Microchip Invention: Kilby, Noyce, and the Integrated Circuit

On September 12, 1958, a Texas Instruments engineer named Jack Kilby carried a sliver of germanium the size of a paperclip into a room full of skeptical managers and showed them a working circuit. Five months later, 1,500 miles away in California, Robert Noyce sketched a different solution to the same problem in his lab notebook at Fairchild Semiconductor. Neither man set out to invent the device that would eventually power every phone, car, and satellite on Earth — they were trying to solve a wiring problem. This guide explains what that problem was, how each man solved it differently, why the two solutions had to merge before either one worked commercially, and how the resulting technology scaled from a handful of components to tens of billions on a single chip.

The wiring problem nobody could out-solder

By the mid-1950s, the transistor — invented at Bell Labs in 1947 — had already replaced the bulky, power-hungry vacuum tube in radios and early computers. Transistors were smaller, cooler, and more reliable. But engineers ran into a wall that had nothing to do with the transistor itself: connecting one to the next.

Every circuit needed transistors, resistors, and capacitors wired together by hand, one solder joint at a time. Military and aerospace engineers called this "the tyranny of numbers." A circuit with a few hundred components meant a few hundred hand-soldered connections, and every single joint was a potential point of failure. Double the number of components in a system and you didn't just double the assembly time, you multiplied the number of ways the whole thing could fail.

The U.S. Air Force and NASA's forerunner agencies were pushing hardest on this problem in the late 1950s, because guided missiles and spacecraft needed thousands of components crammed into a nose cone that could not be serviced once it launched. Committees studied "molecular electronics" and micro-modules as possible fixes. None produced a workable answer before two engineers, working independently and for different reasons, found one.

Jack Kilby had been at Texas Instruments for barely five weeks and, as a new hire with no accrued vacation, was one of the few engineers left in the building during the July 1958 mass vacation shutdown. Alone with unrestricted lab access, he used the quiet weeks to sketch the idea that became the integrated circuit. He demonstrated a working version to TI management on September 12, 1958. ([Computer History Museum](https://www.computerhistory.org/siliconengine/all-semiconductor-solid-circuit-is-demonstrated/))

How Kilby's germanium sliver actually worked

Kilby's insight was simple to state and hard to execute: if every component in a circuit — transistors, resistors, capacitors — could be made from the same semiconductor material, there was no reason they all had to be separate pieces wired together. They could be fabricated as regions on a single continuous block, eliminating most of the hand-wiring altogether.

His first working demonstration was a phase-shift oscillator built on a bar of germanium measuring 11.1 millimeters by 1.6 millimeters (about 7/16 inch by 1/16 inch), roughly the size of a paperclip. On that single sliver he combined one transistor, one capacitor, and the equivalent of three resistors. When TI engineers powered it up on September 12, 1958, the oscilloscope showed a clean sine wave. The "tyranny of numbers" had a working counter-example. (Computer History Museum)

There was a catch, and it was a serious one. Kilby's components still needed tiny gold wires bonded by hand between the different regions of the germanium block, because germanium alone did not provide a way to isolate and interconnect the components cleanly. It was an integrated circuit in principle, but it was still handmade in practice, which meant it inherited some of the same reliability and manufacturing problems it was supposed to solve. Think of it as proving a bridge design works by building a single, hand-fitted model span. The concept was sound, but nobody could mass-produce that bridge from the blueprint yet.

TI filed a patent application for Kilby's "solid circuit" in February 1959, and the company publicly unveiled the concept a month later. It was a genuine breakthrough, and it also happened to be missing the piece that would make it manufacturable at scale.

Why did Kilby's germanium circuit still need hand wiring?

Because germanium didn't offer a practical way to grow an insulating layer directly on the chip. Without that layer, Kilby couldn't lay down interconnects on top of the components; he had to bond fine gold wires between them by hand, one at a time, which limited how many components a single chip could realistically contain.

