Supply Chain Statecraft

By the time Mark Shepherd flew into Taipei in the late 1960s, he had spent most of a decade flying somewhere. Texas Instruments had named him president in 1967 and would name him chief executive in 1969. More than any other executive, Shepherd had committed TI to the idea that the United States could not absorb the labor cost of assembling its own chips. He had already moved transistor lines to Taiwan’s neighbors. Japan first, then a leased shed on Singapore’s Kallang Basin where four thousand workers, most of them women in their twenties, fed lead frames through bonders on three shifts.

Taipei was the next stop. The man waiting for him there was Li Kwoh-ting, Taiwan’s Minister of Economic Affairs, a Cambridge-trained physicist who had once worked under Ernest Rutherford on superconductivity and had since become the chief salesman for an island state that, by every diplomatic measure, was running out of friends. Li went by his initials in English. His business card said K.T. Li. He had the unhurried courtesy of someone who had sat through worse rooms than this one.

The dinner that followed, or the series of meetings that surrounded it, cannot be reconstructed line by line. Morris Chang, then a TI vice president, has spoken publicly about that period only in glancing terms. In remarks at the American Chamber of Commerce in Taipei in 2023, he credited Li with personally clearing the obstacles that allowed TI to “take root” on the island, naming the two specific frictions that had stalled the negotiation: intellectual property rights, where Taiwanese law had no clear protection for an integrated-circuit design, and land use, where the bonded zones available to a foreign assembler were either too small or wrongly classified. Chang did not describe a handshake or a banquet. He described a minister who treated the deal as if it were his own.

That, by the standards of postwar industrial diplomacy, was the unusual part. American executives flying into developing countries in the 1960s were used to ministers who delegated, hedged, or extracted bribes. Li did the opposite. He treated the TI plant as a prestige project of the Republic of China, one whose terms he would personally guarantee against the inertia of his own bureaucracy. The Statute for the Encouragement of Investment, which the legislative yuan had passed in 1960, gave him the legal instruments: five-year tax holidays, accelerated depreciation, free repatriation of profits, exemptions on imported equipment. Of the statute’s thirty-five articles, twenty concerned taxation, and twelve specifically reduced or removed it. What Li offered in person was the political will to make those instruments actually work for a foreign firm in a hurry.

The plant TI built sat in Chung Ho, on the western flank of the Taipei basin, in what is now Zhonghe District of New Taipei City. It opened in August 1969. It assembled and tested integrated circuits; wafers arrived from Dallas, were sawn into individual die, bonded onto lead frames, encapsulated in plastic, tested, and shipped back across the Pacific to be sold by TI’s American sales force. The labor was nearly all female, nearly all young, and nearly all rural. In its broad outlines the operation looked indistinguishable from what Fairchild had set up in Hong Kong six years earlier, what General Instrument had begun in Taiwan in the mid-1960s, and what TI itself was running in Kallang. By 1968, electronics had become Taiwan’s second-largest export category, behind textiles, and was growing at fifty-eight percent a year.

What made the TI deal different was who signed it and what they thought they were signing.

For Shepherd and his board, the plant was an arithmetic decision. TI was making integrated circuits in volumes that, by the late 1960s, were beginning to be measured in hundreds of millions of units. Each unit had to be packaged. Packaging was painstaking, semi-skilled, and at American wage rates, ruinous. A single hour of an American assembler’s time, fully loaded, was running well above two dollars. The same hour in Taiwan, fully loaded, could be had for roughly nineteen cents. Every quarter that an integrated-circuit maker continued to assemble its products in Texas or California was a quarter of margin handed to whichever competitor went offshore first. Fairchild had moved early. TI was simply moving faster.

For Li and his colleagues in the Executive Yuan, the same plant was something larger. It was a piece of statecraft. Taiwan in 1969 was a country whose government had been expelled from the United Nations Security Council in everything but name. In two years Beijing would take the seat outright. The Mutual Defense Treaty with Washington still held, but the Nixon administration’s overtures to the People’s Republic were already underway. The KMT understood, with the clarity of a regime that had watched one continent slip away, that diplomatic recognition was a wasting asset. What it needed in its place was something harder for Washington to walk away from. It needed Americans with payrolls in Taiwan. It needed quarterly earnings calls in which an American company would have to explain to its shareholders why the loss of Taiwan had just cost it twenty cents a share. It needed, in other words, to weave itself into the cost structure of American business.

Texas Instruments was the prize of that strategy. TI was not the first American electronics company in Taiwan. General Instrument had set up bonded export factories in 1964, and by the time the Kaohsiung Export Processing Zone opened on December 3, 1966, the world’s first such zone had attracted a dozen smaller assemblers. But TI was different in scale, in technological prestige, and in political proximity to Washington. Pat Haggerty, TI’s chairman, sat on advisory committees that reached into the Pentagon. The company had supplied the guidance circuits that made the Vietnam-era smart bombs work. To anchor TI in Taiwan was to anchor a piece of the American military-industrial system in Taiwan. The bonded factory in Chung Ho was, in that sense, a cousin of Tainan Air Base, where US aircraft damaged over Vietnam were being repaired and flown back into combat. Both were forms of indispensability. One was state-to-state. The other was firm-to-firm.

