The Potato Chip King

On a weekday morning in 1980, two brothers sat across a Formica table at the Royal Café in downtown Boise and tried not to sweat through their shirts. Joe Parkinson, the older twin by minutes, was a lawyer who had clerked for a federal appeals judge in New Orleans and had been moonlighting as president of a four-person company that fit inside the basement of a dentist’s office. His brother Ward was the engineer, the quiet one, the one who actually knew how a memory chip worked. The man they were waiting on was seventy-one years old, drove a white Lincoln with vanity plates that read SPUD, owned more potato acreage than any private individual in North America, supplied roughly half of McDonald’s french fries from his frozen-foods plants, and had no obvious reason in the world to give them money.

The brothers had already been turned down by every venture capital firm they had been able to reach. Sand Hill Road thought DRAM was a deathtrap in the spring of 1980. The Japanese had crossed seventy percent of the 64-kilobit market in a single product generation. Mostek, the former American champion that had taught Ward most of what he knew about chip design, was visibly breaking under the pressure of Hitachi and NEC. The men in California had looked at the Parkinsons’ pitch deck, looked out the window at the firms already losing money on memory, and politely declined. Idaho had no semiconductor industry. Idaho had no semiconductor venture capital. Idaho had potatoes.

J.R. Simplot walked in, ordered, listened, asked questions, and decided. According to the version Joe Parkinson would tell for the rest of his life, the meeting was not long. Simplot did not need a finished spreadsheet. He had spent his life buying low-margin commodities and selling them in volume. He had built his fortune on the proposition that if you could make the cheapest, most uniform potato product in the country and ship it on time, you would never lose. He looked at the brothers and understood that a memory chip, like a potato, was a fungible commodity whose buyers cared only about price, quality, and supply. He invested one million dollars in cash for forty percent of the company. He would later put in twenty million more. The company was Micron Technology, and at the moment Simplot wrote his check, it had no fab, no production engineers, and no customer left.

Two years earlier, in October 1978, Ward Parkinson and two former colleagues from Mostek’s design floor, Dennis Wilson and Doug Pitman, had walked away from a lawsuit. Mostek had sued the British-financed startup Inmos for raiding its engineers, and the three Boise-bound men had spent a brief unhappy stretch caught between the two companies. They quit Inmos five days before incorporating Micron Technology in the basement of a Boise dental office on East Park Boulevard. Joe came on as the fourth founder, the lawyer-administrator who would handle the parts of a company that Ward had no interest in handling. The plan was modest. They would consult on memory-chip design for whomever wanted to pay them. The company that initially wanted to pay them was Mostek itself, the same firm Ward had recently left.

Then United Technologies bought Mostek in 1979 and canceled the consulting contract. Micron’s only customer evaporated. Ward, who had been working on a redesigned 64K cell that he believed could be shrunk dramatically below what Mostek and the Japanese were producing, decided that the only way out was to build the chip themselves. That required a fab. A fab in 1980 cost on the order of one hundred million dollars at industry standard. Micron had nine.

What turned a Boise dental basement into a manufacturer was a peculiar coalition of rural Idaho money. Allen Noble was a potato farmer. Tom Nicholson was a sheep rancher. Ron Yanke ran a machine shop. Rudolph Nelson was a building contractor. None of them knew what a dynamic random-access memory chip was. They knew the Parkinsons. They put in tens and hundreds of thousands of dollars apiece. The brothers worked the local network the way commodity entrepreneurs had worked Idaho networks for half a century. Then they walked into the Royal Café and asked the richest man in the state to underwrite the rest of it.

Simplot’s logic was the cleanest piece of investment thinking in the early history of American DRAM. He had been born in Iowa in 1909, dropped out of school in eighth grade, and built a vertically integrated potato empire by understanding that commodity producers compete on cost. He had cut a handshake deal with Ray Kroc in 1967 to supply McDonald’s with frozen french fries, a contract that by the early 1980s was supplying more than half of the chain’s volume and had made Simplot one of the wealthiest men in the United States. To Simplot, what the Parkinsons were proposing was not a new technology bet. It was a commodity-manufacturing bet. The Japanese had not won the 64K DRAM by inventing some unreproducible process. They had won by running clean fabs at high yield, with disciplined operators and patient capital, and by being willing to dump prices to clear competitors. Simplot understood that game. He had played it on potatoes. He told the Parkinsons that DRAM was a gamble, and that he would take it.

By 1981, Micron had finished its first wafer fabrication unit, modestly named Fab 1, on the Boise outskirts. The company built it for roughly twenty million dollars, about a quarter of the typical industry cost. Some of that came from Idaho’s relative cheapness in land, labor, and electricity. Idaho Power’s hydroelectric grid sold electricity at rates a third of California’s. Some of it came from Yanke’s machine shop, which fabricated unglamorous fab equipment in town that other companies had been buying from specialty suppliers in California and Massachusetts. Some of it came from sheer refusal to spend money. Ward Parkinson designed equipment in-house when Micron could not afford to buy it. The fab opened with a single production line.

