"Transistor Girls"

One afternoon late in 1963, on the Kowloon side of Hong Kong harbor, four American managers walked through a low concrete building on Hang Yip Street. The structure had recently been a rubber-shoe factory. The smell of latex still clung to the walls. Now Fairchild Semiconductor was leasing the space and stripping it down for a different kind of work. A few weeks later, the first of several hundred young women would file in through the front gate, take a seat at a long wooden bench, and lower their eyes to a microscope.

The wafers came from California by airfreight. They had been grown, doped, masked, and diced in Mountain View. Each wafer carried thousands of tiny transistor dies, each one no larger than a grain of black pepper. The job in Hong Kong was what the industry called the back end. A worker picked up a die with tweezers, set it on a metal lead frame, bonded gold wires from the die’s pads to the frame’s posts, and encapsulated the assembly in a glob of epoxy. Then the next worker tested it. Then it went into a box bound back across the Pacific, where Fairchild’s salesmen would ship it to IBM, Honeywell, the Air Force.

The wage at the Bay Area assembly bench was $2.50 an hour. The wage in Hong Kong was about ten cents. Within a year, the plant had shipped one hundred and twenty million transistors. Within three years, it would employ five thousand workers, more than Fairchild’s entire California payroll combined. None of this had been planned in any precise way. It had begun with a hunch and a vacation.

The hunch belonged to Robert Noyce, by then Fairchild’s research director. Noyce had quietly invested in a small Hong Kong company that assembled transistor radios from Japanese components and was struck by how cheap the labor was. He suggested to Charlie Sporck, who ran Fairchild’s manufacturing, that the company think about doing the same thing for chips. The vacation belonged to Jerry Levine, Fairchild’s manager of administration and planning, who happened to spend a holiday in Hong Kong in 1962 and came back arguing the same case with numbers attached. Sporck was open to it. He had spent four years trying to industrialize semiconductor production in Mountain View and had run head-on into a culture that, he later complained, had no concept of mass manufacturing. Hong Kong looked like a way around the problem.

Sporck flew to Hong Kong with Julie Blank, who handled facilities, to scout buildings. They picked the old shoe plant on Hang Yip Street because the rent was cheap and the airport was close. They hired four Americans to manage. Everyone else, several hundred at first and thousands within two years, was hired locally. Almost all were women between roughly eighteen and twenty-four. The plant was running production within months. By 1965 a trade reporter visiting the operation noted, with mild surprise, that a female worker at a Hong Kong electronics plant earned about a dollar a day for her first three months and roughly a dollar thirteen after she was off probation.

The arithmetic of the move was almost embarrassingly simple. A transistor in 1963 cost a few dollars to make. A meaningful share of that cost was the human labor of bonding a wire to a die and sealing the package. Replace a Bay Area woman earning two and a half dollars an hour with a Hong Kong woman earning ten cents an hour and that share collapsed by more than nine-tenths. Wilf Corrigan, the British engineer who would later run Fairchild’s overseas manufacturing, used to point out that a dollar a day was, in nominal terms, what Chinese laborers had been paid to lay railroad track in California a century earlier. The numbers had not changed much. Only the work had.

There was a second piece of arithmetic that competitors would understand only slowly. American semiconductor firms in 1963 were debating how to bring labor costs down at home. Texas Instruments and Motorola were both spending heavily on automated bonders, machines designed to replace the human at the microscope. Fairchild’s Hong Kong move quietly invalidated those investments. Why build a million-dollar machine to replace a worker when you could hire ten Hong Kong workers for the same year and have nine left over? Within five years every American merchant chip company would be following Fairchild east. Industry research dollars on assembly automation slowed sharply for a decade.

