Intel's Early Success with Memory (1968-1979)

• Founded to produce standardized low-cost MOS memory chips for computing systems
• 1968: Robert Denard patented dynamic RAM concept (one transistor, one capacitor)
• 1969: Intel 1101 - first high-volume MOS semiconductor memory (256-bit SRAM)
  • Cost 3 cents per bit, too expensive vs core memory
  • Required 6 transistors per cell
• 1970: Intel 1103 DRAM breakthrough
  • 1 kilobit capacity at 1 cent per bit
  • Used 3-transistor design with buried contact innovation
  • "Core killer" became world's best-selling semiconductor by 1972
• Late 1970s: Intel shifted focus from DRAM capacity race to higher-margin products
  • EPROM (erasable programmable ROM) invented by Dov Frohman
  • Microprocessors introduced November 1971
• 1979 financials: $661M revenue, $78M profit, 14,300 employees globally

Japanese Competition Emerges (1960-1980)

• 1960: Bob Noyce visits Japan, judges their semiconductor industry "years behind"
• 1974: All 4K DRAM generation was American-made
• 1979: Japan captured 35% of 16K DRAM market
• 1978: Noyce returns to Japan, shocked by superior Japanese quality
  • Fewer defects than Intel products
  • Fortune Magazine article "Japanese Spies in Silicon Valley"
• Customer feedback deteriorating:
  • IBM repeatedly told Intel "yield sucked"
  • HP VP: "I go out of my way to design Intel stuff out"
• March 1980: HP presentation showed best American 16K DRAMs had 6x more defects than worst Japanese
• Japanese advantages: better yields, not dumping or cherry-picking

Microprocessor Success and Operation Crush (1980-1981)

• 1980: Operation Crush launched against Motorola 68000
  • Goal: reduce Motorola to 15% market share
  • Generated 2,500 design wins
• Intel 8088 processor: 16-bit internal, 8-bit external data bus
  • Rushed development in 18 months
  • Lower cost alternative to full 16-bit systems
• IBM PC development (1980-1981):
  • Earl Whetstone sold 8088 to IBM Boca Raton facility
  • Extreme secrecy: Intel support through black curtains
  • IBM demanded AMD as second source
  • Cross-licensing agreement signed with AMD
• August 1981: IBM PC launched, instant hit
• 1981: 33% of Intel's sales came from IBM alone

DRAM Market Collapse (1982-1985)

• 1982: Intel knew it was "hopelessly behind" in DRAM
• 1980: Intel had only 3% DRAM market share
• End 1979: Fujitsu unveiled 64K DRAM at volume
• Intel's 64K DRAM strategy:
  • Chip-level redundancy with blown fuses
  • Fuses didn't work properly initially
  • 2-year delay vs Japanese competition
• Japanese 64K production: 9M pieces (July 1981) → 66M pieces (August 1982)
• 1981: Grove implemented "125% solution" - extra 2 hours/day for 6 months
• 1982: IBM bought 12% Intel stake to shore up supplier
• Mid-1982: Intel sold more logic than memory for first time
• Scott Gibson (memory GM) advocated exiting DRAM business
• Arguments for staying: one-stop shop, process driver, Intel's "baby"

The Crisis Deepens (1984-1985)

• 1984: Economy and silicon cycle crashed
• DRAM and EPROM prices collapsed
• Japanese cut EPROM prices 70-90%
• Intel lost money 16 consecutive months (1985-1986)
• 1985: 16% revenue decline, profits vanished
• Hitachi memo: "10% below their price, if they requote go 10% again, don't quit until you win"
• October 1985: AMD, Intel, National Semiconductor filed dumping petition
• Led to 1986 US-Japan Semiconductor Trade Agreement
• Late 1984: Intel's 256K DRAM share fell to 4%

The Famous Exit Decision (Mid-1985)

• Grove-Moore meeting looking at Great America ferris wheel
• Grove: "If board brought in new CEO, what would he do?"
• Moore: "Get us out of memories"
• Grove: "Why shouldn't we walk out and come back and do it ourselves?"
• Craig Barrett's perspective: "We were effectively out of DRAMs when they walked out"
• Implementation took another year
• New GM kept holding onto 1-megabit R&D program
• Intel left DRAM "kicking and screaming"

Legal Victory Against NEC (1979-1986)

• 1979: Intel licensed 8086/8088 production to NEC for Japan
• NEC created "Intel compatible" V20/V30 chips
• Clean room reverse engineering of hardware (legal)
• NEC copied microcode (embedded CPU firmware)
• Court ruled microcode can be copyrighted
• Combined with 1984 Semiconductor Chip Protection Act
• Created legal cudgel against unauthorized copying

The 386 Strategy and Sole Sourcing (1985-1986)

• 386: Intel's first 32-bit processor, 2x transistors vs 286
• Offered to IBM, they refused
  • Feared cannibalization of minicomputer business
  • Wanted control with PS/2 system
  • Preferred 286 for manufacturing rights
• Grove pushed for sole sourcing 386 (major risk)
  • Antagonized AMD partnership
  • Challenged IBM's power
  • Required massive manufacturing scale-up
• Worked with Compaq on DeskPro 386
• Strategy required manufacturing excellence

Manufacturing Revolution Under Barrett (1985-1986)

• Craig Barrett: Stanford PhD, former professor, track star
• 1985: Made head of all fabs
• Mission: Cut manufacturing costs 50% by 1985, then 50% again by 1986
• Closed plants in California, Oregon, Puerto Rico, Caribbean
• Cut nearly 30% of workforce
• Studied Japanese techniques, cleanliness standards
• Adopted "Copy Exactly" methodology
  • R&D experiments on production lines
  • Eliminated fab-to-fab variations
  • Replicated entire plants identically
  • Inspired by McDonald's consistency
• Oregon DRAM fab converted to R&D facility
• Yields soared, enabled global microprocessor ramp

Recovery and Long-term Impact

• Three key factors in Intel's recovery:
  1. Trade petition halted Japanese EPROM dumping
  2. Microcode victory + 386 sole sourcing locked in x86 monopoly
  3. Copy Exactly fixed manufacturing problems
• DRAM exit less important than commonly believed
  • Intel effectively lost DRAM at 4K generation
  • Market share "closer to zero than 10%"
• Strategy created multi-billion dollar company
• Established foundation for Intel's dominance through PC revolution