Noyce's planar answer, and why it won the manufacturing race

Robert Noyce came at the same problem from a different direction, and from a different company culture. He was one of the "traitorous eight" engineers who had walked out of Shockley Semiconductor Laboratory in 1957 over founder William Shockley's management style, and together they started Fairchild Semiconductor. (PBS American Experience)

The missing piece came from Noyce's colleague Jean Hoerni, who had developed the "planar process" — a way to build transistors with a flat, protected surface by growing a layer of silicon dioxide directly on top of the silicon. That oxide layer acted like a protective coat, sealing and stabilizing the sensitive junctions beneath it instead of leaving them exposed. Hoerni presented the latest version of the planar process to Noyce and a Fairchild patent attorney on January 14, 1959. Nine days later, Noyce wrote in his notebook that if you could grow that oxide layer over an entire wafer, you could then etch a pattern of metal wires directly on top of it, connecting the components underneath without a single hand-soldered joint. (EDN, ETHW Milestones)

This is the detail that separates a clever idea from an industry. Kilby had proven that multiple components could live on one piece of semiconductor. Noyce showed how to connect them without wires at all, using the same photographic etching techniques already used to pattern the components themselves. Once the wiring became just another layer printed onto the wafer, the whole process could be automated and repeated identically, thousands of times, on a single sheet of silicon.

Noyce filed his patent for a "Semiconductor Device-and-Lead Structure" on July 30, 1959. Fairchild engineer Jay Last's team built the first working planar integrated circuit the following year, in 1960. (Computer History Museum, History of Information)

What made the planar process different from Kilby's approach?

The planar process grew an insulating oxide layer on the silicon wafer and then etched metal interconnects directly onto that layer, so components never needed hand-wired gold leads. Kilby's original design still relied on wires bonded by hand between regions of germanium, which limited how densely and reliably it could be manufactured.

The patent fight that took eight years to settle

Two companies, two patents, and one invention meant a legal collision was inevitable. Texas Instruments filed for patent interference against Fairchild in 1962, each side arguing its engineer deserved primary credit and the royalty stream that came with it. (San Antonio Report)

The U.S. Patent Office ruled first, in 1964, and split the decision: Kilby got credit for the broader concept of building multiple components from one semiconductor block, while Noyce's planar interconnection method stood as a separate, later invention. Because each patent covered a piece the other side needed, neither company could manufacture integrated circuits at scale without a license from its rival. Courts later leaned toward Noyce's claim, but by the time appeals ran their course the practical outcome hadn't changed: nobody could build modern chips using only one man's patent. (San Antonio Report)

In 1966, Texas Instruments and Fairchild settled and cross-licensed their patents, with a payment flowing to Fairchild. The dispute never produced a single winner, and that turned out to be the healthiest possible outcome for the industry. A cross-license meant every semiconductor company that wanted to build chips could license both patents through the same two firms, instead of one company controlling the entire technology outright. Historians of technology have since converged on a two-name credit: Kilby proved the concept was possible, and Noyce made it manufacturable. Both are usually described jointly as co-inventors of the integrated circuit rather than one man simply beating the other to the patent office.

Robert Noyce died in 1990, ten years before the Nobel committee recognized the integrated circuit. Nobel Prizes are not awarded posthumously, so when the Royal Swedish Academy of Sciences finally honored the invention in 2000, only Kilby could receive it — even though the committee's own citation credits Kilby specifically "for his part in the invention of the integrated circuit," acknowledging it was a shared achievement.

Why the microchip changed the shape of every industry after it

The transistor had already made electronics smaller and more efficient. The integrated circuit changed the economics of building anything electronic at all, because it collapsed manufacturing, not just component size.

Before the integrated circuit, adding a feature to a device meant adding physical parts and physical solder joints, each one a new cost and a new failure risk. After it, adding a feature increasingly meant changing a pattern on a photographic mask, something that got cheaper and faster with every generation of equipment rather than more expensive. That is the single mechanism behind everything that followed: pocket calculators in the 1970s, home computers in the 1980s, the smartphone in your hand today. None of them required more skilled hands soldering more wires. They required a smaller, more precisely patterned wafer.

