Intel 4004 (1971): The First Commercial Microprocessor

The Intel 4004 was the first commercially available microprocessor, a complete central processing unit (CPU) on a single silicon chip. Announced by Intel in November 1971, it was designed for a Japanese desktop calculator but soon sold as a general-purpose part. The 4004 opened the way to cheap, programmable computing and, eventually, to personal computers and smartphones.

Background: calculators, chips and Hoff’s idea

By the late 1960s, integrated circuits were allowing more and more transistors to be packed onto a single piece of silicon. Electronic calculators were a booming market, especially in Japan, and manufacturers wanted custom chips to make them smaller and cheaper.

Intel was a young company, founded in 1968 by Robert Noyce and Gordon Moore, which concentrated on memory chips. In 1969 the Japanese calculator maker Busicom (Nippon Calculating Machine Corporation) asked Intel to produce a set of chips for a family of printing calculators. Busicom’s design called for around a dozen specialised chips.

Ted Hoff, an Intel engineer assigned to the project, thought Busicom’s design was too complex for Intel to build at a reasonable cost. In late 1969 he proposed a different approach: a simple, general-purpose 4-bit processor that would run the calculator’s functions as a program stored in read-only memory.

Working with Stanley Mazor, Hoff developed the overall architecture of what became the MCS-4 family. Masatoshi Shima, a Busicom engineer who had come to Intel to work on the project, helped develop the logic and the calculator software. Busicom accepted the Intel proposal in early 1970.

The concept was sound, but for months nobody at Intel was turning it into a real chip. The company was busy with memory products, and the project stalled.

Faggin builds the chip

In April 1970 Intel hired Federico Faggin, an Italian-born physicist and engineer who had developed silicon-gate MOS technology at Fairchild Semiconductor. Silicon gates allowed transistors to be faster, denser and more reliable than the older metal-gate design.

Faggin found the project well behind schedule, with Shima arriving to check on progress and discovering that little had been done. Faggin designed a new methodology for random logic using silicon gates, laid out the four chips in the family and solved many detailed engineering problems. He worked closely with Shima, who stayed at Intel for several months.

The first chips came off the production line in late 1970 but failed because a processing step had been missed. A corrected batch in early 1971 worked. Faggin etched his initials, “F.F.”, in a corner of the 4004 design.

The MCS-4 family

The 4004 was the processor in a set of four chips known as the MCS-4:

The 4004 itself held about 2,300 transistors on a die measuring roughly 3 by 4 millimetres, made with a 10-micrometre process. It ran at up to 740 kilohertz and could execute roughly 60,000 to 90,000 instructions per second, depending on the instruction mix. It came in a 16-pin package and had a set of 46 instructions.

From calculator part to product

Under the original contract, Busicom owned the rights to the chip design. In 1971 the calculator market was under heavy price pressure, and Busicom asked Intel for lower prices. Intel agreed in return for the right to sell the MCS-4 to other customers for uses other than calculators, refunding Busicom 60,000 US dollars of development costs.

Intel’s marketing team, initially doubtful, launched the chip set with an advertisement in Electronic News on 15 November 1971 announcing “a new era of integrated electronics”. The 4004 was soon used in cash registers, scales, traffic signals, test instruments and other embedded systems.

Competing claims

The 4004 was not the only early attempt at a computer on a chip. The Four-Phase Systems AL1 (1969) served as part of a processor in that company’s computers, and Texas Instruments developed single-chip calculator processors around 1971. The Garrett AiResearch MP944 chip set, built for the US Navy’s F-14 fighter by 1970, remained classified for decades.

Historians usually describe the 4004 as the first commercially available, general-purpose single-chip microprocessor, a narrower claim that is widely accepted.

Consequences and legacy

Intel quickly built more powerful processors. The 8-bit 8008 followed in 1972 and the 8080 in 1974; the 8080 powered the Altair 8800, one of the first personal computers. Faggin left Intel in 1974 to co-found Zilog, where Shima later joined him; Zilog produced the popular Z80 in 1976.

Intel’s later x86 processors, beginning with the 8086 in 1978, went into the IBM Personal Computer and its successors, making Intel the world’s leading chipmaker. Faggin, Hoff and Mazor later received the US National Medal of Technology and Innovation for their work on the microprocessor. The 4004’s 2,300 transistors compare with tens of billions on the largest chips today, a growth broadly in line with Moore’s law.

Why a microprocessor changed things

Before the 4004, most electronic products were built from logic designed for one purpose. If a manufacturer wanted a new feature, engineers had to design new circuits and often new chips. That was slow and expensive, and it made every product a separate engineering project.

A microprocessor reversed the logic. The hardware stayed the same, and the behaviour of the product was set by a program held in memory. A single chip set could run a calculator, a cash register or a laboratory instrument, depending on the software loaded into its ROM. Changing a product meant rewriting a program rather than redesigning hardware.

This shift also changed the economics of the chip industry. A general-purpose processor could be sold in large numbers to many customers, spreading its design costs and driving prices down. Memory makers such as Intel benefited twice, because every processor needed memory chips to go with it.

Myths and facts

The 4004 is sometimes credited with being far more powerful than it was. In practice it was slow even by the standards of 1971 minicomputers, and its 4-bit design suited decimal arithmetic in calculators rather than general computing. Its importance lay in the idea and in its low cost, not in raw speed.

It is also often said that one person “invented” the microprocessor. The 4004 was a team effort: Hoff and Mazor defined the architecture, Shima contributed the logic for Busicom’s needs, and Faggin designed and debugged the chip. Each of them has been honoured for the work.

Timeline

• 1969: Busicom asks Intel to make chips for a range of printing calculators.
• 1969, late: Ted Hoff and Stanley Mazor propose a general-purpose 4-bit processor.
• 1970, April: Federico Faggin joins Intel and takes charge of the chip design.
• 1971, early: The first fully working 4004 chips are produced.
• 1971: Intel wins the right to sell the chips beyond Busicom’s calculators.
• 1971, 15 November: Intel advertises the MCS-4 chip set in Electronic News.
• 1972, April: Intel introduces the 8-bit 8008 processor.

Why it matters

The Intel 4004 showed that a computer’s brain could be mass-produced on a single chip and programmed for almost any task. That idea underlies nearly every digital device in use today, from washing machines to laptops.