Sporck himself moved on quickly. In early 1967 he was poached out of Fairchild to run a struggling small chip outfit called National Semiconductor. He took a number of Fairchild manufacturing veterans with him, fired half of National’s headquarters staff in his first week, and immediately began looking for his own offshore site. Hong Kong was already saturated with Fairchild and Japanese competitors. Sporck instead set up National’s first overseas plant in Singapore, in a converted shophouse in the Kallang industrial estate, in 1968. Texas Instruments arrived in Singapore the next year, opening a five-acre plant in Kallang Basin with a ribbon cut by Goh Keng Swee. Within five years of Fairchild’s Hong Kong opening, the Asian assembly map had three nodes: Hong Kong, Singapore, and Seoul, where Fairchild had built a wholly owned subsidiary called Semikor in 1966.

In Penang, the same logic arrived almost a decade after Hong Kong, in concentrated form. George Town, the colonial port at Penang’s northern tip, had lost its free port status in 1969, and unemployment in the state had climbed past sixteen percent. The Gerakan government under Lim Chong Eu commissioned a development plan, accepted the recommendation that Penang go after foreign electronics investment, and pushed through a Free Trade Zones Act in 1971. The first phase of the Bayan Lepas zone opened in 1972 on a strip of land between the airport and a river, much of it reclaimed from rice paddies. Eight multinationals signed on in the first wave. Penangites would later call them the Samurai Eight: Intel, Advanced Micro Devices, Hewlett-Packard, Clarion, National Semiconductor, Hitachi, Osram, and Bosch.

Intel’s plant, code-named A1, was the most consequential. By late 1972, the company that had been founded only four years earlier in Santa Clara had committed one and a half million dollars to a small assembly building in the middle of a Malaysian field. Andy Grove and Keith Thomson, two of Intel’s manufacturing principals, flew over with their wives to visit the site shortly after groundbreaking. The car got stuck in monsoon mud on the way in. The first hundred employees were Malay women, many recruited from kampungs, the rural villages along the coast. They came to work in batik. Within a few years Penang’s plant supplied more than half of Intel’s worldwide assembly capacity, and the company would draw more revenue from Malaysia than from any other single offshore site for two decades.

What made Penang different in flavor from Hong Kong was the deliberateness of the recruitment. A young woman in Penang in 1973 was the daughter of a smallholder rubber tapper or a coastal fisherman. She had finished primary school, possibly some secondary. She had grown up in a kampung where wage labor was uncommon and where her economic role had been confined to household and field work. The Free Trade Zone offered her a salary in Malaysian ringgit roughly equivalent to thirty US dollars a month, paid weekly in cash. That figure was below what an American assembly worker earned in a single shift. In the kampung, it was an enormous sum. Daughters who took factory jobs frequently became the highest earners in their households within a year, a phenomenon documented in painstaking detail by the Berkeley anthropologist Aihwa Ong, whose 1978 fieldwork in a Selangor village she called Sungai Jawa traced the economic and cultural shock of that inversion.

The labor came with conditions that the firms tended to describe in the language of efficiency. Plant managers and personnel officers in interviews from the era explained that they preferred women to men because women had finer motor skills and more patience for repetitive precision. The dexterity rationale, repeated almost verbatim from Hong Kong to Penang to Manila, traced the supposed advantage to needlework and other domestic skills girls were said to have learned at home. There was almost no empirical work behind it. No one had run a controlled experiment showing that women bonded gold wires faster than men. But the rationale flattered the firms’ preferred hiring policy and was repeated until it became a folk truth of the industry. Underneath it sat the harder fact, which the personnel officers occasionally admitted off the record: women in Southeast Asia could be paid less than men, were less likely to leave for a competitor, and were less likely to attempt to unionize.

Most of the workers were between roughly eighteen and twenty-four. The lower bound was set by labor law, by the firms’ worry about child-labor headlines, and by the practical problem that younger eyes were better at the microscope. The upper bound was social: a woman who married was expected to leave the plant or, if she stayed, to handle a different role. Many of the firms ran their own dormitories beside the plant, single-sex, six to twelve workers to a room, with a curfew. Others lived in private hostels in nearby kampungs, leaving the village on a Sunday evening to return to the dormitory and walking home Friday night with the week’s pay. The shift pattern across the region was remarkably uniform. Eight-hour shifts, three shifts a day, six days a week, with a rest break that a 1980 Multinational Monitor survey of regional plants tallied at about forty-five minutes total over an eight-hour run.