Fairchild engineer Gordon Moore, another of the "traitorous eight" and later Noyce's co-founder at Intel, put a number on that trend in 1965. Writing in Electronics magazine, Moore observed that the number of components that could be fit onto a chip at the lowest cost per component had roughly doubled every year since 1959, and he projected the pattern would hold for at least a decade. He revised the pace to roughly every two years in 1975. Over the following five decades, the rough doubling trend held often enough that it became known as Moore's Law, less a law of physics than an industry-wide target that chip designers, factory planners, and equipment makers organized themselves around. (Computer History Museum)

The scale of that compounding is easy to state and hard to feel. Kilby's original circuit held one transistor. Apple's M3 Max chip, built in 2023 on a 3-nanometer manufacturing process, packs roughly 92 billion transistors onto a single piece of silicon a few centimeters across. (Guru3d) That is not a difference of degree. It is the difference between a single hand-built prototype and a mass-manufactured object more complex than any machine built before it, produced by the millions every year.

What people get wrong about the "microchip inventor" story

The most common mistake is treating this as a simple priority race with one winner, the way popular science stories often frame the transistor (Bell Labs, 1947) or the light bulb (Edison, 1879). That framing does not survive contact with the patent record. Kilby solved a real problem with a real, working device five months before Noyce did anything on paper. But Kilby's solution could not be mass-produced the way it existed in September 1958. Noyce's planar process, which came later, was the missing manufacturing method, and without it the integrated circuit would likely have stayed a laboratory curiosity, the same fate that met earlier "molecular electronics" proposals from the mid-1950s.

A second common error is assuming the two men were rivals who never spoke. In reality, both worked for decades afterward in the same small, overlapping world of American semiconductor engineering. Kilby stayed at Texas Instruments and later helped develop the handheld electronic calculator, one of the first consumer products built around his invention. Noyce co-founded Intel with Moore in 1968 and became one of the central figures credited with defining the culture of Silicon Valley itself. Their companies fought over royalties. The men themselves are described in most historical accounts as cordial, not bitter.

A third misconception is thinking "microchip" and "transistor" are two words for the same invention. They are sequential, not interchangeable. The transistor is a single switch. The integrated circuit is a method for building many transistors, along with resistors and capacitors, onto one piece of material and wiring them together without solder. You need the transistor to exist first. You need the integrated circuit to make millions of transistors practical to build and connect at once.

How the microchip actually gets built today

Modern chip manufacturing is a direct descendant of Noyce's planar process, scaled up by six decades of refinement rather than replaced by something fundamentally different. Here is roughly how the core steps compare, from Kilby's 1958 prototype to a contemporary chip fab.

Step	Kilby's 1958 prototype	Modern planar-descended process
Base material	Single bar of germanium	Silicon wafer, typically 300 mm (12 in) diameter
Component isolation	None; components close together on raw material	Silicon dioxide layer grown across the wafer surface
Interconnects	Hand-bonded gold wires	Photolithographically etched metal layers, patterned in dozens of stacked steps
Components per chip	5 (1 transistor, 1 capacitor, 3 resistors)	Tens of billions on advanced processor dies
Feature size	Millimeter scale	Nanometer scale (commercial "3 nm"-class processes as of the mid-2020s)
Production method	Hand-built, one at a time	Automated batch fabrication, hundreds of identical dies per wafer

Table: How chip fabrication evolved from Kilby's hand-built prototype to modern planar-process manufacturing.

The throughline across every row is Noyce's core idea: pattern the wiring instead of hand-placing it. Every shrink in feature size and every jump in transistor count since 1960 has come from finding ways to print that pattern more precisely, not from abandoning the method itself.

The takeaway

The integrated circuit is a rare case where crediting "the inventor" honestly requires naming two people who never worked in the same lab. Kilby proved that combining components on one piece of semiconductor was possible; Noyce showed how to wire them without solder, which is the part that actually made mass production possible. The patent office split the credit, the companies fought for eight years, and the industry that emerged from their settlement went on to double chip complexity on a schedule regular enough to earn its own name. Every device built on a chip today, from a hearing aid to a data center GPU, is running on the manufacturing logic Noyce patented in 1959, executed at a scale neither man could have specified in his own patent application.

