About 1971, not 1967
The microprocessor, a complete central processor on a single chip, dates from about 1971, and not 1967. There was no microprocessor in 1967. What did exist was the technology that made one possible, and understanding what computer hardware looked like in 1967 helps explain why the single-chip processor took until the early 1970s.
What a processor was in 1967
In 1967 the central processing unit of a computer was not a chip. It was a collection of circuit boards, which were in turn populated by transistors, diodes and resistors, joined together by wiring in a cabinet. A mainframe’s processor could fill several cabinets. A minicomputer’s might fit in a single rack, but it was still dozens or hundreds of separate parts.
Vacuum tubes had mostly given way to transistors by the early 1960s. Some machines of the period used what IBM called Solid Logic Technology, small hybrid modules combining several components, in the System/360 range announced in 1964, rather than full monolithic integrated circuits. Memory was commonly magnetic core: grids of tiny ferrite rings, each storing one bit, threaded with wires. It was reliable and kept its contents without power, but it was costly to build and comparatively slow.
The integrated circuit arrives
The integrated circuit placed several components on a single piece of semiconductor. Jack Kilby at Texas Instruments demonstrated a working one in 1958, and Robert Noyce at Fairchild Semiconductor developed a practical approach using a flat, or planar, process in 1959. Early chips held only a handful of components.
During the 1960s they spread. Military and space programmes were large early customers, since small size and reliability mattered more than price. The Apollo Guidance Computer, built for the crewed lunar programme, reportedly made extensive use of integrated circuits on the road to Apollo 11. Commercial minicomputers followed, and by the later part of the decade integrated circuits were replacing discrete transistor boards in new designs.
These were small-scale and medium-scale chips. A typical chip might contain a few logic gates, or perhaps a small register, rather than a processor. Building a computer still meant wiring together many integrated circuit chips.
Minicomputers in 1967
The minicomputer was the leading edge of compact computing. Digital Equipment Corporation’s PDP-8, introduced in 1965, was the best-known example, and rivals such as Data General were soon to follow. These machines were small enough to be installed in a laboratory, a factory or a ship, and affordable by an organisation which could not buy a mainframe.
They used short word lengths and modest memory, and they were programmed for specific tasks, such as controlling an instrument or gathering data. In a sense they were the nearest ancestors of the microprocessor, because the early microprocessor designers often had minicomputers in mind. But the processor in a minicomputer was still made of many parts.
Chip density begins to climb
A crucial trend was visible by 1967. In April 1965, Gordon Moore, then at Fairchild, published an article observing that the number of components that could be placed economically on an integrated circuit had been roughly doubling every year, and he suggested that this might continue for at least a decade. He later revised the pace to about every two years, and the idea became known as Moore’s law. It was an observation of industry behaviour, not a law of nature.
The practical effect was that by the end of the 1960s, manufacturers could put hundreds, and then thousands, of transistors on one chip. This growth in density was the precondition for the microprocessor: a processor needed a few thousand transistors, and until around 1970 that was beyond what could be made reliably at an acceptable cost. Moore and Noyce left Fairchild to found Intel in 1968, focusing at first on semiconductor memory.
Calculators as a clue
One late-1960s project hints at the direction of travel. In 1967 a team at Texas Instruments, including Jack Kilby, built a prototype handheld electronic calculator using integrated circuits. It was a technical demonstration rather than a product, but it showed that chips could bring serious calculating power into a portable device. The calculator business would soon drive demand for custom chips, which is where the microprocessor story actually begins.
What 1967 had instead
In summary, 1967 had mainframes and minicomputers, core memory, hybrid and early integrated circuits, and a rapidly improving semiconductor industry. It did not have a microprocessor. The idea of fitting a processor on one chip became practical only as density rose, and the first chips of that kind appeared in the early 1970s.