A short history of British computing and why it still matters.

Earlier this month, on 6 May 2026, a Cambridge-headquartered company reported that its fiscal-year revenue had reached approximately $4.92 billion, up 23% on the year before. The company's chip designs, not the chips themselves; the architectural blueprints from which other companies make the chips, reach, by the company's own estimate, one hundred per cent of the connected global population. Every smartphone on the planet, and increasingly every laptop, car, data centre, and connected sensor, contains silicon whose design genealogy traces back to a small group of engineers working in a barn in a Cambridgeshire village in 1985.^[1]

This piece is a short history of how that came to be the case, and why it still matters. The aim is not nostalgia. The British computing tradition is not, primarily, a story about old machines; it is a story about the structural reasons a particular country has continued to produce a disproportionate share of foundational digital infrastructure across seven decades of compute. The reasons are still operative. Several of them help explain the figures in the companion piece on the UK digital economy that opens this short series.

I  ·  The wartime substrate

The conventional opening line of any British computing history is Bletchley Park. It is the correct opening line. The codebreaking work conducted at Bletchley between 1939 and 1945, and the design of the Colossus machines that supported it, produced the first generation of programmable electronic computers and, more consequentially, the first generation of engineers who knew how to think about programmable electronic computers. The machines themselves were dismantled and the work classified for thirty years; the engineers, and the institutional habits of mind they had developed, were not.

One of the direct descendants was the Manchester Baby of 1948 and the Manchester Mark 1 of 1949, the first stored-program computers, designed at the University of Manchester. The same year, the Cambridge EDSAC ran its first program, designed by Maurice Wilkes. The Ferranti Mark 1, the first commercially-built computer in the world, was delivered by Ferranti of Manchester in 1951. The first computer used in business applications was the LEO (Lyons Electronic Office) which began running payroll calculations for the J. Lyons tea-shop company in 1951.

That sequence (Bletchley, Manchester, Cambridge, Ferranti, LEO) is not folkloric. It is the documentary record of the first decade of stored-program computing, and the United Kingdom is the principal locus on the map for every year of it. The country emerged from the Second World War with the deepest available concentration of programmable-computing expertise anywhere on the planet, and with a set of universities and industrial laboratories prepared to act on that inheritance.

II  ·  Acorn, the BBC Micro, and the literacy project

The line from the 1950s computing pioneers to the present day runs through Cambridge, and through one company in particular. Acorn Computers was founded in Cambridge in December 1978 by Hermann Hauser, Chris Curry and Andy Hopper.^[2] Hauser was an Austrian-born physics postgraduate at the University of Cambridge; Curry had worked alongside Clive Sinclair at Sinclair Radionics. The company began as a consultancy, one of its first paid projects was a microprocessor-based controller for a Welsh fruit-machine manufacturer, and rapidly developed its own line of microcomputers.

The pivotal moment came in 1981. The BBC, acting under a government initiative known as the Computer Literacy Project, was looking for a hardware partner to build the official computer for an accompanying television series intended to introduce British schoolchildren and adults to computing. The competition went to Acorn. The resulting machine, the BBC Microcomputer, became the standard computer of British schools for most of the 1980s.^[3]

Acorn Computers' rise to fame is intimately tied to the BBC and its Computer Literacy Project, which selected the Acorn Proton, later renamed as BBC Micro, as the official computer for its programmes. — UCL Institute for Innovation and Public Purpose

Two things about the BBC Micro deserve specific attention. The first is its commercial scale: approximately 1.5 million units were sold over the life of the model, a striking figure for a country of 56 million. The second is its educational consequence: a generation of British schoolchildren, including a meaningful fraction of the engineers who would build the next thirty years of British software, learned programming on a BBC Micro.^[4] The Computer Literacy Project is the single best example in modern industrial history of a public-service broadcaster, a government policy initiative, and a private engineering company combining to produce a stable demand-side foundation for an industry.

The Computer Literacy Project also illustrates a structural feature of the British system that is easy to miss. The BBC, by virtue of being a public-service broadcaster with universal reach, was able to seed the country with a single common piece of computing hardware across a single common educational experience. No other country with a comparable population had an institutional apparatus capable of doing the same thing. The cohort effect this produced has reverberated through British software for forty years.

III  ·  Inventing the chip in the chip in every phone

By the early 1980s, Acorn faced a problem of its own success. The BBC Micro had been built around the venerable 6502 microprocessor; planning its successor, the engineering team realised that none of the available 16-bit successors on the market (from Motorola, Intel, National Semiconductor) performed acceptably. The available chips were complex, power-hungry, and slow on the memory bandwidth that the team had concluded was the real determinant of a computer's performance.^[5]

The decision Acorn took next was, in retrospect, one of the most consequential single technical decisions of the late twentieth century. Two engineers, Sophie Wilson and Steve Furber, both Cambridge graduates, were set to designing the company's own 32-bit microprocessor, using the then-emerging Reduced Instruction Set Computer (RISC) approach. Hauser would later remark that he had given the team "two things no-one else would give their teams: no time, and no resources." The first prototype silicon, the ARM1, was completed in April 1985, in a former barn at Swaffham Bulbeck near Cambridge that Acorn was using as its design office.^[6]

"ARM" originally stood for Acorn RISC Machine. The processor first appeared in a commercial Acorn product (the Archimedes desktop) in 1987, with 25,000 transistors. The chip turned out to have a property the design team had not specifically optimised for: it consumed very little power. That property, which mattered little for desktop computing, would turn out to matter enormously when somebody eventually wanted to put a computer in a pocket.