---

Read next:

• History of Computing: From the Abacus to Artificial Intelligence
• How the Transistor Changed Everything
• The Invention of the Internet

Frequently Asked Questions

Q: Who actually invented the microchip, Kilby or Noyce? A: Both are credited as co-inventors. Kilby built the first working integrated circuit at Texas Instruments in September 1958, proving multiple components could share one piece of semiconductor. Noyce, at Fairchild Semiconductor, patented the planar process in 1959 that let those components be wired together without hand-soldered leads, which made mass production possible.

Q: Why did Jack Kilby win the Nobel Prize and not Robert Noyce? A: Noyce died in 1990, and the Nobel Prize is never awarded posthumously. When the Royal Swedish Academy of Sciences recognized the integrated circuit in 2000, Kilby was the only living co-inventor eligible to receive it, even though the citation credits him only "for his part in" the invention.

Q: What was the "tyranny of numbers" that the microchip solved? A: It was an engineering term from the 1950s for the problem of hand-wiring growing numbers of individual transistors, resistors, and capacitors. As circuits grew more complex, the number of solder joints grew with them, and each joint was a potential failure point that made large circuits unreliable to build and maintain.

Q: What material was Kilby's first integrated circuit made from? A: Germanium. His September 1958 prototype was built on a bar of germanium measuring about 11.1 by 1.6 millimeters, containing one transistor, one capacitor, and the equivalent of three resistors, connected by hand-bonded gold wires.

Q: How is the planar process different from Kilby's original design? A: The planar process, developed by Fairchild's Jean Hoerni and applied to integrated circuits by Noyce, grows a protective silicon dioxide layer across the wafer and etches metal interconnects on top of it. That eliminated the need for hand-wired connections between components, which is what made the integrated circuit manufacturable at scale.

Q: What happened in the Texas Instruments vs. Fairchild patent dispute? A: TI filed for patent interference against Fairchild in 1962. The Patent Office split the decision in 1964, giving each company a patent the other needed a license for. The two companies cross-licensed their patents in 1966, which let the broader semiconductor industry build on both inventions instead of one company controlling the technology outright.

Q: What is Moore's Law and how does it relate to the microchip's invention? A: Moore's Law is Gordon Moore's 1965 observation that the number of components that could be fit onto a chip at minimum cost was roughly doubling every year, later revised to about every two years. Moore worked alongside Noyce at Fairchild and later co-founded Intel with him; the "law" describes how far the manufacturing technique the two invented could be pushed.

Q: How many transistors are on a modern computer chip compared to Kilby's original? A: Kilby's 1958 prototype had one transistor. A modern high-end processor, such as Apple's M3 Max chip built on a 3-nanometer manufacturing process in 2023, packs roughly 92 billion transistors onto a single piece of silicon a few centimeters across.

Q: Did the transistor and the integrated circuit get invented by the same people? A: No. The transistor was invented at Bell Labs in 1947 by John Bardeen, Walter Brattain, and William Shockley. The integrated circuit came over a decade later, built on top of the transistor, and is credited to Kilby and Noyce working independently at different companies.

Q: What was the first commercial product to use an integrated circuit? A: Early integrated circuits were adopted first by military and aerospace programs, including guidance computers, because of the reliability and space savings they offered over hand-wired circuits. Texas Instruments, Kilby's employer, later used the technology in the first handheld electronic calculator in the early 1970s.

Sources

• Computer History Museum: All-Semiconductor "Solid Circuit" is Demonstrated (1958)
• Computer History Museum: First Planar Integrated Circuit is Fabricated (1960)
• Computer History Museum: Practical Monolithic Integrated Circuit Concept Patented (1959)
• Computer History Museum: Moore's Law Predicts the Future of Integrated Circuits (1965)
• EDN: Noyce Conceives Planar IC, January 23, 1959
• Engineering and Technology History Wiki: Semiconductor Planar Process and Integrated Circuit, 1959
• History of Information: Robert Noyce Invents the First Practical Monolithic Integrated Circuit
• Nobel Prize: The Nobel Prize in Physics 2000
• PBS American Experience: Early Silicon Valley Timeline
• San Antonio Report: How a Texas Instruments Engineer's Problem-Solving Skills Led to the Creation of the Integrated Circuit
• Guru3d: Apple Introduces M3, M3 Pro, and M3 Max Processors on TSMC 3nm