EE 292P: Atoms, Bits, and National Interest (Mar 10)

Attendees: Dustin J Ross Date: March 10, 2026 Type: Class Session

Summary

Professor Robert Bergelman’s longitudinal study with Andy Grove
- 12-year field research from 1988-2000
- Documented Intel’s transition from memory company to microprocessor leader
Strategic vectoring concept
- Strategy as direction + magnitude (vector)
- Andy Grove’s approach to aligning organizational forces
- Creates momentum but can lock in strategic direction

Resource allocation rule: “maximize margin per wafer start”
- Highest margin products get priority during capacity constraints
- Reinforced Intel’s identity as leading-edge technology company
- Microprocessors had higher margins than commodity DRAM
Middle management drove technical decisions
- Limited top management’s decision space
- Engineers like Lescom developed RISC processor without full leadership awareness
- Bottom-up innovation critical for strategic pivots

Japanese competition commoditized memory market
- Superior manufacturing capabilities
- Strategic focus on semiconductor dominance
Intel’s declining market share
- Fell to 3% market share in DRAM
- R&D allocation remained with memory division despite declining revenues
- Process technology leadership maintained but volume advantage lost to competitors

IBM PC partnership created installed base
- Intel chosen for technical marketing excellence and roadmap clarity
- “Crush the competition” campaign secured 2,000 design wins
Network effects and increasing returns
- Large installed base attracted independent software vendors (ISVs)
- More software increased platform value
- Self-reinforcing cycle of adoption
Intel Inside branding strategy
- Direct customer connection bypassing OEMs
- Reduced investment risk for $2B+ fab investments

Successful transformation requires resolving identity tension
- Three-year period where Intel was both memory and microprocessor company
- Leadership must clarify strategic direction
Vectoring creates both strength and vulnerability
- Alignment enables execution but creates strategic lock-in
- Moore’s Law became “prison” - incorporated everything into CPU rather than specialized chips
Hubris from success
- “Volume begets standards” worked in PC industry
- Failed in telecommunications where committees set standards
- Missed mobile opportunity due to margin requirements

Missed major platform shifts
1. Mobile computing - rejected low-margin smartphone processors
2. AI acceleration - could have bought Nvidia for $20B in 2005
3. Communications - failed WiMax and other telecom ventures
Manufacturing challenges
- Stuck at 10nm process for 5+ years while TSMC advanced
- Lost process technology leadership
Pat Gelsinger’s tenure (2020-2024)
- Four-front battle: manufacturing vs TSMC, x86 vs AMD/ARM, foundry business, AI vs Nvidia
- Valuation dropped from $500B peak to $100B when fired
- Cost-cutting measures: 20,000 layoffs, dividend suspension

New horizontal structure emerged
- Hyperscalers (Amazon, Google, Microsoft) designing custom chips
- TSMC as dominant foundry serving multiple customers
- Specialized chip designers (Broadcom, Qualcomm) thriving
Market valuations (2024-2026)
- Nvidia: multi-trillion company
- TSMC: $2T+ valuation
- Broadcom: 10x growth in 2 years due to data center interconnects
- Intel: struggling with $100B valuation vs $20B+ fab costs

New CEO Lip-Bu Tan’s unclear strategy
- Inherited four-front competitive battle
- Limited options: get big, become niche player, or exit
Potential focus areas discussed
- Abandon manufacturing, focus on design like AMD
- Target specific markets rather than compete everywhere
- Leverage existing talent (many Intel alumni now at hyperscalers)
Structural challenges
- Lost innovative culture through decades of Moore’s Law focus
- Optimized for scale rather than agility
- Activist investor pressure for short-term returns