Did you know that your smartphone has anything between 20 and 30 processors inside it designed by a British company? Arm Holdings doesn’t build the chips, but licenses its technology to virtually every semiconductor company on the planet. By the end of 2020, an estimated 160 billion processors had been built with its intellectual property (IP). They’re used in everything from your phone’s main processor to the WiFi chip, Bluetooth and even the battery charger.

To find out how this came to be and why it matters, let’s wind back to the late 1970s when computers were still big, expensive, remote devices. I was at school at the time and remember the faff of posting off punch cards to a distant computer for processing. They’d come back in the mail two weeks later – invariably with a syntax error on line one so you’d have to start again from scratch. You can imagine the magical feeling when my school bought its first computer – a Commodore PET – in 1979.

Arm Holdings’ chips are used in everything from your phone’s main processor to the WiFi chip, Bluetooth and even the battery charger

So when a device called the ZX80 came on the market in 1980 for just £99 – an order of magnitude less than the PET – I simply had to get my hands on one. Launched by the entrepreneur Clive Sinclair, it was the first of many home computers sold in Britain and was followed in 1981 by the BBC Micro. That machine was bought in huge numbers and even had an accompanying BBC TV show, which explained to an eager public how computers would change the way we live and work.

The tender to build the BBC Micro had been won by Acorn Computers, a company co-founded in Cambridge in 1978 by the inventor Chris Curry and the physicist Hermann Hauser, who had just completed his PhD at the Cavendish Laboratory. Acorn gave the BBC exactly what it wanted for the TV series, prototyping a machine in just five days. More than 1.5 million units were eventually sold, helping Acorn to reach a turnover of £100m by 1983. The company was floated on the UK stock exchange that year with a market capitalization of £135m, with Hauser’s and Curry’s stakes in it worth £64m and $51m respectively.

Going for growth

Acorn was, however, keen to build serious business machines that could take on the Apple II and the ubiquitous IBM PC, which had dominated the business market since its launch in 1981. Looking for a suitable microprocessor, Acorn approached Intel to use its 286 processor but didn’t like the package pin-out, so asked to buy the chip only. When Intel refused, Acorn decided to build its own chip.

Two Acorn engineers – Steve Furber and Sophie Wilson – decided to focus on “reduced instruction set computer” (RISC) processors, the ideas for which had been developed by John Hennessy at Stanford University and David Patterson at the University of California, Berkeley. Smaller and faster than conventional “complex instruction set computer” (CISC) processors developed by the likes of Intel, RISC processors needed more memory and a more complex compiler. But with memory costs falling quickly, Acorn decided to adopt the new approach and the Acorn RISC Machine (ARM) project began in October 1983.

Hermann Hauser has written an open letter to the British prime minister, strongly objecting to the acquisition

The first ARM microprocessor chip was tested in April 1985, requiring less than 5% of the power of a comparable CISC processor, which meant it could be put in a cheap plastic package without overheating a PC. Later, with Apple wanting to use ARM in its forthcoming Newton mobile device – but not wanting the chip produced by a competitor – Acorn decided to spin out ARM as a joint venture with Apple and another US firm (VLSI Technology) in 1990.

Renamed Advanced RISC Machines, it developed a business model in which other companies were allowed to build ARM chips by paying an upfront cost and a modest royalty fee per chip (usually a few per cent). Rivals could therefore manage their own supply chains, knowing they had access to advanced and widely used processor technology under reasonable terms. Attracted by the low power consumption, Nokia (then the market leader in mobile phones) used the ARM chip, as does Apple in its iPods, iPhones, iPads and, more recently, its computers too.

Where next?

In 1998 ARM floated on the London stock exchange and became the world’s most successful licensing firm, dealing even-handedly with every semiconductor manufacturer; it’s the Switzerland of semiconductors as Hauser puts it. Eventually, in 2016 Japanese telecoms firm SoftBank Group bought ARM for £23.4bn, seeking to become the leader in the “internet of things” and machine learning. There was some resistance to the sale from regulators, but ARM’s neutral “Switzerland” status was still retained, so the deal went ahead (with ARM rebranded Arm in 2017).

Read more

Hot time for hard disks: why magnetic-recording technology is still going strong

All seemed to be going well until the US chip maker Nvidia – one of Arm’s licensees – announced plans last year to buy Arm from SoftBank for $40bn. Most other licensees oppose the deal, not least because Nvidia is American, which means that Arm could be restricted from, say, Chinese companies if the US government so decreed. Hauser has already written an open letter to the British prime minister, strongly objecting to the acquisition.

The UK’s Competition and Markets Authority (CMA) is investigating the proposed takeover, focusing on the potential impact on competition in the UK and whether Arm, if sold, would have an incentive to “withdraw, raise prices or reduce the quality of its IP licensing services to Nvidia’s rivals”. I am expecting that the CMA will block the deal, given the huge opposition from almost all licensees and sovereign nations. But if the acquisition does go ahead, Arm’s business model may be damaged and its technology, I fear, may be turned into a potential political weapon. I hope they make the right decision.