writing in the 1990s would call it a developmental state. Lee Byung-chul, at the Okura in February 1983, was its most successful product.

By the time Lee placed the call to Hong Jin-ki, the technical balance had shifted in a way that gave Samsung an opening the chaebol model could exploit. The Japanese had broken the American DRAM industry over the previous five years, as the Anderson Bombshell in March 1980 and the collapse of Mostek had made plain. American firms were scrambling to exit memory or to find some way to survive in it. One of those firms was Micron Technology, a small Idaho company funded by the potato baron Jack Simplot, which had made enough of a 64K DRAM design to have something to license but not enough cash to scale it. In 1982 a Samsung delegation, working through Korea Semiconductor’s old Sunnyvale connections, arrived at Micron’s Boise headquarters and offered cash for the design. Micron’s leadership, calculating that an Asian producer making chips on a Micron architecture would expand the architecture’s footprint without directly competing with Micron in U.S. customers, took the deal. The price reported in subsequent Korean accounts was a few million dollars, modest by the standards of any later licensing transaction, but the Boise team was happy to take it. Micron got the cash to build its next fab. Samsung got a working memory design.

The deal looked transactional. It was, in retrospect, the first stitch in the alliance that would define Korean memory for the next twenty years. The Japanese, by 1983, were licensing nothing of consequence to anyone. American firms, looking down the barrel of bankruptcy, were willing to license almost anything to a foreign producer who paid in dollars. Sharp signed a parallel deal with Samsung the same year, providing the SRAM and ROM technology Samsung’s group affiliates needed for consumer products. Both contracts were, on paper, ordinary commercial transactions. In aggregate they amounted to Samsung receiving, from two opposing sides of an industry that was about to enter a trade war, the technical inputs to enter that war as a third party.

Samsung complemented the licensing with a Korean engineering corps assembled out of two streams. The first stream came from KAIST and Seoul National University, where a generation of engineers trained in Kim Choong-Ki’s semiconductor program had been waiting for a domestic employer that could use them. Kim, who had returned from Fairchild and Columbia in 1975 to teach at what was then KAIS, would later be described as the godfather of Korean semiconductors; his students included men like Chin Dae-Je, who would lead Samsung’s 16M DRAM program in the late 1980s, and Kwon Oh-Hyun, who would later run the company. The second stream came from Silicon Valley itself. In 1983, simultaneously with the Tokyo Declaration, Samsung incorporated DSA, Samsung Semiconductor America, in San Jose, and began recruiting Korean-American engineers out of Intel, Fairchild, and National. The pitch was domestic: come home. Build a Korean industry. Be a founder rather than a middle manager. The pitch worked because, on either side of the Pacific, the people who heard it had grown up in a country that had been a Japanese colony from 1910 to 1945, and the prospect of out-engineering the Japanese in their own decisive industry resonated at a frequency that no Silicon Valley salary could match.

The Giheung fab, broken ground south of Seoul in March 1983 and reportedly brought online in the second half of 1984, was the visible artifact of all of it. Korean accounts emphasize that the construction was completed in roughly six months, half the time a comparable Japanese or American line would have taken; the foreign chip-industry press greeted that claim, when they noticed it at all, with the polite skepticism due to chaebol publicity. What the line produced, however, was harder to dispute. Samsung delivered its first 64K DRAM samples in late 1983, scarcely ten months after the Tokyo Declaration. The 256K followed in October 1984, fabbed off the Micron design. By 1986 Samsung had moved to the 1Mb generation, by 1988 to the 4Mb. The technological gap with Japan, which had stood at more than ten years when Lee Byung-chul flew to Tokyo, had compressed inside a half-decade to something like two.

The compression was paid for in losses. Between 1984 and 1986 the global DRAM market collapsed under the same Japanese price-attrition strategy that had finished off Mostek and pushed Intel out of memory. The 64K chip, which had sold above three dollars when Samsung’s first parts came off the Giheung line, was selling under fifty cents by mid-1985. Samsung’s per-chip production cost was higher than the price. The company’s accumulated losses on memory by the end of 1986 ran past three hundred million dollars, a figure that exceeded its semiconductor equity. By any straightforward accounting, Samsung Semiconductor was insolvent. By the unstraightforward accounting of the chaebol model, Samsung Semiconductor was a line item inside a conglomerate that also sold ships, sugar, fertilizer, life insurance, and televisions, and the rest of the conglomerate kept the chip line alive on internal cash flow while the founder waited for the cycle to turn. The Korean state contributed too: through preferential bank lending arranged by the Ministry of Trade and Industry, and through a national 4M DRAM consortium initiated in 1986 by the Electronics and Telecommunications Research Institute, ETRI, that pulled Samsung, Hyundai, and Goldstar into a shared program along the lines MITI had pioneered for Japan a decade earlier.

What turned the cycle for Samsung was not, in the end, anything Korean. It was an agreement signed in Washington in September 1986 between the United States and Japan, intended to do something else entirely. The US-Japan Semiconductor Trade Agreement put a floor under Japanese DRAM prices in the American market, monitored Japanese export pricing to third countries, and raised the cost to Japanese firms of the dumping strategy that had nearly finished the American memory industry. It was negotiated in Washington and Tokyo by people who barely thought about Korea. The Reagan administration’s trade representatives wanted to discipline Japan; the Japanese negotiators wanted to limit the discipline; neither side was in the room for Korea. But the agreement set in motion a price recovery in DRAM that began in the second half of 1987 and accelerated through 1988, and the immediate beneficiaries of that recovery were the producers who could ship chips into the United States without a price floor on their backs. That was Micron. And it was Samsung.

