On Friday, April 26, 1985, a box of chips arrived at Acorn Computers’ lab in Cherry Hinton, on the eastern edge of Cambridge, England. Steve Ferber, a Cambridge physicist turned hardware engineer, opened it around 1:00 p. m. and lifted out the first silicon of a processor he and Sophie Wilson had spent 18 months designing with a team of about a dozen people.
He pushed the chip into a development board hooked to a BBC Micro. A line of text appeared: “Hello world. I am ARM. ” By 3:00 p.
m. , the chip was doing something it wasn’t strictly supposed to do yet — working at all on the first attempt. By late afternoon, Ferber picked up a multimeter and put the probes on the chip’s power pins. The needle sat at zero.
He checked the probes, the meter, and the chip. The chip was running. He traced the power rail back to the development board and found it wasn’t connected. Nothing was wired to the power pins at all.
The processor was executing instructions on parasitic leakage currents coming through the input and output lines, drawing a few milliwatts out of signals meant only for data. That accident on a workbench in an industrial unit on the outskirts of a small English city would have enormous consequences. The processor on that board was the ARM1, the first ancestor of a chip family that, 40 years later, has shipped more than 250 billion units into phones, watches, cars, hearing aids, and data centers. At that very moment, Acorn was nine months into a cash crisis so acute that an Italian firm was already drafting paperwork to buy a controlling stake.
This is the story of Acorn Computers. Before there is an acorn, there is a 13-year education in surviving the English electronics trade taken by Chris Curry. Born in 1946 in St. Neots, he left school with two A levels in math and physics and no interest in a degree.
In April 1966, at age 20, he went to work for Clive Sinclair at Sinclair Radionics in Cambridge. Curry spent 13 years there, working on hi-fi, pocket calculators, an electronic watch, and eventually a small electric vehicle project that would one day be sold as the Sinclair C5. In 1972, the firm shipped Britain’s first consumer calculator, The Executive, and Curry was one of the engineers who coaxed more battery life out of its seven-segment LED display. By 1976, Sinclair Radionics was in trouble, and the Labour government’s National Enterprise Board was brought in to stabilize it.
Clive Sinclair, who disliked being stabilized, encouraged Curry to leave quietly and run a sister company called Sinclair Instruments. Curry did the work — and started to notice he could do it on his own. Meanwhile, in a graduate common room half a mile away, an Austrian physicist named Herman Hauser was thinking about microprocessors. Born in Vienna in October 1948 and sent to Cambridge at 16 to learn English, he was back at King’s College by the mid-1970s, finishing a PhD at the Cavendish Laboratory on the mechanical activation of chemical reactions.
He finished in 1977, having already caught the microcomputer bug. At Selwyn College on the other side of the city, a mathematics undergraduate from Leeds named Sophie Wilson was spending the summer of 1977 designing a small system around a MOS Technology 6502 microprocessor to control the automatic feeding of dairy cows. The system worked. A friend of a friend told Herman Hauser about the undergraduate who built it, and Hauser wrote to her.
The Britain these three people were moving through had three television channels, no home computers in schools, and a prime minister warning that if the country did not learn to use computers, the economy would not survive the century. The company arrived on December 5, 1978. Herman Hauser and Chris Curry filed papers for a firm called Cambridge Processor Unit Limited, which did consulting work on microprocessor designs out of a small office on Market Hill in Cambridge. A third founder, engineer Andy Hopper, joined shortly after.
The company had no real capital, no real products, and a name nobody would remember. In March 1979, Hauser and Curry rebranded the trading side as Acorn Computers Limited. There is a story Hauser has repeated across the decades that the name was chosen to appear ahead of Apple in alphabetical trade listings. The documentary record on the exact timing is thinner than the story, and honest history should flag that.
The first Acorn product shipped in January 1979 as the Acorn System 1, a 6502-based microcomputer sold in kit form for around £80. It was a hobbyist’s board, not a consumer product, but it sold well enough to pay the office rent. In March 1980, Acorn shipped the Atom, its first full home machine with a built-in keyboard, a cassette slot, and a dialect of BASIC that Sophie Wilson had written herself. Wilson had joined the firm as its first programmer, working out of what she would later describe as a converted room above a bedding shop.
Turnover for the first full year was around £3,000. By the end of 1980, Acorn was quietly profitable, and Hauser was already preparing files for a successor machine internally called the Proton. The Proton was intended to be a 16-bit descendant of the Atom with a 6502 running at twice the clock speed of the hobby machines of the moment, hardware expandable in ways nothing else in Britain yet was. Nobody at Acorn had built one yet.
