Intel 4004: How a Calculator Project Produced a Commercial Microprocessor

Intel announced the 4004 in November 1971 as part of the MCS-4 chipset. The 4-bit CPU put the main functions of a small central processor onto one silicon chip, but it was not conceived as a stand-alone personal-computer brain. It emerged from a Japanese calculator company’s request for a family of custom chips and from Intel’s decision to replace special-purpose logic with a programmable architecture.

Busicom asked for calculator logic

In 1969, Nippon Calculating Machine Corporation—whose products used the Busicom name—approached Intel for chips implementing a calculator design. The proposal called for numerous specialized circuits. Intel engineer Marcian “Ted” Hoff saw that a simpler general-purpose processor, combined with program memory and input/output, could support a family of calculators with fewer chip designs.

Hoff and Stan Mazor developed the architecture and instruction concepts with Busicom engineer Masatoshi Shima. The eventual MCS-4 family included the 4004 CPU, 4001 ROM, 4002 RAM, and 4003 shift-register I/O. A system still required these companion chips; “CPU on one chip” did not mean every computer function was on one chip.

Federico Faggin turned architecture into silicon

Federico Faggin joined Intel in 1970 and led detailed logic and physical implementation using silicon-gate MOS technology, working closely with Shima. Layout had to fit the processor into the process, package, pin count, and schedule. Faggin’s initials appear in the chip artwork, a material trace of implementation credit often compressed in simplified accounts.

The 4004 had a 4-bit data path, a 12-bit program address, a stack and register set suited to decimal calculator work, and roughly 2,300 transistors according to Intel’s historical account. Its clock and instruction throughput were modest by later standards. The achievement was integration and programmability within early MOS manufacturing constraints.

Busicom initially held exclusive rights for calculator applications. As its finances and contract changed, Intel negotiated rights to sell the chipset to other markets, reportedly with exclusions. Intel then advertised the MCS-4 as a general microprogrammable computer set. Commercial positioning after the design was crucial: a reusable processor could address control, instrumentation, terminals, and other embedded applications.

“First microprocessor” depends on definition

The 4004 is commonly described as the first commercially available single-chip microprocessor. Other contemporary and earlier projects complicate broader claims. Garrett AiResearch’s multi-chip Central Air Data Computer for the F-14 was designed under secrecy; Four-Phase and others built MOS processor systems; Texas Instruments developed single-chip processor work and patents around the same era.

Definitions can emphasize single-chip integration, general programmability, completion date, commercial availability, or public documentation. The historically defensible claim is narrow: Intel shipped and marketed the 4004 as a commercial single-chip CPU within a documented chipset in 1971.

Intel followed with the 8-bit 8008 and 8080, whose address spaces and interfaces suited broader computing. The 4004 itself did not “invent the personal computer,” but it demonstrated that a programmable CPU could be a manufactured component purchased like memory or logic.

The calculator origin is the central lesson. A customer wanted fixed-function electronics; an architecture reduced the number of custom designs; process engineering made the architecture physical; contract rights made it a product. The microprocessor became transformative when those four pieces aligned.

Related:

Sources:

Busicom asked for calculator logic

Federico Faggin turned architecture into silicon

“First microprocessor” depends on definition

Comments