"Sharing God's Love with the Chinese"

In August 2000, on a flat parcel of reclaimed delta soil in Pudong, fifteen kilometers east of the Bund, a man in his early fifties stood with a ceremonial spade in his hands and a small group of Shanghai officials at his shoulder. The site was almost featureless. The municipality had bulldozed the rice paddies a year before, run a few service roads through the mud, and named the resulting grid the Zhangjiang Hi-Tech Park. There was nothing else to see. The man with the spade was Richard Chang Ru-jing, born Zhang Rujing in Nanjing in 1948 and very nearly born into the Chinese civil war. He had been on the island of Taiwan since he was an infant, in the United States for most of his adult life, and back in Asia for three years. Four months earlier, on April 3, he had incorporated a Cayman Islands holding company called Semiconductor Manufacturing International Corporation. He had with him roughly $1.48 billion in committed equity, a list of second-hand wafer-fab equipment he intended to buy from the rubble of Asia’s last memory-chip recession, and a personal conviction, which he had stated openly to anyone who asked, that God had brought him here.

The conviction was unusual. In a 2002 interview with BusinessWeek, Chang would put it plainly: the Lord, he said, wanted him to come to China and share God’s love with the Chinese people. He was an evangelical Christian, the kind who tithed and prayed before meetings and quoted scripture without embarrassment. He believed that the chip industry he was about to build in Pudong, the school he would open the following September for the children of his engineers, and the church he intended to seed in the same district were three legs of a single project. The phrase his colleagues used, in private and eventually in print, was a three-fold vision: technological, educational, spiritual. None of the Shanghai officials at the groundbreaking objected, at least not audibly. They needed the fab.

He looked, that morning, like exactly what he was: a Texas Instruments lifer in a borrowed suit. Chang had spent twenty years at TI in Dallas. His first manager, when he joined in 1977, had been Jack Kilby, whose Nobel Prize for the integrated circuit was still two decades away but whose reputation inside the company was already mythical. Chang had moved up by being deployed: he ran the construction and ramp of fabs in Texas, then in Sherman, then in Italy, in Japan, in Singapore, and finally in Taiwan, where in the early 1990s he had supervised the build-out of TI’s Acer joint venture. He had done it ten times. By the late 1990s he was, by his own count and his colleagues’, one of the few people on earth who had personally taken ten green-fields from groundbreaking to volume yield. Inside TI he was respected. Outside it, almost no one had heard of him.

He left TI in 1997, returned to Hsinchu, and founded a foundry of his own. He called it the World Wide Semiconductor Manufacturing Corporation, WSMC, in a fit of ambition that turned out to be mistimed. Hsinchu in 1997 already had Morris Chang’s Taiwan Semiconductor Manufacturing Company, a decade old, profitable, dominant. It had United Microelectronics, the runner-up. It had no obvious appetite for a third foundry built by a TI exile with no local capital base. WSMC limped through three years, completed a single fab, started a second, and in January 2000 was acquired by TSMC for roughly $550 million in stock. The deal was a foundry consolidation; for Richard Chang, it was something closer to a mugging. He had not wanted to sell. The TSMC offer, made in the language of stock-for-stock fairness, gave him no plausible way to refuse without burning his investors. He took the paper and walked out. He was fifty-two years old, abruptly liquid, and unemployed.

What followed, in his telling and his colleagues’, was a kind of resolution. A man who had built ten fabs and lost his own would build the next one in the country he had been born in and never lived in. The mainland’s semiconductor industry in 2000 was a graveyard of state attempts. Project 908, launched in 1990 to put China on the integrated-circuit map, had taken eight years to produce a Wuxi fab whose process was a generation behind by the day it opened. Project 909, started in 1996, had paired the state-owned Huahong with NEC to assemble Japanese kits in Shanghai, a transfer that produced chips but no transferable knowledge. Beijing had spent something on the order of two billion dollars across the two projects and had nothing to show for it that an actual customer wanted. The lesson the Chinese leadership took from the failures was administrative rather than industrial: the next attempt would have to be run by an industry insider, not a ministry, and the industry insider would have to be someone with American capital and Taiwanese know-how. The Shanghai municipal government, under party secretary Huang Ju and his successor Chen Liangyu, had spent two years quietly looking for the right insider. Chang, suddenly available, fit every requirement they had.

The deal Shanghai offered him would have been unimaginable in any market economy. The municipality contributed land in Zhangjiang at terms no foreign investor could have replicated, exempted SMIC from the corporate income tax for five years and halved it for the next five, gave the company a fifteen-year ride on import duties for capital equipment, and arranged an electricity priority that, when the Yangtze Delta began its rolling blackouts in 2003, kept SMIC’s wafers moving while textile mills upstream went dark. Local press releases described these provisions as policy. They were better understood as a single bilateral contract: the city would carry the capital cost of an industry, and in return it would have an industry. By the time Chang turned the first spade of earth in August 2000 the contract was already in force. He spent the first nine months of SMIC’s existence on airplanes, raising the rest of the round, signing equipment orders, and recruiting.