There were drawings, block diagrams, and chips that existed on paper, in Steve Ferber and Sophie Wilson’s heads, and in some index cards pinned to a corkboard. In the autumn of 1980, Herman Hauser had lunch in London with a producer from the BBC. The British Broadcasting Corporation was quietly planning a television series on microcomputers for the following spring and had decided it needed a British machine at the center of it. Hauser did not mention that the one Acorn had in mind did not yet exist in silicon.
The BBC’s Computer Literacy Project was a strange, almost specifically British idea — partly a television series, partly a government training program funded by the Manpower Services Commission, partly national guilt about falling behind the Americans and Japanese on silicon, and partly a straightforward BBC education mission. Its original plan was to use a British machine called the New Brain, developed inside Sinclair Radionics before being spun out. By late 1980, the New Brain was behind schedule, over budget, and visibly not going to make the broadcast date. The BBC opened the contract to other British firms.
Acorn was one of several candidates. The others included Sinclair, now trading as Sinclair Research, which believed it owned the market on moral grounds. On a Friday in early February 1981, a BBC delegation was due in Cambridge to see what Acorn had. Herman Hauser, by all accounts, had told the BBC that the Proton existed.
He had also told Steve Ferber and Sophie Wilson separately that the other one had already got most of it working. Neither had got any of it working. They had one week. Ferber laid out a wire-wrap prototype on his bench.
Wilson rewrote her Atom BASIC from the ground up as the richer structured language that would become BBC BASIC. The two worked through nights that spilled into days. The night before the demo, the machine still refused to boot. According to Wilson’s later recollection, Ferber traced the fault through the small hours and got the processor talking to its video circuitry at around 4:00 in the morning.
The BBC team arrived that afternoon. They saw a working computer. They handed Acorn the contract in February 1981. Across the city, Clive Sinclair took it personally.
He had already shipped the ZX-80 and would soon ship the ZX-81. He regarded the BBC Micro contract as something that should have been his by right, and he said so. Acorn, for its part, ran an advertisement comparing Sinclair’s build quality unfavorably to its own. The feud escalated through 1983 and into 1984.
By legend, it ended with Chris Curry and Clive Sinclair in a Cambridge pub called The Baron of Beef in August 1984, where a rolled-up newspaper is said to have made contact with somebody’s head. The story is repeated so often it has become part of British computing folklore. Even the pub’s own staff, when pressed, cannot entirely confirm it. The documented record is thinner than the legend, and honest history has to say so.
What is not disputed is that the Proton, now officially the BBC Microcomputer System, shipped on December 1, 1981, in two configurations. Model A cost £235 and came with 16 kilobytes of RAM. Model B cost £335 and came with 32. Inside the beige case was a 2 MHz 6502, a Tube expansion bus that nobody else in the industry had, and a BASIC with features no other 8-bit computer would ship for years.
The decisions that went into the BBC Micro look in retrospect like a series of quietly radical choices made by engineers who didn’t realize they were being radical. The BBC originally wanted a Zilog Z80, the processor at the heart of the Sinclair machines and most of the American CP/M business market. Acorn’s engineers refused. They had learned the 6502 from the Atom and thought it ran rings around the Z80 on anything that mattered.
The BBC wanted a BASIC compatible with the dialect it had just finished writing a teaching curriculum around. Sophie Wilson looked at that dialect and quietly rewrote it. BBC BASIC as it shipped had long, fully significant variable names, REPEAT… UNTIL loops, named procedures and functions with locally scoped variables, structured error handling, and an inline assembler for the 6502.
It was years ahead of Microsoft BASIC, which was what most of the rest of the world was running. Steve Ferber laid out the Model B around something he called the Tube — a high-speed expansion bus running at the full memory speed of the processor, designed to let the BBC Micro act as a glorified terminal and input/output machine for a completely different processor sitting on a sideboard. In 1981, this was a strange thing to design into a home computer. Nobody else was doing it.
Ferber was doing it because he already suspected the 6502 would run out of road and wanted a hardware architecture that would let Acorn drop any processor it liked into the machine without redesigning the main board. He did not yet know in 1981 that the processor Acorn would eventually plug into that Tube bus would be a processor Acorn would itself design. The BBC Micro as it shipped in 1981 was not really a product. It was scaffolding — the frame on which the next chip, the next operating system, and the next company could be built.
On launch day, Hauser, Curry, Wilson, and Ferber were too busy counting orders to notice. By January 1982, orders had outrun production. By the summer of 1982, the Department for Education and Science was paying 90% of the cost of a Model B for secondary schools under a national scheme to put a computer in every classroom. By 1983, Acorn’s turnover had gone from £3,000 in 1979 to £8.
