Key Principle
Moore's Law delivers ~60% annual improvement in processing power, but designers' ability to utilize transistors grows at only ~20% per year. This structural mismatch is the quantitative engine of overshooting. The critical question is not whether Moore's Law will continue but whether it will continue to matter -- once mainstream customers are overshot, the basis of competition flips from raw speed to convenience and customization, forcing a wholesale reorganization of the semiconductor value chain.
Conservation of integration governs this reorganization: whatever needs to be optimized must be surrounded by something configurable. The alternating architecture of interdependent/optimized and modular/configurable layers is structurally necessary, not accidental. When overshooting hits, previously proprietary and profitable stages commoditize while previously marginal stages become the new profit pools.
Why This Matters
- Overshooting triggers value chain reconfiguration, not just market share shifts. New entrants do not simply steal customers; the entire architecture of how the industry creates and captures value restructures around specialist firms.
- Incumbents' strengths become liabilities. Intel's exacting processes, high margins (49-62%), and massive scale make $20M customization markets invisible -- the same excellence that sustains dominance creates footholds for attackers.
- Applies beyond semiconductors. The pattern -- Moore's Law-style overshooting, basis-of-competition flip, conservation of integration driving value chain reconfiguration -- is a general template for any industry with exponential improvement trajectories.
Good Examples
- Transistor disrupting vacuum tubes. Initially inferior on power handling, but small size and low power consumption opened nonconsumption markets (hearing aids, pocket radios). Sony's transistor radio delighted teenagers who had no portable alternative. None of the vacuum tube giants became major semiconductor players. (Ch. 7)
- Intel's disruptive black belt responses. Celeron (separate organization in Israel countering AMD/Cyrix low-end attack), Centrino (optimized for mobile convenience -- wireless and battery life over raw speed), StrongArm (Hudson, MA facility for custom-configured value network). Each used an autonomous organization with distinct processes and values. (Ch. 7)
- Fabless model emergence. Xilinx reached >$1B revenue by 2003 on programmable logic. Tensilica enabled engineers to customize systems-on-a-chip via a website by mixing IP blocks -- modular, convenience-based competition made concrete. (Ch. 7)
- Foundry power reversal. TSMC and UMC began dictating design rules to chip designers and interface standards to equipment suppliers, inverting historical power dynamics -- power accrues to whoever controls the interdependent, not-good-enough stage. (Ch. 7)
Counterpoints
- Commoditization risk for fabless firms. Being disruptive is necessary but not sufficient. As architecture modularizes and assembly becomes rules-based, design risks becoming undifferentiated. Value migrates to customers, component providers, and manufacturers unless entrants own an interdependent, hard-to-replicate stage. (Ch. 7)
- Performance-demanding tiers persist. Applications still taxing the fastest processors (3D games, digital video editing, real-time speech-to-text) exist but represent a fraction of mainstream volumes. The frontier still requires rule-breaking engineering, not rule-following design tools. (Ch. 7)
- Entrant discipline required. Entrants jumping too soon to big markets face devastating consequences -- stretching underperforming customized products to demanding customers fails just as surely as incumbent cramming. (Ch. 7)
Key Quotes
- "But what if performance continues to follow Moore's Law, and it doesn't matter?" (Ch. 7)
- "The very thing that makes Intel great is the very thing that creates opportunities for the attacking firms." (Ch. 7)
- "The wafer fabs that succeed will be those that figure out how to make integrated circuits in a system similar to the one in which Toyota makes cars." (Ch. 7)
- "Conservation of integration means optimizing the links between each step in the process. This will require deoptimizing each process step." (Ch. 7)
- "Without the benefit of theory, these developments would likely be viewed as meaningless noise; with the benefit of theory, these developments can be viewed for what they are -- harbingers of massive change." (Ch. 7)
Rules of Thumb
- Diagnose overshooting by the design gap. If improvement rate exceeds utilization rate, overshooting is structural and inevitable -- plan for the basis-of-competition flip, not for continued performance races.
- Rule-following vs. rule-breaking marks the boundary. Whether engineers must break rules or can follow codified ones is a diagnostic signal for whether a market tier is overshot. EDA tools are disruption enablers, not just productivity aids.
- Fast fabs may be the future profit pool. VCE predicts value migrates from microprocessor design toward fast fabrication. Converting slow batch fabs to continuous-flow fabs is the superior strategic move over outsourcing to foundries.
- Autonomous organizations are the incumbent's best weapon. Intel's Celeron, Centrino, and StrongArm succeeded because each had separate processes and values -- the prescription for incumbents facing disruption.
- Watch for conservation of integration inversion. When a previously commoditized stage becomes the bottleneck, it becomes interdependent, proprietary, and profitable. The power shifts to whoever controls that stage.
Related References
- Core Framework -- disruption theory fundamentals, overshooting, basis-of-competition shifts
- Value Chain Evolution -- conservation of integration, modularity-interdependence cycle, profit pool migration
- RPV Theory -- resources, processes, values framework applied to incumbent constraints
- Strategic Choices -- autonomous organization prescription, co-option failure conditions
- Competitive Battles -- asymmetric motivation, cramming response pattern