The product Ward had been designing in his basement was a 64K DRAM whose physical area was about half the size of the Japanese leaders and roughly a third smaller than Texas Instruments’ comparable part. Smaller chips meant more dies per wafer; more dies per wafer meant lower cost per unit at any given yield. It was an engineering accomplishment that mattered for a specific commercial reason. Micron could not outspend NEC, and it could not match Hitachi’s clean-room discipline overnight. But if Ward could squeeze the cell smaller than anyone else, the math of cost-per-bit would work in Micron’s favor even at lower yields. The chip went into volume in 1981. By 1982, Micron had shipped over a million of them. Some went into the Commodore 64, the home computer that introduced a generation of American teenagers to BASIC. The buyers did not know or care that the part came from a fab in Idaho. They cared that it was cheap and that it ran.

In June 1984, Micron went public, listing on Nasdaq under the ticker MU. The offering was 2.1 million shares; the stock traded between fourteen and forty dollars across the back half of the year. The 1984 financials looked outright spectacular. Micron posted twenty-nine million dollars of after-tax profit on eighty-four million dollars of revenue, a margin ratio that placed the company among the most profitable electronics firms in the world. A photograph from the listing day showed Joe Parkinson in a suit and Simplot standing beside him, eighty-one years old, grinning at the cameras. The Boise Idaho Statesman ran the photograph above the fold. The Parkinsons had taken five years to turn a basement consulting shop into a publicly traded chip company. They had done it by ignoring the entire venture-capital orthodoxy of Silicon Valley, by leveraging Idaho’s hydroelectric grid and ranch capital, and by shipping the smallest 64K cell in the industry.

The triumph lasted about twelve months. In the back half of 1984, Japanese DRAM producers began the price-cutting campaign that would liquidate the American memory industry. The 64K part that had been selling for around three and a half dollars in late 1984 sold for around twenty-five cents by mid-1985. Micron’s earnings collapsed from twenty-eight point nine million dollars in 1984 to roughly one hundred fifty thousand dollars in 1985, a number rounding error that did not technically count as a loss but that everyone inside the company understood as one. The 256K generation, the next product line, was already being dumped at similar slopes. Micron was running out of cash.

In the spring of 1985, Joe Parkinson laid off half the workforce. Roughly six hundred people walked out of Boise fabs and offices over a few weeks. One of Micron’s two production lines went cold. The company was, by internal accounts, down to about two weeks of cash on hand. Simplot, by then a director, did what he had said he would do in the Royal Café five years earlier: he wrote another check. Twenty million dollars more flowed in across 1985 and 1986, propping the company through the trough that finished off most of Micron’s American competitors. The board began meeting at six in the morning every Monday at Elmer’s Pancake House on Fairview Avenue, a coffee-shop diner where Simplot would walk in, sit down, and ask any latecomer if he had slept in. He would steal bacon strips off the plate of whoever was next to him. He would not raise his voice. The decisions were made between eggs and refills.

In June 1985, Micron filed an antidumping petition with the Commerce Department against Japanese 64K DRAM producers. The decision to file was Joe Parkinson’s, and it was treated by the SIA, the trade association of the larger American chipmakers, as both useful and embarrassing. Useful because it gave Washington a domestic injured party to point at; embarrassing because the small Idaho firm everyone in California had ignored was now leading the political effort that the bigger companies had been politely skirting. The Commerce Department initiated the formal investigation later in 1985, the Reagan administration self-initiated a separate case on 256K and future generations in December, and by the fall of 1986 the United States and Japan had signed the bilateral Semiconductor Trade Agreement. The full politics of that agreement belong to the broader American policy story; what mattered for Micron was that the cliff stopped sliding. By 1988, with the cycle inverted and prices recovering, Micron’s revenue had reached three hundred million dollars and earnings had jumped to nearly a hundred million. By the second quarter of that year, the company was clearing seventeen million dollars in profit against a ten-million-dollar loss in the same quarter of 1987. The DRAM industry had not been saved. Most of it had been incinerated. But the small subset of it that was still American-headquartered was now anchored in Idaho.

The longer story of Micron’s survival is not a single dramatic rescue. It is twenty years of repeated trough management. Through the late 1980s and the 1990s, Micron’s revenue and earnings cycled in patterns that would have terrified any board accustomed to predictable consumer goods. Sales of three hundred thirty-three million in 1990 became eight hundred twenty-eight million in 1993. Profit cratered, recovered, cratered again. In 1996, when chip prices fell seventy-five percent and the industry shed a quarter billion dollars of Micron’s earnings overnight, the board fired its young chief executive, Steve Appleton, who had risen from the production floor at four dollars and forty-six cents an hour to the corner office over eleven years. Eight days later, after a near-revolt by senior managers, the board rehired him. Appleton returned with terms that ended the board’s daily interventions in operations. He kept the fabs running close to ninety percent capacity utilization while competitors held theirs to sixty or seventy, on the theory that if the trough was inevitable you might as well hold output and force the cycle’s pain onto everyone else’s balance sheet. The strategy was, in spirit, J.R. Simplot’s potato strategy applied to silicon. The lowest-cost producer of the most uniform commodity wins on volume.