6 million in the year to July. In September of that year, the company went public on the Unlisted Securities Market at a valuation of around £135 million. Across the lab in Cherry Hinton, in the quietest corner of Acorn’s engineering office, Sophie Wilson was writing an instruction set for a processor that did not yet exist, simulating it one instruction at a time in BBC BASIC on a BBC Micro. The British state in the early 1980s behaved briefly as if it believed in domestic computing.
Mrs. Thatcher’s government is not usually remembered as a sponsor of national technology projects, but the Department for Education and Science wrote the checks to put BBC Micros in schools, and the Manpower Services Commission wrote more checks to retrain adults in programming on the same machine. The combined effect was a national subsidy no American company had access to. In the 12 months to July 1979, Acorn turned over £3,000.
In the 12 months to July 1983, it turned over £8. 6 million. In September 1983, the company listed at a market capitalization of around £135 million. Herman Hauser personally walked out of that flotation worth around £64 million on paper.
Chris Curry walked out worth around £51 million on paper. The company had existed for less than five years. By 1983, British homes were buying computers at a higher rate than any other country in the world on a per-household basis. Sinclair was shipping the ZX Spectrum at £175.
Commodore was shipping the 64 for just under £200. The Spectrum sat on kitchen tables. The 64 sat on kitchen tables. The BBC Micro at £335 for the Model B sat in classrooms — and because it sat in classrooms, every British child who learned to type a program typed one on the Model B.
The home market was Sinclair’s. The classroom market was Acorn’s. For a short time, this looked like two companies carving up a nation. It was not.
Across the Atlantic, something was quietly reshaping the ground. IBM shipped the Personal Computer in August 1981, running an Intel 8088 at 4. 77 MHz and an operating system called MS-DOS. Within three years, the IBM PC and its compatibles would be the standard against which every business computer was measured.
By 1985, the home computer was already being redefined in American minds as a cheaper IBM PC. Britain’s lead in per-household home computing became, in the same year, Britain’s irrelevance to the global desktop market. Acorn was still thinking about British schools. The world was already thinking about Lotus 1-2-3.
Hauser and Curry could see the horizon. They could not change it from Cambridge. What they could do was build their own silicon, so that when the next transition came, they would not have to buy their processors from the Americans or the Japanese. In October 1983, three months after the IPO, Acorn gave Sophie Wilson and Steve Ferber a mandate: design a chip.
At the peak around 1985, somewhere close to 80% of British secondary schools had a BBC Micro on a trolley. Over the Model B’s full product life, which ran from 1981 into the late 1980s, Acorn shipped more than 1. 5 million units. Every state school in England got one, often several.
BBC BASIC became, for a decade, the first programming language a British child learned. The verdict of the users is written into the fact that working Model B units still turn up at every retro computing meeting in Britain and routinely sell for between £200 and £500 on auction sites, on the understanding that they still work, that the keyboard still has its distinctive dense click, and that they will still boot into a cursor reading “BBC Computer 32K. ”
Hold the Model B in your hands and the engineering choices are visible. The beige case is deeper than it needs to be because Ferber gave it room for a user port, a 1 MHz expansion bus, an Econet connector for the earliest classroom networks, a cassette interface, an analog port, a disk interface slot, and his own Tube bus.
For £335 in 1981 — around £1,200 in today’s money — a British family was buying a machine that would survive drops, spilled Ribena, and the next 14 years of educational curriculum changes. In August 1983, Acorn tried to carry that success down into the price range Sinclair owned. The Acorn Electron shipped on August 25, 1983, at a launch price of £199. It was a cost-reduced Model B with Sophie Wilson’s BBC BASIC in ROM, most of the custom hardware folded into a single Ferranti ULA chip, and a smaller case meant to sit on a bedroom desk.
The British press loved it on paper. By October, Acorn had over 150,000 confirmed orders. By October, Acorn could also manufacture only around 25,000 Electrons a month at Astec in Malaysia, because Ferranti’s yields on the ULA were much worse than expected. Acorn admitted publicly it would have only around 60,000 units in shops before Christmas.
In Oxford Street, only two W. H. Smith stores received stock. Shoppers queued.
Shoppers left empty-handed. Shoppers bought ZX Spectrums instead. Christmas 1983 was a quiet disaster disguised as a success — the shipments that did make it through sold well, which encouraged Acorn to assume the following Christmas would be bigger. Christmas 1984 did not come.
The home computer market in Britain softened through the autumn. Sinclair cut the Spectrum. Commodore cut the 64. Acorn was left with warehouses full of Electrons nobody was buying at £199.
