Read the latest edition online
The 50 hottest disruptive technology companies
The real Arm story from womb to world dominance – written by Dr Hermann Hauser, co-founder of Acorn Computers and its ‘love child’ Arm. Dr Hauser is a serial technology entrepreneur and globally influential venture capital investor and syndicate leader.
When we made the crazy decision at Acorn Computers to create our own microprocessor nobody, including the greatest optimists at Acorn (that was probably myself), predicted the phenomenal success Arm would become.
We did not intend to build our own processor. We had made a detailed study of all available 16-bit and 32-bit processors for our next Acorn computer design. We examined the Intel 80286, National 16032, Motorola 68000, among others. We built second processors cards with each of these processors to test as add-ons to the BBC Micro.
We felt that the Intel architecture could work but the pin-out of the bus architecture needed redesigning. We approached Intel about it so that we could have independent access to the data and address buses which we felt were essential to our design. Intel, with hindsight perfectly understandably, refused, so with no workable solution to the processor problem we decided to build one ourselves.
Although Acorn was then the most successful computer company in Britain growing to revenues of ~£100m in just five years the resources we could bring to bear on a new microprocessor architecture were miniscule compared to Intel, National Semiconductors or Motorola.
So why has the Arm architecture become many times more successful than all of these companies combined? Arm is outselling Intel roughly 20:1 in numbers every year!
The answer is to be found in the two advantages I managed to give our design team at the time which none of the others had:-
Consequently the design had to be really simple. We had the good fortune that John Hennessy at Stanford and David Patterson at Berkley had just invented a new architecture type called RISC for Reduced Instruction Set Computer.
Their published research showed that less is more; that reducing the number of instructions implemented in a processor actually improves its performance, reduces the die size and power consumption.
Surprisingly, the first implementation and commercialisation of this US RISC breakthrough happened in Cambridge.
Professor Andy Hopper, one of the Acorn directors at the time and later the head of the Computer Department at Cambridge University, had heard of the RISC research and brought the papers into the Acorn offices for Sophie, Stephen and myself to study.
It fitted our requirements like a glove and because it was so much smaller and simpler, the small team which we grew around our two architects was able to implement it.
It was unusual for a computer company to have its own silicon design team at that time, but I was obsessed with “designing on silicon” and gave talks about how there will only be two types of computer companies in the future – “those that know how to design on silicon and those that are dead.”
This had already served us rather well as Acorn’s clever design of silicon chips based on Ferranti ULAs (Uncommitted Logic Arrays) underpinned the success of the BBC Micro which beat the Apple II by more than a factor of 2x in speed and much more in graphics.
On April 26, 1985, the first Arm chip, arrived at Acorn. The acronym Arm was coined by Sophie Wilson as an abbreviation for Acorn RISC Machine.
Always the optimist, I bought two bottles of champagne in the unreasonable expectation that the chip would work first time. They can still be seen at the museum in the Arm headquarters.
Stephen plugged the chip into the carefully prepared circuit board and –nothing! However, it took only a few hours to discover the mistakes on the board; we tried again, and the screen lit up with the message: “Hello world, I am an ARM!”
We celebrated with the champagne as intended and Stephen decided it was time to measure the current that the chip was using to check the power consumption.
As he was trying to disconnect the power pin for measurement he discovered that it was disconnected already as this was another fault in the circuit board.
The chip was working without being connected to the power supply! This was the first indication that we had a power efficient design. (The explanation for this miracle was that the leakage current from other pins was enough to power up the chip and make it work).
The ARM then became the processor used in the Acorn Archimedes range of computers. Because it had a much better price/performance than the Intel range, Acorn managed to compete with IBM PCs for another 15 years –especially in its graphics capability due to the barrel shifter being included in the basic architecture.
Sophie argued that it would increase our graphics performance enormously; Stephen argued that it would take up far too much silicon area. So, on one of the many design walks to Cherry Hinton Hall Park, I cast my vote for the barrel shifter which was incorporated into the design. As it turned out that was one of my better decisions – although it was a close call.
Acorn’s fortunes took a dive when CD players took over from home computers as the most desired Christmas gift and the company had to be rescued by Olivetti.
Because of Acorn’s outstanding R & D team, I became Vice-President of Research for the entire Olivetti group, which was then the leading PC producer in Europe with revenues of $7 billion.
I reported to the Vice-Chairman, Elserino Piol, who also became chairman of the Acorn board on which I continued to serve. I convinced Elserino and the board that if we wanted to make a success of Arm we had to spin it out as nobody would want to buy a microprocessor from a competitor. This presaged the issues encountered by NVIDIA during its failed takeover bid for Arm many years later.
Elserino and I than tried for years to convince European companies to make Arm the European educational processor. We made good progress with Siemens and almost had a deal with Thomson CSF.
However, it was not until I met with Larry Tesler who ran the Newton group at Apple at the time, that we managed to get a deal. Larry was looking for a low power, high performance processor for the Newton pen computer and did not trust AT&T to support the Hobbit, their RISC processor which he was using at the time.
He argued that a small Cambridge company would keep supporting the Arm as it had no other option whereas AT&T would abandon their design if the sales did not materialise fast enough – which is what actually happened.
So we managed to spin out ARM in 1990 with £1.5m of money from Apple for 43 per cent of the equity, 43 per cent for Acorn, 5 per cent for VLSI technology – the first manufacturer of ARM chips – and the rest for the 12 Acorn engineers who became the founding team for ARM.
