Microprocessor 1971 Step 19 License Second Source Architecture Industry Practice
# License the Design Itself: Second-Sourcing as a Business Practice This Series Has Not Yet Named
## 1. A Different Kind of Reuse Than Step 16's Cost Model Describes
Step 16 of this series describes how a reusable processor design amortizes its engineering cost across many customers who buy chips built from that one design. It does not describe a related but distinct business practice this series has not yet named: second-sourcing, in which Intel licensed the right to manufacture compatible versions of its own processor designs to other semiconductor companies, who then sold their own physically fabricated chips under their own names, built from licensed masks or specifications rather than bought from Intel's own production line.
This practice became standard for Intel's processor family descended from the 4004 lineage this series documents — the 8080, architecturally downstream of the 8008 named in Step 14, was second-sourced to multiple other manufacturers during the 1970s. The business reason was not generosity: large customers, particularly in military and industrial markets, routinely refused to commit to a single-source component whose supply could be disrupted by one company's own production problems, so Intel licensed competitors to manufacture compatible parts specifically to make its own architecture an acceptable, multi-supplier-backed choice.
## 2. Why a General-Purpose Architecture, Specifically, Made Second-Sourcing Possible
Second-sourcing an entire chip set only makes commercial sense for a design whose value lies in behavior reproducible by any competent fabricator working from the same specification — which is exactly the property Step 1's thesis establishes. A fixed-function chip's value is largely in its specific physical layout, since its behavior and its silicon are the same thing; licensing a second manufacturer to build "the same" fixed-function chip mostly just licenses a copy of one customer's own narrow design. A general-purpose processor's value is in the instruction set and architectural behavior Step 17 describes as the real design artifact — once a second manufacturer reproduces that behavior faithfully, customers can run the identical programs on either company's chip, making the two chips genuinely interchangeable in a way two differently licensed fixed-function designs rarely are.
Second-sourcing therefore depends on the same program-behavior-separated-from-silicon property this series opened with, applied at an industry level rather than a single-chip level: because the 4004's descendant architectures defined behavior as a specification independent of any one company's own fabrication line, that specification could be licensed and reproduced by a competitor without each company needing to agree on anything about layout, process, or mask geometry — only on the instruction-level behavior a program would observe.
## 3. What This Adds to the Series Without Changing the 4004 Itself
This step describes an industry practice applied to processor architectures broadly descended from the lineage this series traces, not a specific second-sourcing arrangement for the 4004 itself, which remained a less widely second-sourced part than its own 8080-generation successors. Nothing here revises any step's technical account of the 4004; it completes the business picture Step 11's contract renegotiation and Step 16's cost model began, showing that the architectural choice Step 1 made — behavior defined by a portable specification, not fixed by one company's own silicon — had consequences for industry structure that extended well past the original Busicom order, eventually making the processor architecture itself, not any single company's fabrication line, the thing buyers evaluated and chose.