The work was monotonous and microscopic. A bonder operator picked up a die with tweezers, lined it up under a microscope, fed gold wire through a capillary, and tapped a button with her foot. The bonder welded the wire to a contact pad on the die and then to a pin on the lead frame. She did this several thousand times a shift. A test operator pressed a finished package against a row of probes, watched a meter, and sorted the package into a pass or fail bin. By the late 1970s, Karen Hossfeld, doing parallel labor research in Silicon Valley among the largely Vietnamese, Filipino, and Mexican women still doing the same job in California, would conclude that the personnel formula had not changed at all between San Jose and Penang. The firms preferred immigrant or rural women because they were cheap, available, and unlikely to make trouble. Geography simply changed which women were closest at hand.

The trouble came anyway, in unexpected forms. In the late 1970s, a wave of spirit-possession incidents swept through Malaysian electronics plants, particularly the Japanese-owned ones in the Selangor and Penang zones. A worker would scream at her station, fall to the floor, claim to see a hantu, a spirit, in the clean-room ceiling. The episode would spread to several other workers in the same shift, then to dozens. Lines would shut down for hours while imam were called in to bless the building. Plant managers found these incidents bewildering. Ong, who interviewed both workers and managers, read them as a refusal that could not be expressed in any other vocabulary, an objection to the discipline of the corporate clock that an unmarried Malay woman could lodge only by becoming, briefly, someone other than herself. The firms eventually accommodated the practice. They hired bomohs, traditional spirit-handlers, on retainer. Production resumed.

By the early 1980s, when an International Labour Organization survey looked at electronics workers across East and Southeast Asia, the count was something like two hundred and fifty thousand women across the region. In Penang, ninety percent of the assembly-line workforce in the multinational electronics plants was female. In Singapore the figure was similar. South Korea’s plants in Kuro and Masan ran on the same demographic. The Philippines added Bataan in 1972 and Baguio later in the decade. By the time the wave reached Bangkok and Surabaya, the template that had been refined in Hong Kong fifteen years earlier was being copied without significant modification at every stop.

The wage differential that had brought Sporck to Kowloon in 1963 narrowed slowly. By 1980 a Hong Kong assembly worker was earning closer to a dollar an hour, a tenfold rise. By the late 1980s, Hong Kong itself was being abandoned for the lower-wage interior of southern China. Singapore and Malaysia followed an upward path. Each country, having captured the bottom rung of the assembly ladder, would either climb or be left behind by the firms moving on. Singapore climbed by deliberately upgrading the work, drawing in fabrication and then design, as Linda Lim documented in her 1978 dissertation. Malaysia stayed largely with assembly and test, but managed to keep most of the multinationals through tax incentives and infrastructure. The Philippines lost ground in the 1980s when political instability scared off a generation of investors, then regained it later. Korea built its way up through DRAM into something altogether different.

What survived through all of this was the structural pattern. Chip manufacturing, by the early 1970s, was no longer a single integrated process taking place in a single building. It had been pulled apart into a front end and a back end, with the front end, the wafer fabrication, kept close to design and engineering in California or Texas, and the back end shipped to wherever labor was cheapest. The seam that ran across the Pacific would, over time, get longer and more elaborate. Photomasks would migrate. Test would migrate. Eventually fabrication itself would migrate, though that took another twenty-five years and required a different kind of company. The original logic, that some part of chipmaking would always live in Asia because the labor there was abundant and the firms could pay it less, was set in 1963 in a converted shoe factory off Hang Yip Street.

Charlie Sporck would later be asked whether he had understood, at the time, that he was setting a template for the rest of the century. He claimed he had not. He had been trying to bring a single year’s manufacturing costs down. The women on the line in Kowloon, and later in Bayan Lepas, also did not particularly care about templates. They had taken the work because it paid more than anything else available, often a great deal more. They sent most of the money home. They roomed with strangers. They made the chips that flew back across the ocean and went into telephones, missiles, the early minicomputers, and the radios their American customers carried to the beach. The chip industry’s first global supply chain was theirs.