By March 1985, the price was £129 and still falling. Across the whole life of the product, Acorn would ship somewhere between 200,000 and 250,000 Electrons, against an expectation that was multiples of that. And in a corner of the same engineering office, Sophie Wilson had written out the instruction set for a brand new processor on paper, simulated the whole thing in BBC BASIC on a BBC Micro, and handed the simulation to Steve Ferber to turn into silicon. The project was called the Acorn RISC Machine.
It was the first commercial reduced instruction set processor designed by anyone outside the big American research labs. The team was around a dozen people. The budget by the time the project finished was somewhere around £5 million. Wilson and Ferber had earlier in the year flown to Phoenix, Arizona, to visit the Western Design Center — a boutique 6502 shop run largely out of a suburban house — and watched a small team ship real processors.
They came back convinced that an Acorn-sized team could do the same thing. The ARM1 was their answer. The first silicon came back from VLSI Technology on April 26, 1985. At 1:00 p.
m. , it was in the box. At 3:00, it was on the development board. By the end of the afternoon, it was drawing current from the input and output pins because the power rail was not wired up.
This was the moment the character of the ARM processor family was decided for the next 40 years. ARM by accident was cheap enough on power to run on parasitic leakage. It would later be deliberately engineered to run on batteries. The accident proved it was possible.
They had, in the summer of 1985, invented their own future. They had, in the same summer, run out of money. Over the first six months of that year, Acorn’s share price fell. The warehouses still had Electrons.
The headcount, which peaked at around 480, was on its way down to around 270 by the middle of the year. On February 20, 1985, the Italian industrial group Olivetti took a 49. 3% stake in Acorn Computers in exchange for £10. 39 million.
A further round in July brought in another £4 million. Acorn’s creditors agreed to write off a further £7. 9 million in debts. The company was saved.
Olivetti was a serious industrial firm. It built typewriters and PCs out of Ivrea in northern Italy, and it was not interested in a British 8-bit educational manufacturer. It was interested in whatever Acorn had that was next. For a short time, it was interested in the ARM.
For the engineers inside the Cherry Hinton labs, the first year after Olivetti was a quiet relief. Research continued. Sophie Wilson refined the instruction set. Steve Ferber brought the ARM2 to silicon in 1986, running at 8 MHz, hitting around 4 million instructions per second, drawing less than a watt.
In June 1987, Acorn shipped the Archimedes. Three models launched at once — the A305, the A310, and the A440 — at prices from £799 up to £2,299. The Archimedes was the first mass-market RISC personal computer on Earth, many times faster than any Commodore Amiga or Atari ST at comparable price points. It ran an operating system called Arthur, soon replaced by RISC OS, with a graphical user interface and a window manager that did things the Amiga could not.
It was backward compatible with the BBC Micro by software emulation. By 1989, the A3000 variant captured more than a third of the British school computer market in a nine-month window. British educators saw the Archimedes as the natural successor to the BBC Micro and bought it. For everyone else, the Archimedes was impossible.
It was nearly twice the price of an IBM clone of comparable speed. It ran no business software anyone outside Britain cared about. It was not sold in American department stores. Microsoft did not write a word processor for it.
The company that could have changed the trajectory of personal computing in 1987 shipped a machine that only British schools, British software hobbyists, and a few European enthusiasts ever heard about. Acorn’s management under Olivetti decided to protect the educational margin rather than attack the general consumer market. One reading is that Olivetti’s board, sitting in Ivrea, declined to spend money on an American marketing push for a British operating system that would have had to compete with MS-DOS, Windows, and the Macintosh. The other reading is that Acorn’s own engineers understandably protected the only market in which they had sold anything at all.
Either way, the Archimedes stayed British. The ARM, however, did not. In 1990, Apple Computer in California began work on a handheld computer it called the Newton. It needed a low-power processor.
The processor Apple found was the ARM. In November 1990, Acorn, Apple, and VLSI Technology jointly incorporated a new company called Advanced RISC Machines Limited, pulling the ARM design team out of Acorn and into a separate entity. Apple invested around $3 million US for a founding stake. The design team, which had been working above a bedding shop a decade earlier, moved into a converted barn outside Cambridge and started selling its chip designs to anyone who would buy them.
For most of the 1990s, Acorn and ARM lived separate lives. ARM, the small spinout, licensed its processor designs to Texas Instruments, Nokia, and anyone else building a mobile telephone. By the middle of the decade, the first digital mobile phones were starting to run ARM cores. By 1998, the installed base of mobile ARM was in the tens of millions and climbing fast.
