The first integrated circuit was wired by hand

On September 12, 1958, Jack Kilby applied power to a strip of germanium at Texas Instruments. An oscilloscope showed it oscillating at about 1.3 megahertz. The first working integrated circuit worked.

It was also connected with tiny gold wires.

Kilby’s idea was that a transistor, resistor and capacitor did not need to be made from different materials and assembled as separate packaged parts. They could be formed in one piece of semiconductor. For the demonstration, technicians cut germanium wafers into bars roughly 1.6 by 10 millimetres, alloyed metal tabs to them and used hand-applied black wax while etching the component regions.

The regions still had to become a circuit. Fine “flying wires” joined them. A Computer History Museum account notes that this was not a practical production technique; only a few dozen such devices were shipped for evaluation before deposited-metal connections became available.

This does not make Kilby’s circuit a counterfeit first. It clarifies what his demonstration proved. The components could share one semiconductor body and still perform different jobs. Integration of the wiring was a separate problem.

Robert Noyce at Fairchild soon proposed using the oxide layer of the planar manufacturing process as insulation, with metal conductors deposited on top. That turned interconnection from delicate assembly into something that could be patterned on the wafer. A later patent dispute even separated the contributions this way: as Kilby recounted, the court ruled that Noyce was first to teach metal connections adherent to the oxide.

The famous little bar therefore sits at an awkward but useful boundary. It condensed several kinds of component into one material while preserving the old-fashioned act of wiring them together. The first integrated circuit was integrated enough to establish the idea, but not enough to manufacture the future.