Twenty-five years that turned a sliver of germanium pressed against gold foil into the substrate of the modern world. A chronological reading of the inventions, founders, processes, and deployments that built the chip industry — December 1947 through November 1971.
In Room 1E455 of Building 1, Brattain and Bardeen demonstrate to Bell Labs management an amplifier built from a sliver of germanium and two gold contacts separated by a razor-blade slit, mounted on a triangular plastic wedge held in place with a paper clip. Power gain measured at about eighteen.
"This circuit was actually spoken over and… a distinct gain in speech level could be heard." — Brattain notebook, Dec 24, 1947Stung by Bardeen and Brattain's breakthrough, Shockley locks himself in a Chicago hotel room and over four weeks scribbles ~30 pages sketching a three-region doped sandwich — vastly more elegant and manufacturable than the point-contact device.
Bell unveils a row of small metallic cylinders containing two fine wires touching a pinhead-sized chip of germanium. The New York Times buries the story on page 46.
Sparks succeeds in producing a working junction transistor — moving Shockley's January 1948 sketches off paper and onto a workbench. Mass production becomes thinkable.
Under antitrust pressure, AT&T licenses the transistor patents broadly for a flat fee of $25,000. Among early licensees: a small Texas firm called Texas Instruments — and a newly incorporated Tokyo company that will later rename itself Sony.
Two years later, 97% of all U.S. hearing aids will be transistor-based.
First commercial transistor radio. RCA, Philco, Emerson all turn down the prototype. ~150,000 units sold; Consumer Reports pans the audio. By 1956 Regency walks away from the business.
Japan's first commercial transistor radio. Yields hover in the low single digits — engineers cull bins by ear. The opening Sony walks through after the American consumer-electronics establishment dismisses the technology.
Newly Nobel-laureled, Shockley returns to Palo Alto and recruits Moore, Noyce, Hoerni and others. Within a year, lie-detector tests, salary publication, and a pivot to a four-layer device nobody else thinks practical will drive them out.
The three meet in Stockholm, share one cordial dinner, and almost never appear in a room together for the rest of their lives.
Trying to fit a working radio inside a mortar shell, two Army engineers invert a microscope, shrink a Kodak photoresist mask through its optics, and project light onto germanium. The patent (filed Oct 31, 1957) becomes U.S. 2,890,395 — the printing press of the chip industry.
Within six months: the National Defense Education Act, ARPA, NASA, and a federal R&D line that climbs from ~$0.5B to over $10B. The chip industry's first customer has just been created.
Eight men sign ten one-dollar bills at the Clift Hotel. Fairchild Camera and Instrument provides $1.38M in working capital. Noyce, 29, becomes director of R&D. The defining American technology company of the next half-century is born.
Skipping the company-wide summer break (no vacation accrued, six weeks on the job), Kilby builds the first integrated circuit: a phase-shift oscillator with five components etched and diffused into a single piece of germanium. A clean sine wave traces itself across an oscilloscope. There is no record of cheering.
The eight ship a commercial silicon mesa transistor twice as fast as anything competitors offer. The price is roughly 30× the industry norm — and the Pentagon, behind IBM, pays it.
Hoerni leaves the silicon-dioxide oxide layer in place — heretical at the time — and produces a transistor flat against the wafer surface, sealed under glass. The leftover oxide is an electrical insulator; metal can be deposited and patterned over it. Photolithography now has a substrate it can love.
Noyce sees what Hoerni has done and writes out, on two notebook pages, the architecture of nearly every chip the world has built since: many planar transistors on one wafer, isolated by reverse-biased p-n junctions, wired by aluminum lines deposited over the protective oxide.
Filed July 30, 1959 — five months after Kilby. Cross-licensed by 1966.Five months ahead of Noyce's filing. The flying-wire drawings will, a decade later, be the hinge on which the 1969 federal court ruling turns.
Photolithography becomes the property of the U.S. Government. The Army awards Lathrop and Nall a $25,000 prize. Lathrop, characteristically, uses his share to buy a station wagon.
The first integrated circuit fabricated entirely in the planar style. Sold commercially as the 2N1613 in April. The architecture and the manufacturing process now match.
The first commercial integrated circuit family. Absurdly priced compared with discrete transistors, but — Noyce understands — the price has no fundamental floor. By 1965, ICs sell for less than the discrete parts they replace.
Eldon Hall's first AGC chip order. Within five years, NASA's reliability discipline and the Apollo volumes will pull the unit price below $30, then below $1.
Three months later — August 8, 1962 — Khrushchev signs the postanovlenie that founds Zelenograd, the planned Soviet Silicon Valley. It will run two generations behind, permanently.
The first major custom IC design program. By 1965 the Minuteman program is the largest single buyer of integrated circuits in the world.
Hall walks into a NASA meeting and proposes that the entire AGC be built from a single integrated circuit, ordered by the hundred thousand. Block I uses ~4,100 NOR gates per machine. Across the program, the AGC consumes ~1 million flat-pack ICs.
Bay Area assembly wage: $2.50/hour. Hong Kong wage: ten cents. Within a year the plant ships 120 million transistors. Within three years it employs 5,000 — more than Fairchild's entire California payroll. The Asian semiconductor supply chain begins.
No commercial market existed at any price a chip maker could economically produce. Apollo and Minuteman paid the prices that justified the yields.
Moore observes a doubling of transistors per chip roughly every 18–24 months. Soviet engineers will spend the next 25 years cloning Western chips two-to-three years late, never realizing the curve they are chasing has a velocity, not just a position.
Zelenograd's first integrated circuit — a recognizable knockoff of a TI part. The next is a Fairchild knockoff. By the early 1970s, "copy it" becomes formal Soviet policy.
A quadrant photodetector and a few ICs in a coffee-can nose cone. CEP collapses from ~400 ft (Rolling Thunder) to ~20–25 ft. The chip rewrites the mathematics of bombing.
Passed over for the Fairchild Camera presidency, Noyce calls Rock. Rock raises $2.5M in a single afternoon on a one-page document. The compensation structure that Fairchild pioneered — equity for everyone — comes with them.
The first plainly visible link of the supply chain that, by 2026, the United States cannot extricate itself from. Shepherd thought he was saving on labor. Li thought he was buying anchor weight. Both were correct, by an order of magnitude.
The Block II computer flies on every Apollo mission and on Skylab. Across the program, not one of its 2,800 in-flight integrated circuits is recorded as having failed.
The Fairchild diaspora gathers force — Sporck to National (1967), Hoerni & Last to Amelco (1961), Sanders to AMD, Valentine to Sequoia, Kleiner to KP. By decade's end, the family tree has begun to bear fruit.
By that point neither side cares: Fairchild and TI have been cross-licensing since 1966.
2,300 transistors at 10 µm. A general-purpose CPU on a single die. The germanium sliver from Murray Hill has, in 24 years, become a programmable computer the size of a fingernail.
After 800 sorties and 11 lost aircraft over seven years, a single strike package severs the Thanh Hoa Bridge. The chip industry's first geopolitical demonstration. The next half-century of warfare can now be drawn.