Acorn, the British home computer company that still owned a controlling share of ARM, was losing money in the computer business every quarter. The Archimedes line gave way to the RISC PC. In 1994, the RISC PC kept the British school market and the enthusiast market — and nothing else. In November 1996, Acorn started work on the next machine, a RISC PC 2 codenamed the Phoebe.
It was to use the StrongARM processor at its full potential, support PCI expansion, and drive a modern graphics card. It was shown at trade shows across 1997 and 1998. In May 1998, Acorn opened pre-orders at a discount price of £950 against a planned retail price of around £1,500. By the time the first silicon motherboards powered up on September 15, 1998, Acorn had somewhere between 150 and 300 Phoebe pre-orders in hand.
The first silicon had sound bugs. The system was unstable. Two days later, on September 17, 1998, Acorn’s chief executive, Stan Boland, walked into a staff meeting and announced that the workstation division was closed. The Phoebe was cancelled.
Around 40% of the staff was to be made redundant. Those pre-orders became some empty injection-moulded cases sold off to enthusiasts. Most of the remaining stock was later destroyed in a warehouse fire. The Acorn community came to call that day Black Thursday.
Outside in the share listings, ARM Holdings, Acorn’s spinout, floated on the London Stock Exchange and the NASDAQ during 1998 at a valuation that let Acorn book around £18 million from its shareholding across the year. In the first nine months of 1998, Acorn lost around £9 million on its computer business. The spinout it owned a chunk of was already more valuable than the parent company had ever been. In January 1999, Acorn was renamed Element 14 Limited, a holding structure for its remaining research assets.
On June 1, 1999, the Acorn Computers name was formally wound up. The brand was eventually licensed out and lived on for a few years on budget set-top boxes imported from overseas. The company that had held a market capitalization of around £135 million at its 1983 listing was erased in a Cambridge office over an afternoon. The company it spun out would, by 2023, be relisted on the NASDAQ at an initial public offering valuation of around $54.
5 billion US. What survives is, first of all, the hardware itself. BBC Micros still boot. There is a steady auction market in Model Bs and Masters, with working units routinely moving between £200 and £500 depending on condition.
There are retro computing fairs across Britain where owners bring their machines, plug them into period monitors, and run software they wrote at 13. There is an active RISC OS community that, over the quarter century since Acorn went down, has kept the operating system maintained and running on modern ARM hardware, including the Raspberry Pi. What survives, second, is the educational mission. The Raspberry Pi Foundation, founded in Cambridge in 2008 by a group of engineers who had grown up on the BBC Micro, builds a credit card-sized computer intended specifically to put programming back into British schools.
It uses an ARM core. It can run BBC BASIC. By the mid-2020s, it has shipped over 60 million units into schools, homes, and industrial controllers. The BBC itself, in 2016, handed out 1 million BBC micro:bit boards to Year 7 children.
The micro:bit is ARM-based. The echo is deliberate. What survives, third, is the chip. By April 2025, 40 years after the first ARM silicon came back from VLSI Technology, over 250 billion ARM cores have shipped.
ARM is in almost every smartphone sold anywhere in the world. ARM is in every Apple Silicon Mac from 2020 onwards. ARM is in Amazon’s Graviton data center processors, in Qualcomm’s basebands, in Microsoft’s Surface laptops, in cars, in hearing aids, in car keys. What survives, fourth, is the people.
Herman Hauser becomes one of Cambridge’s most prolific venture investors across the following 30 years, backing a long line of British technology startups and quietly building what would come to be called the Cambridge Cluster. Steve Ferber joined the University of Manchester in August 1990 as professor of computer engineering and spent the next three decades building a neural computing project called SpiNNaker — essentially a brain simulator built out of over a million ARM cores. Sophie Wilson stayed in Cambridge, moved to Broadcom, and continued to design processors, including the family of chips inside most residential cable modems in the United States. Return to the needle at zero.
On April 26, 1985, in an industrial unit in Cherry Hinton, Steve Ferber put the probes of a multimeter on the power pins of a chip he and Sophie Wilson had drawn on paper and watched the needle refuse to move. Nothing was wired. The chip was running anyway. In 1985, that reading meant one chip, one bench, one accident.
By 2025, it means more than a billion smartphones built every year — each one small and warm enough to sit in a pocket because the processor inside it barely drinks any power at all. The accident was the product. Acorn did not fail to see the future. Acorn designed the future, shipped it in a beige British home computer, and ran out of time before the world caught up.
The management textbooks will say the company missed the transition to the IBM-compatible PC. The engineers will say the company built the processor that outlived the IBM-compatible PC by 30 years. Both are true, and honest history holds both.