Picture the scene properly. A farming cooperative outside a village called Sárisáp, tractors parked in the yard, and somewhere in one of its outbuildings a row of workers hunched over kitchen tables, soldering irons in hand, hand-fitting capacitors onto circuit boards meant to become a home computer. Not a garage in California. A collective farm in central Hungary, in 1984, building the machine that would introduce a generation of Hungarian teenagers to programming.
It sounds like a joke. It wasn’t. It was the only way the thing was ever going to get built.
A research institute with no factory of its own
The Primo came out of the Computer Science and Automation Research Institute of the Hungarian Academy of Sciences, known by its Hungarian acronym MTA SZTAKI, where a small team of engineers had spent the early 1980s watching the rest of Europe fill up with home computers Hungary had no easy way to buy or build. Getting hold of Western semiconductors from behind the Iron Curtain was never simple, and a state institute wasn’t set up to run a production line even if the parts had been sitting on a shelf.
So the SZTAKI team did something faintly subversive for a planned economy. In 1983 they formed the Microkey Research, Development and Production Association, a joint venture stitched together from a state electronics distributor called Elektromodul and, improbably, an agricultural cooperative named Új Élet (“New Life”) based in Sárisáp. Development stayed with SZTAKI’s own commercial arm, Cosy. Assembly went to the farm.
It worked, in the sense that computers came out the other end. It worked less well in every other sense. Quality control on those early production runs reportedly happened on the same kitchen tables the boards were built on, and of the first hundred units to leave the line, sixty had to be sent back. Precision electronics and a working farm’s idea of a clean room turned out not to mix especially well.
Before any of that, the machine wasn’t even called Primo. Its engineers had been calling it Microter internally, a smash of “Microkey” and “computer” that nobody outside the building particularly loved, and its power supply had the equally functional working name Energoter. Both got quietly retired before launch in favour of Primo, Italian for “first”, which is either a nicely on-the-nose choice for the country’s first retail home computer or exactly the kind of thing an engineering team reaches for when the deadline has arrived and nobody has a better idea. Possibly both. The old name lingered on a few early boxes and warranty cards anyway, printed before the memo went round.
A computer built almost entirely out of not having chips
What SZTAKI’s engineers actually put inside the case was a lesson in doing more with less, because less was genuinely all there was. At its heart sat the U880, an unlicensed East German clone of the Zilog Z80 built by VEB Mikroelektronik in Erfurt, the same processor family powering half of Eastern Europe’s home computers, running here at a modest 2.5 MHz. There was no dedicated video chip. No sound chip either, beyond a single one-bit speaker Hungarian manuals sincerely called “the horn.” The Primo’s Z80 clone generated its own black-and-white picture, 256 by 192 pixels, entirely in software: the same resolution a ZX Spectrum owner already had on their screen, just without the colour.
Every one of those decisions was a chip somebody didn’t have to source, license or pay import duty on. It also meant the CPU was doing double duty as its own graphics card, which is a neat trick right up until you need every clock cycle for something else, at which point it’s simply the ceiling on what your machine can do.
The Primo shipped in three RAM tiers, badged in a way that still trips people up decades later. An A-32 wasn’t 32 kilobytes of RAM; it was 16 kilobytes of RAM plus the machine’s 16 kilobytes of ROM, added together to make the number on the box. An A-48 carried 32 kilobytes of real RAM, an A-64 carried 48. It’s the sort of marketing arithmetic that makes complete sense to whoever wrote the spec sheet and absolutely none to the customer standing in the shop doing the sum in their head.
The keyboard that gave people a nervous breakdown
The single most divisive decision on the whole machine came from Gábor Örley, tasked with designing a keyboard for a country that couldn’t reliably get its hands on mechanical switches. His answer had no moving parts at all: a flat, copper-etched capacitive panel built straight onto the motherboard, sealed under the case, that registered a keypress from the touch of a fingertip rather than the click of a switch.
On paper it was clever. Nothing to wear out, nothing to jam, spill-proof by design. In practice, typing on it meant pressing your fingers down hard enough to register against a plate that had no interest in telling you whether it had noticed. Hungarian owners came up with their own names for it fast: tapi pad, roughly “the tap-board”, and masszír pad, “the massage-board”, the kind of nickname that only gets invented by people who have spent real time being frustrated by something. Sensitivity drifted with the board’s calibration, too, so characters would occasionally repeat themselves with nobody’s finger anywhere near the key.
Series B fixed it the honest way, with real push-button switches, but only around a thousand of those ever got made, a small fraction of a production run that, by the time the line shut down in 1986, had reached somewhere between seven and nine thousand machines in total, depending on which archive’s count you trust.
The machine that won the contest and still lost the sale
There was a better Primo waiting in the wings, and it’s the part of this story with the sharpest sting in it. For the second national School Computer Tender, SZTAKI’s engineers built the Pro/Primo: a proper mechanical keyboard using magnets instead of springs, a colour graphics card capable of 32 shades from a 256-colour palette, and internals redesigned around fewer, denser RAM chips. It placed at the top of the judging in its category. Schools liked it. The tender panel liked it.
They bought the other one anyway.
Commodore, selling into Hungary through a distributor called Novotrade, put its C16 in front of the same schools at 8,775 forint. The Pro/Primo the judges had just ranked first cost 17,000, very nearly double, for a machine built by a small team in a country that couldn’t manufacture at Commodore’s volumes and had no interest in trying to compete on price against a company that could. Around fifty Pro/Primo prototypes were built. None were ever sold. The regular Primo, meanwhile, had been priced at roughly 10,000 forint against an average monthly wage of about 5,500, not quite double a month’s pay, but close enough that nobody needed convincing it was a stretch, especially set against a loaf of bread at under eight forint and a small washing machine at two thousand.
None of that was a failure of engineering. Nobody outsold Commodore’s factories in 1986 by being cleverer at the drawing board. It was a factory in the West against a farm cooperative in Sárisáp, and everyone in that room already knew which one could print machines faster than the other could solder them.
What’s left of it
The Primo never became the Hungarian ZX Spectrum its engineers might quietly have hoped for. What it did become, for a few thousand Hungarian teenagers who got one of those beige boxes for their birthday, was a first computer, the machine that taught them to type BASIC, argue with a keyboard that fought back and occasionally get lucky with a FORTH loader typed in from a magazine listing. A dedicated community still keeps that alive: DOS and Windows emulators, an open-source project called OS-Primo, and a modern 512-kilobyte multicart built by a hobbyist called Sándor Vass that loads a shelf’s worth of old tape games through a 3D-printed cartridge shell.
In 2024, that same community marked the fortieth anniversary of the machine’s launch with an exhibition in Hungary called “Primo 40: An Attempt to Conquer the Market.” The title does the honest thing and calls it exactly what it was. An attempt. Not a conquest. Just a genuinely clever computer, built by people with almost nothing to build it with, that lost to the one company in the world that had everything.
