Key Principle
Industries cycle between integration and modularity depending on where performance is "not good enough." The golden rule: integrate to improve what is not good enough; outsource what is more than good enough. When one layer of a value chain commoditizes, an adjacent layer becomes the new site of proprietary integration and attractive profits. This is the law of conservation of integration -- integration never disappears from an industry, it migrates.
Three tests determine whether a value-chain interface is ready for modular specialist entry:
- Specifiability -- can the interface parameters be clearly defined?
- Verifiability -- can exchanges across the interface be measured and confirmed?
- Predictability -- are interactions well understood with no unpredictable interdependencies?
All three must hold. If any test fails, specialist entry at that interface struggles and integration retains its advantage.
Why This Matters
Profits concentrate wherever the hardest unsolved problem sits in the value chain. As performance trajectories overshoot customer needs, that locus shifts -- and profits migrate with it. Firms that integrate at yesterday's hard problem become "thin integrators" assembling modular products, while firms that integrate at the new hard-problem interfaces capture disproportionate value.
This means organizational design is a moving target. Yesterday's integration advantage becomes tomorrow's rigidity penalty. The critical strategic question is not "should we integrate or modularize?" but "where is the current not-good-enough problem, and is it shifting?"
Good Examples
- Semiconductors (fabs to fabless): IDMs once integrated design and manufacturing because overall chip performance was not good enough. As design rules codified and manufacturing became modular, fabless firms (Xilinx, Tensilica) entered by outsourcing production to foundries (TSMC). TSMC then gained power, dictating design rules to chip designers and interface standards to equipment suppliers -- integration migrated from IDMs to foundries. (Ch. 7)
- Semiconductor manufacturing layers: Conservation of integration produces an alternating architecture: modular components, interdependent manufacturing steps, modular fab design/WIP inventory, interdependent microprocessor, modular PC. "Whatever needs to be optimized needs to be surrounded by something that is configurable." (Ch. 7)
- Telecom (transport to routers): AT&T integrated end-to-end telecom when functionality/reliability was not good enough. After the 1968 Carterfone decision, specialists entered customer premise equipment (spawning fax machines, modems, PBX systems). But specialists could not compete at AT&T's core electronic switches, where integration still managed intricate interdependencies. (Ch. 1)
- Dell and Wal-Mart: When PCs became modular and functionality overshot, Dell integrated across supply chain and customer interface (where speed and customization were not good enough) while outsourcing component design. Wal-Mart integrated similar supply-chain pieces for cost. Both captured value at the new hard-problem interfaces. (Introduction, Ch. 1)
- Credit scoring: Banking evolved from expert loan-officer judgment to codified rules (Fair, Isaac's scoring tool, 1956), enabling vertical disintegration -- separating loan origination, servicing, and sourcing into specialist functions. (Ch. 1)
Counterpoints
- Mistiming modularity: Modularizing too early sacrifices performance before the market is ready. CLECs after the 1996 Telecommunications Act failed because no clean modular interface existed into local telephony networks -- the three tests were not met. (Ch. 1)
- Mistiming integration: Integrating too long sacrifices speed and flexibility once the performance dimension is overshot. Vacuum tube manufacturers saw the transistor early and invested heavily, but tried to force it into their existing integrated value networks. None became major semiconductor players. (Ch. 7)
- Assuming integration disappears: "It is not accurate to characterize an industry as integrated or disintegrated" -- multiple forms of integration always coexist at different points. Specialists are integrated too, just across different interfaces. (Ch. 1)
- Commoditization risk for disruptors: Being modular and disruptive is necessary but not sufficient. Fabless firms risk margin collapse if design becomes undifferentiated: "Modular assemblers typically find it difficult to capture value when what they do is undifferentiated and replicable." Entrants must own an interdependent, hard-to-replicate stage. (Ch. 7)
Key Quotes
"Integrate to improve what is 'not good enough' and outsource what is 'more than good enough.'" (Introduction)
"When an interdependent system architecture is necessary to optimize performance at a stage of value added that is not good enough, the architecture of the product or service at the adjacent stage of value added must be modular and conformable." (Ch. 1)
"Companies make attractive money when they solve the hardest problems. The hardest technical problems mandate solutions that are tightly coupled integrated systems." (Ch. 1)
"Conservation of integration holds that whatever needs to be optimized needs to be surrounded by something that is configurable." (Ch. 7)
"The very thing that makes Intel great is the very thing that creates opportunities for the attacking firms." (Ch. 7)
"Conservation of integration means optimizing the links between each step in the process. This will require deoptimizing each process step." (Ch. 7)
Rules of Thumb
- Follow the "not good enough": Wherever performance falls short of customer needs, that is where integration creates value and profits concentrate.
- Watch for overshooting: When improvement outpaces customer needs (60% vs. 20% in semiconductors), the basis of competition flips from performance to convenience/customization/price.
- Apply all three modularity tests: Specifiability, verifiability, predictability. If any fails, specialist entry at that interface will struggle.
- Expect profit migration, not profit destruction: When one layer commoditizes, look to adjacent layers for the next locus of integration and attractive profits.
- Rules-based codification enables disruption: Incumbents "often unwittingly develop the rules that put their competition into business." (Ch. 1) Watch for codified standards as a leading indicator of modular entry.
- Deoptimize parts to optimize the whole: When end-to-end flow becomes not good enough, individual steps must become configurable rather than maximally performant -- simplicity beats cutting-edge at synchronized-flow stages. (Ch. 7)
Related References
- RPV theory (resources-processes-values) -- explains why incumbents cannot reprioritize even when they see the shift
- Disruption patterns (low-end and new-market) -- modularity often enables subsequent disruption by letting entrants assemble value chains in novel ways
- Motivation/ability framework -- nonmarket forces (regulation, standards) can accelerate or block modular entry
- Basis of competition shifts -- the demand-side trigger that causes value chain restructuring