Samsung’s 1Mb DRAM came to market in 1986 into a market that, eighteen months later, was paying for it. By 1988, with the price floors holding the Japanese above their own marginal cost and the Korean producer free to undercut them, Samsung’s memory line was profitable for the first time. The lost three hundred million was earned back in eighteen months. By 1992 Samsung had introduced the world’s first 64M DRAM and overtaken Toshiba for the global memory crown, a leading position no Japanese, American, or European firm would dislodge in the next thirty years.

The American attitude through this rise had two faces, and historians arguing about how to characterize the period are still arguing about which face was the real one. The face the Korean firms saw was tolerant. Lee Byung-chul died in November 1987, before the trade picture had fully resolved, but during his last years his American interlocutors had been unfailingly cordial. Bob Noyce of Intel, by then chairman of the Semiconductor Industry Association and a public hawk on Japan, made no effort to extend the SIA’s Section 301 case to include Korea. Charles Sporck of National Semiconductor, the loudest voice in the trade-war lobby, did not call for restraints on Korean DRAM. When the U.S. Commerce Department finally did open an antidumping case against Korean DRAM in 1992, at Micron’s request, the duties imposed were tiny: 0.22 percent against Samsung, single digits against the others. The case was procedurally identical to the dumping cases that had crushed Toshiba and Hitachi six years earlier, and procedurally unrecognizable in its outcome. American officials in private conversations of the period, recorded later by Korean industrial historians, described the Korean producers in language no one had ever used about the Japanese: not as adversaries but as second sources, useful counterweights, partners in keeping memory chips affordable for American computer makers.

The other face was less generous. The Americans had not extended the trade peace to Korea out of friendship; they had extended it because Korea was, in 1986, still a small enough player that going after it would have looked petty, and a useful enough one that disciplining it would have only handed market share back to Japan. Samsung’s rise was tolerated because the alternative was that Tokyo would benefit from any Korean misstep. The American calculation, never formally articulated, was the one Lee Byung-chul had counted on at the Okura: an enemy’s enemy is a customer’s preferred supplier. The relationship was strategic on the Korean side and convenient on the American side, and convenience, in the politics of trade, has at least the durability of strategy.

The man who would inherit the Korean side of that relationship was the same son who, thirteen years earlier, had bought Korea Semiconductor with his own money against his father’s wishes. Lee Kun-hee assumed the Samsung chairmanship in December 1987, weeks after his father’s death, at the age of forty-five. The next five years passed in a steady consolidation. Samsung became Korea’s largest chaebol; its DRAM business overtook NEC, then Toshiba; its profit margins, which had been negative for most of a decade, turned the corner the trade agreement had drawn for them. By 1993, Samsung was the world’s leading memory producer in volume terms. Lee Kun-hee had inherited the company in time to preside over the victory.

He chose, in the spring of 1993, to declare a crisis anyway. On June 7, in a function room at the Kempinski Hotel near the Frankfurt airport, Lee gathered roughly two hundred Samsung executives, division heads and branch managers and the senior international sales force, every leader the chairman had been able to summon to Germany on short notice, and delivered what would become the most quoted speech in Korean corporate history. He told them that the company was complacent, that its products were second-rate by international standards, that its sales force still thought in Korean while its customers thought in dollars and yen, and that nothing short of a complete reorganization of mindset would prevent Samsung from sliding back into the second tier the moment the trade agreement’s protective umbrella expired. The line that survived translation was characteristically blunt. Change everything except your wife and children, he said. The assembled executives took notes that would later be transcribed into the company’s New Management doctrine and circulated through every Samsung subsidiary on earth. The Frankfurt Declaration, as it came to be known, became the founding document of the version of Samsung that would later challenge Apple in phones, Sony in displays, and TSMC in foundry.

It was also a quietly ungrateful document. The chairman who had been handed a global memory crown by an American trade fight that Korea had not started was telling his executives that the position was already at risk. He was, on the evidence, correct. Within five years the Asian financial crisis would force the Hyundai-LG memory merger that became Hynix; within ten years the Japanese memory industry would have collapsed entirely, taking with it the rationale for American forbearance toward Korea; within twenty years Samsung itself would be the largest single chip producer that any American administration had to think about restraining. The trade architecture that had let Lee Byung-chul into the industry would not last.

In Tokyo on the night of February 8, 1983, none of that was knowable. Lee hung up the phone at the Okura, returned to Seoul, and faced his board the following week, watching them bend in the direction his telephone call had already established as inevitable. Two months later the manifesto Samsung published in the JoongAng Ilbo, titled in Korean as Why We Must Enter the Semiconductor Business, argued that Korea’s national survival required a domestic chip industry, that no one else could fund the entry, and that Samsung was the only Korean firm with the scale to attempt it. The argument carried inside Korea because the country was prepared to believe it. It carried abroad because, in the geometry of trade rivalries Washington and Tokyo had drawn around themselves in the spring of 1983, a third party useful to one side and not yet large enough to threaten either was something both sides were prepared, for different reasons, to indulge.

The American chip industry had spent the early 1980s asking what kind of foreign producer it could live with. The unspoken answer turned out to be: not a Japanese one. By the time Washington realized that the Korean producer it had quietly tolerated had inherited the throne the Japanese were vacating, the position was no longer available to be reassigned. Lee Byung-chul had read that calculation in a Tokyo hotel room on a winter evening in 1983, and had decided to act on it before anyone else did. The act was characteristic of the man: blunt, premature by every conventional measure, and right.