How long it took the microprocessor to actually become a business
I like geeking out over this stuff, but I think it’s also a great illustration of a business principle which I unfortunately don’t know what to call yet.
As many geeks know, in 1971, Intel invented arguably the biggest technological breakthrough of the 20th century: the microprocessor. So, of course, they brought it to market and instantly sold millions of their new… nope, that didn’t happen. At first Intel didn’t sell microprocessors at all. Intel’s first CPU, the 4004, was produced under contract for Busicom, who wanted it for a desktop calculator. It was an entire year before Intel bothered selling them openly, and when they did it was largely ignored. There were some embedded applications, like traffic lights and industrial sensors, but as a consumer product it wasn’t really a thing.
(At this time, a guy named Tom Pittman built a computer based around the 4004, all by himself, but nobody noticed except a handful of hobbyists.
It’s a cool story though.
) The next year, in 1972, Intel invented the improved, 8-bit 8008 chip. But they were very determined not to be in the computer business. Computers were these big, clunking machines sold to big companies for huge amounts of money, and that was just too much of a pain; computers were best left to the computer makers. So they were very definitely not going to sell computers. (Eg. the 8008 only had 18 pins, and an Intel engineer spent nine months fighting management to get a computer-friendly 40 pins for the next-generation 8080.) Meanwhile, outside the world of industrial semiconductors, the 8008 was again mostly ignored.
In 1973, a French company used the 8008 to build a computer called the Micral. But that was a French product, marketed and sold in France, so in the US it was basically ignored too.
In 1974, the magazine Radio Electronics advertised schematics for a computer based on the 8008, called the Mark-8. You ordered the schematics and the circuit board, and then you had to buy hundreds of parts from a bunch of different suppliers, and assemble them all yourself. This got some interest, but only a few hundred people actually built it.
In 1975, MITS came out with a computer called the Altair, based on Intel’s newer 8080 processor. The Altair was generally sold in kit form: they gave you all the parts and an instruction book, but you still assembled it yourself. This got some more interest; thousands of people started buying Altairs. Other companies started selling spare parts, add-ons, software, etc.
In 1976, a company called IMSAI cloned the Altair, and sold a new-and-improved version which patched up some design flaws. This was pretty popular, and they were soon selling hundreds of them a month. More people started ordering computer kits pre-assembled, but you still had to buy the peripherals (basically everything but CPU and memory) as add-ons.
Then, in 1977, Radio Shack of all people finally started selling a ready-to-run CPU-based computer, everything pre-assembled, all the parts and peripherals included (at least the essential ones like monitor, keyboard, tape drive and BASIC). That model was called the TRS-80, and Radio Shack expected to sell maybe a few thousand of them. Instead they started flying off the shelves, tens of thousands were sold in the first month alone.
The usual story told here is that Moore’s Law waved a magic wand, and packed the computers with more and more transistors until people wanted to buy them. But that seems mistaken. The original, 1972-vintage 8008 chip had 3,500 transistors. Five years later, in 1977, the TRS-80 had a Zilog Z80 with 8,500 transistors, so there’s some Moore’s Law there. But ten years later in 1982, the most popular computer was the Atari 400/800, which was back at 3,500 transistors (MOS 6502B). Only by 1985 did IBM PC clones, with much larger transistor counts, start dominating the market. Microprocessors succeeded commercially when they came wrapped in packages that were easy to buy, and easy to use.