Li did not theorize the strategy in those terms publicly, but its logic ran through everything he wrote afterward. In his 1988 collection of policy essays, The Evolution of Policy Behind Taiwan’s Development Success, he presented Taiwan’s foreign-investment regime as the deliberate construction of “linkages,” a word he borrowed with the specific meaning that economists Albert Hirschman and Gustav Ranis had given it. A linkage was a relationship that, once formed, became expensive for either side to sever. Tax holidays alone created no linkages. Cheap labor alone created no linkages. What created linkages were factories with thousand-person headcounts, locally trained engineers, multiyear capital commitments, and supplier networks that wrapped local subcontractors around the foreign principal. Each of those, Li believed, was worth more than a treaty, because each could outlast the politicians who had signed the treaty.

The Chung Ho plant was built on those terms. TI was given exemptions worth, in present value, perhaps a tenth of its initial capital outlay. In return, Taiwan got a training pipeline. By the early 1970s the plant was running shifts of more than two thousand workers; production engineers cycled through Dallas for training and returned to Taipei to manage their own lines. The supplier ecosystem that grew around the plant, leadframe stamping shops in Sanchong, plastic molders in Banqiao, gold-wire vendors who learned the TI bonding tolerances by failing them, was infrastructure no government could have ordered into existence by decree. It came as a byproduct of an American payroll.

The TI plant in Chung Ho is, in retrospect, the first plainly visible link of the supply chain the United States cannot extricate itself from. The decision to assemble integrated circuits in Taiwan rather than in Texas was, on its face, a small and rational outsourcing of a low-margin step. Nobody in the room in 1969 thought they were committing the United States to a multigenerational dependency on an island a hundred miles off the coast of mainland China. Shepherd thought he was saving on labor. Li thought he was buying anchor weight. Neither was wrong. Both underestimated, by an order of magnitude, what they were starting.

The dependency grew the way ecological systems grow, by accretion. Once TI had a plant in Taiwan, the plant pulled in suppliers. The suppliers pulled in subcontractors. The subcontractors trained their own engineers. By the mid-1970s, when National Semiconductor, RCA, and Philips were all assembling on the island, Taiwan had a deeper bench of packaging-process engineers than anywhere in the United States outside the half-dozen company headquarters. By the time Li, by then a Minister without Portfolio with a brief over science and technology, persuaded the government to create the Hsinchu Science-based Industrial Park in 1980, the local labor pool was already several layers deep. Hsinchu’s first tenants were not pioneers. They were the next layer of accumulation on top of Chung Ho.

It is tempting, looking back, to read the 1969 decision as predatory: American capital exploiting a captive labor market and a politically vulnerable client state. The historical record does not support a reading that simple. The exchange was real, and both sides knew what they were buying. The Taiwan side, in particular, was not naive. Li and his colleagues at the Council for Economic Planning and Development had read the development literature. They knew the dependency-theory critique of foreign direct investment, and they had decided that, in Taiwan’s specific case, the alternative, political abandonment by a Washington that no longer had any commercial reason to defend Taipei, was worse than the risks of capture. They chose the linkage strategy with their eyes open. By every reasonable measure of Taiwanese welfare across the next half-century, they were proven correct.

What neither side could see, and what no actor in 1969 was positioned to see, was the second-order consequence: that the same accumulation logic which gave Taiwan its anchor weight against geopolitical abandonment would, decades later, give the entire global semiconductor industry its single most concentrated point of failure. The Chung Ho plant did not, by itself, produce that outcome. But it set the pattern. It taught American chip companies that the cost calculus of integrated-circuit manufacturing always pointed offshore. It taught Taiwanese officials that hosting American capital was a way of buying American attention. It taught both sides to keep doing what they had just done. They did, for the next fifty years, with metronomic consistency. By the time the question of how to undo the pattern became urgent, the pattern had become the industry.

Shepherd would run TI for another sixteen years, retiring as chairman in 1988, by which point the company’s Asian operations employed more people than its American ones. He never publicly described the Taiwan decision as a turning point. In his telling, in interviews and in the memorial tributes published after his death in 2009, the international expansion of the 1960s was a single sustained campaign of which Chung Ho was one episode among many. Li lived until 2001 and was lavishly memorialized in Taipei as the architect of the economic miracle, with Stanford professorships endowed in his name. Neither man, in his lifetime, was asked to defend the choice they had made together. Neither needed to. The choice had defended itself by becoming irreversible.

What stands out, in the photographs and the trade-press clippings that survive, is how unspectacular the original act was. No signing ceremony was preserved on film. The Chung Ho plant did not have a famous architect. The first day of operations, in August 1969, produced no front-page news in Taipei or in Dallas. A few hundred women, hired through provincial labor recruiters and bused in from townships south of the city, learned to operate die-attach machines. A few American supervisors, transferred from Sherman, Texas, learned to give instructions through translators. Plastic-encapsulated chips began moving from the loading dock to the port at Keelung, then by sea to Long Beach, then by truck to TI’s distribution center in California. The supply chain had its first link. The link held.

It is still holding.