The recruiting was where the Christianity came in, although a casual observer would have called it ordinary engineering management. Chang did not hire missionaries. He hired the Chinese diaspora. Of the company’s first thousand-odd employees, only about four hundred had grown up on the mainland; the rest were Taiwanese, Singaporean, Malaysian Chinese, and overseas-Chinese American engineers, many of them following Chang from the TI alumni network and the WSMC roster. They moved to Shanghai with families, and the families needed schools, and the schools that existed in Pudong in 2000 were not the kind that would teach in English to the children of returnees who had left Taiwan or California two months earlier. So in September 2001, twelve months after the company itself had been incorporated, SMIC opened a school. It used Western curricula, hired teachers Chang’s network helped recruit, and admitted first the children of SMIC engineers and later anyone who could pay tuition. It would graduate its first senior class in 2006. By 2015, fewer than one in five of its students would have a parent at the company. The school had become a Pudong institution, with a recognizable culture that its neighbors and competitors sometimes called Christian and that the school itself, in its public materials, described in milder language.

The third leg arrived later and quieter. In the early years the company hosted Sunday services in a hotel conference room near the Beijing fab; about two hundred employees attended, hymns projected on a screen, Bibles distributed, sermons given by a state-approved minister whose presence the local religious affairs bureau had cleared in advance. A SMIC spokesman would tell the Christian Science Monitor in 2008 that the founding CEO had made the church an issue and the government had accommodated him “again and again.” In Pudong the arrangement evolved into what was eventually known as the Thanksgiving Church and, in its English-speaking ministry, the Thanksgiving English Fellowship; by the late 2000s, hundreds of SMIC employees and their families were attending. Chang did not pretend the church was incidental to the business. He had said, on the record, that it was the third leg of why he had come.

What the church was not, was a competitive advantage. SMIC’s competitive position, through its first years of operation, was almost entirely a function of the engineers Chang had brought with him from Hsinchu. He had hired more than a hundred people who had previously worked at TSMC. Some had retired and been re-hired; others had been recruited away. Most had signed nondisclosure agreements on their way out of TSMC and had then, in many cases, taken with them what TSMC’s lawyers would later describe as more than fifteen thousand documents and process recipes. SMIC’s own engineers came to refer to TSMC’s most valuable production knowledge by an internal codename, BKM1, for Best Known Method One. SMIC’s first 0.18-micron process, the workhorse it would use to win foundry contracts from Broadcom and others, ramped to volume in something like six months from equipment install. The industry’s normal benchmark for a comparable ramp was eighteen to twenty-one months. The compression was not because Pudong’s engineers were better than Hsinchu’s. It was because they were, in a meaningful operational sense, the same engineers, working from the same notes.

TSMC noticed. In December 2003, Morris Chang’s company filed suit in the U.S. District Court for the Northern District of California, alleging patent infringement on five of its process patents and trade secret misappropriation on a sweeping list of technologies. The complaint named SMIC as the defendant, listed by name a number of former TSMC employees who had moved to SMIC, and alleged that those employees had brought TSMC’s proprietary processes with them. SMIC at first denied everything, then, fourteen months later, settled. The terms announced on January 31, 2005, were a one-hundred-and-seventy-five-million-dollar payment to TSMC over six years, thirty million annually for five years and twenty-five in the sixth, plus a six-year cross-license of the two companies’ patent portfolios through the end of 2010. It was a substantial number, but not a ruinous one for a company that had just completed an initial public offering. SMIC’s American depositary shares had begun trading on the New York Stock Exchange under the ticker SMI on March 18, 2004, and its ordinary shares on the Hong Kong exchange the same day under stock code 981, raising about $1.8 billion in the third-largest IPO of the year. The settlement, in the press releases of both companies, was described as resolving the dispute. In Hsinchu, internally, it was described as a ceasefire.

It did not last. As part of the settlement, SMIC had agreed to allow TSMC’s investigators to verify that it had purged its processes of the borrowed material. When TSMC’s auditors began the work, they found, by their own account, that the contamination ran far deeper than the original suit had alleged: not five patents and a handful of processes but, ultimately, thousands of pages of recipes copied verbatim, “word for word, line for line, diagram for diagram,” in the language of the second complaint. In August 2006, TSMC filed again, this time in the Alameda County Superior Court in California, framing the new action as a breach of the 2005 settlement and asking for additional damages. The case ground on for three years. In 2009, a jury found SMIC liable on sixty-one of sixty-five claims and assessed roughly a billion dollars in damages. The verdict, if it stood, would have been the end of SMIC. It would also have damaged TSMC, by stranding a tier-one Asian foundry whose customers were already integrated into the global supply chain.

