AI summary by TwinMind
Can America’s Tech Giants Get Together & Start a Chipmaker?
Asianometry
20m 34s
Background: The Great Memory Shock of the 1980s
- Dynamic RAM (DRAM) was semiconductor industry's largest market segment
- Used as secondary memory in computers with insatiable demand
- Classified in generations: 1 kilobit → 4 → 16 → 64 → 256 → 1 megabit (each 4x larger)
- American origins and early dominance
- Invented by Robert Dennard (died at 91)
- Intel 1103 chip (1 kilobit) released 1970
- Mostek dominated 4 and 16 kilobit generations, became $350M company
- Mostek had 75% world market share at peak, pioneered 16-pin industry standard
- Japanese entry and rise
- Leveraged open format, superior manufacturing, cheap money
- By 1979 (64 kilobit generation): Hitachi, NEC, Fujitsu caught up
- 1983: Americans still owned 60% market
The 1984-1985 Market Collapse
- Economic slowdown late 1984-1985 caused DRAM market crash
- Pricing collapse during worst period:
- 256 kilobit chips: $2/unit
- 64 kilobit chips: <$1/unit
- Industry losses: $5B total ($4B by Japanese companies)
- Japanese survival factors: financial backing from electronics arms, "long-term thinking"
- American casualties: 11 US DRAM companies reduced to 3 by 1986
- Survivors: Micron, Texas Instruments (R&D only), Motorola (exited 1984)
- Micron became last true American memory maker
Industry Response and Trade Tensions
- Semiconductor Industry Association (SIA) deeply divided on response
- IBM opposed quotas/tariffs (needed cheap DRAM for PC boom)
- Reagan administration split between free trade commitment and industry protection
- Micron's late 1984 price cut triggered worst of 1985 slump
- Dumping charges led to discovery of Hitachi 10% memo (undercut competitors by 10%)
- 1986 Semiconductor Trade Agreement: Japan agreed to self-limit US sales, open domestic market
The 1988 Chip Shortage Crisis
- 1986 agreement failed to revive dead American companies
- Japanese cut 256k production while upgrading to 1 megabit fabs
- Mid-1988 economic surge caught everyone by surprise
- DRAM price explosion Q3 1988: $1.50 → $6 (Jack Tramiel, Atari CEO)
- Supply shortages affected entire electronics industry:
- Apple Macintosh SE delayed 1 year
- Sun workstations: 6-month backlogs
- Affected laser printers, scanners, fax machines, barcode readers
- Blame distributed among Japanese, trade agreements, American companies
Formation of US Memories (June 1989)
- SIA and American Electronics Association collaboration
- Cross-industry joint production venture
- Seven founding partners: IBM, AMD, Digital Equipment Corporation (DEC), Hewlett-Packard, Intel, LSI Logic, National Semiconductor
- Leadership structure:
- Chairman: Wol Corgan (LSI Logic)
- CEO: Sanford Kane (27-year IBM veteran, Sematech co-founder)
- Only full-time employee with up to 25 "borrowed" workers
- Goal: 4 megabit DRAM fab operational by end 1991
- Technology: Licensed from IBM (internal DRAM production)
- Business model: Cooperative where investors buy corresponding percentage of DRAM needs
Financial Structure and Projections
- Initial cost estimate: $1B (~$2.5B today)
- Seven companies contributed $50K each for admin expenses
- Kane's task: Raise remaining funding
- Projected to become $1B company by 1993
- 10% projected market share by 1992 start
Antitrust and Opposition Concerns
- Antitrust concerns due to major computer manufacturer participation
- National Cooperative Research Act of 1984 enabled R&D cooperatives like Sematech
- US Memories different: actual manufacturing for profit
- Most public joint venture since Toyota-GM (1983)
- Kane testified no anti-competitive concerns (only 10% market share)
- Sought legislative protection for joint production ventures
- Avoided government funding to reduce antitrust scrutiny
- Opposition from TJ Rogers (Cypress Semiconductor) and Scott McNealy (Sun)
Customer Commitment Challenges
- Major computer makers declined to join: Apple, Sun, Unisys, Compaq
- HP (original partner) hedged commitment
- Three key concerns emerged:
- DRAM shortage subsided, prices falling
- Startup scaling concerns (2,000 employees, full-scale fab in <2 years)
- Fear of Japanese retaliation (cutting off supply agreements)
The Dallas Meeting and Failure (January 1990)
- Pivotal meeting with 11 electronics companies
- Revised plan: $150M equity from electronics industry needed
- Goldman Sachs could raise additional $400-600M in equity/debt
- Required industry purchase commitments for Wall Street investment
- Only IBM and DEC firmly committed among computer makers
- Kane's poster board moment: "room got real quiet real fast"
- January 15th: Kane announced US Memories cannot proceed
Aftermath and Legacy
- Industry reaction: lack of American long-term strategic thinking
- New York Times op-ed: "US Memories: Too Short"
- Gordon Moore: "not strong endorsement of getting together"
- US never recovered domestic DRAM market share
- Korean memory giants and Intel CPU monopoly settled US-Japan tensions
- Kane later founded Tower Semiconductor (National Semiconductor Israel spinoff)
Modern Comparison to TSMC Competition
- Kane's assessment: US Memories model not competitive with TSMC today
- Intel foundry efforts: potentially $116B available, still insufficient
- Magnificent Seven split: $16.5B each required
- Technology licensing no longer available like IBM arrangement
- Modern requirements: Logic ASIC chips, latest nodes, immense scale
- Historical lesson: American companies unwilling to commit even $50M then
