For two decades, dynamic random-access memory — DRAM — was the most measured, most undifferentiated, most strategically loaded commodity in technology. The country that made it set the pace of computing. Between 1978 and 1998, that country changed twice.
The Japanese did not invent DRAM. They industrialised it. MITI's 1976–80 VLSI Project pooled the engineers of NEC, Toshiba, Hitachi, Fujitsu and Mitsubishi inside a shared lab in Kawasaki and pointed them at the next two memory generations. By 1981 Japanese factories were running at yields of 70–80% while American fabs were stuck at 50–60%. By 1985, the 64K chip that had sold for $3.50 in early 1984 was selling for 35 cents — a ten-fold collapse engineered by a Hitachi sales memo that instructed its U.S. force to "win with the 10% rule": bid 10% under any American price, no matter what the price was.
Then, on June 14, 1985, the Semiconductor Industry Association filed a Section 301 petition. Ten days later, an Idaho startup called Micron filed an antidumping case. By September 1986, the U.S. and Japan had signed a managed-trade agreement that put a floor under DRAM prices. The agreement did not save American DRAM — most of it had already disappeared. What it did instead was leave Korea, freshly entered into the business under Lee Byung-chul's 1983 Tokyo Declaration, an open lane to the global market.