Both companies negotiated through the autumn. The settlement that emerged on November 9, 2009 was, in its way, more revealing than the verdict had been. SMIC would pay TSMC two hundred million dollars in cash. It would also issue TSMC roughly 1.79 billion new shares, plus warrants for nearly seven hundred million more, a stake that, once exercised, gave TSMC about ten percent of the equity in its Chinese rival. The 2005 cross-license was terminated. And in a separate, simultaneous announcement, Richard Chang resigned. He was sixty-one. He told reporters he was leaving to pursue other interests. The board named David Wang, a former Huahong NEC chairman, to replace him as CEO. Chang accepted, as part of the personal terms of the settlement, a three-year prohibition on operating in the chip industry. He went home. SMIC’s American depositary shares, which had touched twenty-three dollars at the IPO peak in 2004 and fallen by something like ninety percent in the years since, were, by the time of his resignation, suspended from trading.

The departure was a kind of summary of what SMIC had become. Chang had founded the company as an act of belief: belief that an outsider with the right industrial knowledge could build, in a year and a half, what Beijing’s ministries had failed to build in a decade, and belief that the Lord had given him the assignment. He had been correct, in a narrow operational sense. SMIC by 2009 was the largest foundry in mainland China, a real customer-facing business with real wafers shipping to real fabless designers in Sunnyvale and Taipei and Tel Aviv. It had done in nine years what the 908 and 909 projects had failed to do in fifteen. But it had done it on borrowed knowledge, and when the lender called the loan, the company had no choice but to pay in equity and in its founder.

The structural limits the affair exposed were not Chang’s alone. SMIC, throughout the 2000s and into the 2010s, would consistently sit two and sometimes three process generations behind TSMC. When TSMC was producing 65-nanometer chips in volume, SMIC was at 130. When TSMC moved to 28, SMIC was reaching 65. The gap was not principally an equipment problem; both companies could buy ASML scanners and Applied Materials etchers and KLA inspection tools out of the same catalogues. Nor was it a capital problem; Beijing, before the decade was out, would launch the National Integrated Circuit Industry Investment Fund, the so-called Big Fund, with capitalization in the tens of billions of dollars, of which SMIC would receive a generous share. The gap was something subtler. TSMC’s edge was not its equipment or its capital; it was the accumulated discipline of a manufacturing organization that had been writing down its own process recipes, in its own language, for twenty-two years, and that had built around itself a network of suppliers, design houses, and customers who together formed an ecosystem none of its competitors could replicate. SMIC had bought, or borrowed, the recipes. It had not bought, and could not borrow, the ecosystem.

This was what made the settlement so quietly damaging. The two-hundred-million-dollar payment was an irritant. The ten-percent equity stake was, in normal industrial-policy terms, a humiliation. But the deeper consequence was that Beijing now understood, in a way that the failures of the 908 and 909 projects had not quite proved, that capability could not be smuggled across the strait in the briefcases of departing engineers. It could be acquired only by reproduction. And reproduction, at the leading edge of semiconductor manufacturing, took longer than the political cycle of any single Chinese leader.

Chang stayed in Shanghai. He honored the three-year ban. In 2014, on its expiration, he founded Shanghai Xinsheng Semiconductor, a 300-millimeter wafer manufacturer; in 2018, he started SiEn Integrated Circuits in Qingdao, where he later accepted the title of lifetime honorary dean at Qingdao University. He continued, in his later years, to speak openly about his faith. When Dallas Baptist University awarded him an honorary doctorate in 2023, the citation described him as “a committed Christian servant leader who demonstrates active love and service to the body of Christ across East Asia.” The Thanksgiving Church and its English-language fellowship in Pudong, by then operating independently of SMIC, continued to draw a few hundred congregants on Sunday mornings: doctors, engineers, returnees from California and Taipei, a slowly shifting expatriate roll. SMIC itself continued to grow, continued to lag, continued to receive state support, and continued to insist, with progressively more state backing and progressively less success, that the gap to TSMC could be closed.

Across the strait, Morris Chang, when asked in interviews how he viewed Richard Chang’s project, would generally answer with the kind of careful courtesy two TI veterans of the same generation owe each other in public. He had, after all, sued Richard Chang and won. He had also taken ten percent of his rival’s company in lieu of damages, an unusual remedy that left TSMC, formally, as a part-owner of the foundry it had been suing. The arrangement was eventually unwound; TSMC sold the SMIC shares over the following years, taking a profit on the way out. But in 2009, on the day of the settlement, the structure was extraordinary. The world’s leading foundry now owned a slice of its largest mainland competitor, by court order in everything but name.

What neither Chang knew in 2009, and what no one watching the settlement could have known, was that the question SMIC was failing to answer would, within a decade, become the central question of global technology policy. China’s first real foundry was a real foundry. It was not, and on the basis of its first decade could not become, a leading-edge one. The reasons for the gap, the long discipline, the deep ecosystem, the cumulative learning that could not be hired or copied, were the reasons that the Pentagon would, by the late 2010s, be willing to bet a generation of strategic policy on the assumption that the gap would hold. They were also the reasons that Beijing would be willing to bet a generation of industrial policy on the assumption that it would not.

For now, the man who had begun it walked away. He had built a school, he had seeded a church, he had founded a foundry, and he had lost a court case that took the foundry from him. The three-fold vision was intact. Two of the three legs, his colleagues would later say, had outlasted him.