The companies that had been Micron’s American peers stopped existing one by one. Mostek had been gone since 1985. Intel, by 1986, had decided the memory business was unwinnable and had walked away from DRAM to become a microprocessor company. National Semiconductor exited memory in stages. Motorola wound its DRAM operations down. By the mid-1990s, what remained of American DRAM came down to two firms: Micron and Texas Instruments. In June 1998, with TI losing money on memory and looking for a way out, Steve Appleton bought TI’s DRAM business in a stock-and-debt transaction worth roughly eight hundred million dollars. The deal handed Micron a wafer fab in Avezzano, Italy, a fab in Richardson, Texas, an assembly and test plant in Singapore, and a second seat at the global leadership table. The transaction closed on October 6, 1998. From that day forward, Micron was the only American-headquartered DRAM producer of any scale, and Idaho was the entire American memory industry.

Simplot was eighty-nine years old when the TI deal closed. He retired from Micron’s board the following year, having sat on it for almost two decades. His Micron stake, which began at one million dollars for forty percent in 1980, had at various points been worth more than three billion dollars. He held it through almost everything. He had told Fortune in a 1995 profile that Micron was his “baby” and that he had bankrolled it, propped it up when it was on the brink, and helped push it over the top. The phrasing was not modest. It was also not wrong. Without his money in 1980 and again in 1985, the company would not have lived to see the trade agreement, the Texas Instruments acquisition, or the consolidation wave that was about to remake the global memory industry around three remaining producers: Samsung, Hynix, and Micron.

The wave kept coming. In February 2010, Micron acquired Numonyx, the flash and phase-change memory venture that Intel and STMicroelectronics had spun off two years earlier, for roughly one and a quarter billion dollars in stock. The deal added NOR and NAND flash to Micron’s portfolio and gave it a new memory technology to develop. In July 2012, Micron agreed to acquire the bankrupt Japanese DRAM maker Elpida, the last domestically headquartered Japanese memory firm, formed in 1999 from the merged DRAM operations of NEC and Hitachi. Elpida, the same conglomerated descendant of the firms that had nearly killed Micron in 1985, had filed the largest Japanese bankruptcy since Japan Airlines, undone by a strong yen and the same DRAM price collapse that had once nearly ended the Idaho company. Micron paid roughly two and a half billion dollars across the Elpida and Rexchip transactions. The deal closed in mid-2013. The buyer was the small American firm that the Japanese had once tried to extinguish with twenty-five-cent chips. The seller was the corporate descendant of the firms that had done the dumping. Joe Parkinson had retired in 1994, three months before Steve Appleton replaced him. He did not run the Elpida deal. But the symmetry was complete enough that he could enjoy it from his cattle ranch in Grand View.

J.R. Simplot died on May 25, 2008, at age ninety-nine, the oldest billionaire on the Forbes 400 at the time of his death. By then he had largely stepped away from active involvement in Micron, though his estate retained a meaningful stake. He had once told an interviewer that he was a fact man, and that if it didn’t add up, he didn’t buy it; that he didn’t believe in hocus pocus. He had also said, when asked about Japanese tariffs on Idaho potatoes, that the Americans could “out-tech ‘em and out-produce ‘em,” and that they would beat the hell out of the Japanese. He had been wrong about potatoes; the tariff fight was already lost. He had been right about chips, in the only way that mattered, which was that he had kept his small Idaho company alive long enough to be on the right side of the consolidation when it came.

What survived in Boise was not just a firm. It was the only piece of American DRAM manufacturing that had managed to remain American across the entire generation of Japanese ascendancy and the subsequent Korean rise. Micron was, by the late 1990s, the institutional memory of an industry that had otherwise dispersed to East Asia. Its fabs in Idaho ran continuously through cycles that drove competitors out. Its workforce, drawn from the same labor markets that had once supplied Simplot’s potato plants, treated chip-making as an industrial discipline rather than as an exotic technology. The company that had been told no by every venture firm on Sand Hill Road, that had been founded in a basement with the help of a sheep rancher and a potato farmer, had outlasted Mostek, Intel’s memory business, National’s memory business, TI’s memory business, and Elpida. None of that was inevitable. Most of it traced back to a one-million-dollar check written across a table at the Royal Café in 1980, and to the conviction of a man with a SPUD license plate that a chip and a potato were not, in the end, all that different.

The lesson Micron drew from its survival was not a glamorous one. It was that in commodity manufacturing, the firm that holds on through the bad years inherits the market when the bad years end. Idaho would soon need to make that bet again. So would the rest of the country.