In November 1990, Acorn spun the chip-design business out as a joint venture with Apple Computer (which needed a low-power processor for its Newton handheld project) and the Californian fabrication house VLSI Technology. The new company, Advanced RISC Machines, took twelve engineers and operated from a different Cambridgeshire barn. Sir Robin Saxby, recruited as its first chief executive in 1991, made two early strategic decisions that shaped the company's next thirty years: first, the new firm would not manufacture chips itself; second, it would license its architecture rather than sell chips, charging customers an upfront fee and a royalty on every chip shipped.^[7]

The licensing-and-royalty model was unusual. Almost every other semiconductor company of the era manufactured and sold its own silicon. The ARM model meant that the small company in a Cambridgeshire barn could let the world's largest foundries do the manufacturing, including, as it would turn out, the giant Asian foundries that did not yet exist in their current form, and capture an economic share of every chip those foundries produced from its designs. The model was the single most important commercial decision in ARM's history.

IV  ·  The web, and the second British contribution to global compute

Two streets in a single Cambridge college could fairly claim to have launched two of the four foundational technical layers of the modern internet: the processor architecture underneath, and the protocol layer on top. The second of these is the work of Sir Tim Berners-Lee, born in London in 1955, the son of two computer scientists who had themselves worked on the Ferranti Mark 1.^[8]

Berners-Lee took a physics degree at Oxford in 1976. After a period in industry, he joined CERN in Geneva as an independent contractor in 1980, returning permanently as a fellow in 1984. On 12 March 1989 he circulated a memo to his manager titled Information Management: A Proposal, suggesting that CERN should adopt a global hypertext system to manage the laboratory's documentation. The memo was annotated by his manager ("vague but exciting") and at first not acted upon.^[9]

Through the following year, Berners-Lee resubmitted the proposal in a revised form and, in late 1990, was given permission to develop it. By the end of December 1990 he had written the first web client (which he called WorldWideWeb), the first web server, the first specification of HTML, the first specification of HTTP, and the first specification of URLs. The first public web page went live on the open internet on 6 August 1991. On 30 April 1993, at Berners-Lee's urging, CERN placed the underlying code in the public domain, a decision without which the web as a universal protocol almost certainly would not exist.

The web is a British contribution to global compute in a slightly different sense than ARM. It was built at a Swiss laboratory, by an English physicist, using an American workstation, on protocols that were intentionally given away. But the foundational design choices, the decision to build on top of an open Internet rather than as a proprietary product; the decision to use a permissive document model rather than a controlled application model; the decision to release the spec rather than the runtime, reflect a particular set of values about openness and standards that map closely to the British academic computing tradition Berners-Lee was raised in.

V  ·  Why none of this is finished

It is tempting, when surveying a history this rich, to treat it as past. That would miss what 2026 in particular is making visible. The Cambridge company spun out of Acorn thirty-six years ago reported earlier this month that its trailing fiscal-year revenue had reached $4.92 billion, with royalties from its architecture flowing in from approximately twenty-two million developers and an effectively complete share of the world's smartphone market.^[10] The company announced in March 2026 the launch of its first own-designed silicon, the Arm AGI CPU, intended for the artificial-intelligence data centres that are now the principal growth surface of the global semiconductor industry, with more than $2 billion of customer demand booked across the following two fiscal years.

Cambridge is, in 2026, the headquarters of a publicly-traded technology company that touches the silicon inside more or less every smartphone, tablet, and increasingly laptop and data-centre server on the planet. The web protocols Tim Berners-Lee wrote in 1990 are now the substrate for an estimated 6 billion people online globally. The British academic computing tradition that produced both of these things continues to produce engineers, founders, and policy frameworks that compound rather than dilute the inheritance.

This matters for a practical reason. The structural advantages the United Kingdom enjoys in the global digital economy, the talent pool, the regulatory provenance, the time-zone bridge, the default-English-language reach, the depth of public-and-private institutional support for technical work, were not designed in the 2020s. They were built over seventy years by a sequence of institutions that happen to have remained, in their present form, recognisable to the engineers who started using them in the 1950s. The current government's stated ambition that the United Kingdom should be "one of the most attractive locations globally for digital and technology businesses" is, on the evidence of the previous seven decades, not a new policy. It is the continuation of a policy that has been in operation, more or less continuously, since the Manchester Baby first ran in 1948.

VI  ·  The thread

What links Bletchley to ARM to the World Wide Web is not a national myth. It is a particular set of institutional habits: deep university-industry coupling, public-service broadcasters and public bodies willing to seed demand for new technologies, a relatively permissive intellectual-property culture, an academic tradition that values openness of specification and standardisation as commercial advantages rather than threats, and a steady stream of engineers and founders raised on the products of the previous generation's work.

These habits are not romantic. They are the documented structural reasons that a mid-sized European country with no semiconductor manufacturing industry came to design the chip architecture inside every smartphone in the world. They are also the reasons that the United Kingdom's digital economy, the subject of the companion piece in this series, is now the third-largest in the world.

The next few years will test whether the institutional habits hold under pressure from the next layer of compute, the artificial-intelligence build-out, the data-centre energy question, the standards battle over agentic and autonomous systems. The history suggests they will. They are older than the technologies they have shaped, and they have outlasted every previous wave.

---

Continue reading

• How British companies built the world’s third-largest internet economy. The present-day economic surface of the seventy-year inheritance described above. UK digital economy ~$1.2 trillion, third-largest globally.
• Where British consumers actually spend their time online: a 2026 reading. The demand-side counterpart. What the median UK adult does with the four-and-a-half hours a day of online attention the infrastructure above makes possible.
• The two-billion-pound shop window. The same structural argument applied to a different sector. Why British television content exports past £2bn for the first time, and why "British content" has come to function as a category brand abroad.

---

Sources