Apple eventually managed to sell its stake for $800m at a time when they really needed the cash. According to John Scully, CEO of Apple at the time, Apple would have failed if it had not been able to receive this cash injection at the time.
Here is John Sculley’s quote from the PC magazine from September 15, 2015: “But the irony is that we had to work with a man named Hermann Hauser, who was the creator of the Acorn computer in the UK – a professor at Cambridge University.
“He had developed a unique processor for the Acorn and we worked with him to modify it to work with a processor called ARM. We owned 43 percent of the ARM at Apple, because it was completely designed as the first Object Oriented programming language application on a low-powered microprocessor.
“Everything in the ARM was designed around the Newton. Newton was not successful but Newton actually made $800 million because Apple eventually sold the 43 per cent it owned in ARM, which, by the way, kept the doors open at Apple, just before Steve Jobs came back.
“It was one of the really important decisions that [Gil] Amelio [the last CEO before Steve Jobs returned] made, and it gave them the cash to buy NeXT. It’s interesting how you can connect the dots by lots of things that happened with a lot of innovation along the way. Things that you don’t often find in history books.”
We convinced Robin Saxby, a very experienced semiconductor executive whom I had met many years earlier when he was trying to sell Acorn the Motorola 68000, to become CEO.
The early years of ARM were not easy for him as he had only two main customers, Acorn and Apple. As the Newton was not a great success neither of his parent companies needed a large number of chips so ARM was living hand to mouth for a few years.
As VP of research for Olivetti I became a member of the Esprit advisory board, the first European Framework program for R & D. Convinced, as I still am, that Europe needs its own Microprocessor capability, I helped to persuade the board of the importance of an Open Microprocessor Initiative for Europe (OMI) and ARM was one of the companies that benefited from this European support during a difficult financial period.
The breakthrough came when Nokia was looking for a new processor and decided on the ARM for their mobile phones. This had two important consequences.
Nokia did not want to buy chips from a small Cambridge company so Robin changed the business model to target licensing companies like Texas Instruments which was a trusted supplier of chips to Nokia.
Nokia became the dominant mobile phone provider in the world with a 60 per cent market share until the arrival of the iPhone. Every other phone manufacturer followed suit as all new software appeared on the ARM first.
This is the reason why Intel with their Atom processor aimed at the mobile phone market never made any inroads into the ARM dominance.
This sequence of events has resulted in Arm having more than a 95 per cent market share in mobile phones through 500+ licensees worldwide.
Under Robin Saxby’s brilliant leadership ARM went public on NASDAQ and the London Stock Exchange in 1998. It became one of the all-time best performing investments on the London Stock Exchange and was ranked 22 in the FTSE 100 when SoftBank took it over for £24bn.
I was quite public about my objections to this takeover at the time but politicians did not understand the strategic significance of retaining the only globally relevant technology company in the UK and celebrated the sale as a sign that after the Brexit referendum the UK was still an attractive country for acquisitions.
As technology sovereignty has become the defining issue of this decade it has turned out to be probably the biggest industrial strategy blunder by the UK government.
I was re-assured by Masayoshi Son at the time that he would keep Arm’s licensing independence and invest in the expansion of Arm into other areas –specifically the Internet of things. I believed him and he kept his word. But this was not the reason why I objected to the latest, now aborted NVIDIA-Arm deal.
My rationale was firstly that there was no reason for the acquisition as Arm had £1bn of cash in the bank, an outstanding management team under Simon Segars and a great and growing market. Secondly, I was worried what would happen when SoftBank wanted to sell it on.
This is exactly what happened in 2020 when NVIDIA announced its intention to buy Arm from SoftBank. I was the first to point out that this would be a disaster for Cambridge, the UK and Europe as it would result in one of the 500 licensees owning Arm and therefore not paying royalties – a highly anti-competitive situation.
I always felt that NVIDIA would have a hard time to convince the US regulator (FTC), the UK regulator (CMA), the EU and China to clear the deal. Fortunately sense has prevailed and the deal was called off. So what next? Clearly the preferred solution would be for Arm to come home to the London Stock Exchange. SoftBank has stated its intention to list Arm on NASDAQ so the UK government should make an effort to correct its past mistake and corral City funds to facilitate an Arm listing on the LSE.
One way of encouraging such a move would be for the Government to invest £1bn in a London IPO and take a golden share as they have in Rolls-Royce, so a foreign takeover can be prevented in the future.
As the Government has spent half a billion to finance OneWeb as part of UK industrial strategy, one could argue that Arm is rather more strategic to the UK as it gives us the only technology bargaining chip – literally – should a future American or Chinese president use technology as a weapon again.
The prospects for Arm’s future are good. Because mobile phones are sold by the billions Arm has also become the processor of choice for low cost, power efficient, embedded control.
Although the main application processor of PCs is still Intel, there are many more Arms in every PC than Intel chips. There are Arms in most hard disk controllers, chips for WiFI, Bluetooth, USB, mouse and keyboard controllers among other devices.
Furthermore, Apple has just replaced the main application processor from Intel with the Arm-based M1 chip in their PC range – giving them a great performance and power consumption advantage.
The most important new sector for Arm, however, is the server market. After many years of development, the latest Arm architecture now produces the same performance in the server market as Intel but at a much lower power consumption.
As cooling data centres is a key priority this is becoming a major selling point. There is also a considerable commercial reason for switching from Intel to Arm for the server market: Intel’s gross margin is ~60 per cent while Arm’s royalty rates are typically in the single digit range.
If Amazon produces its own Arm-based design – as it has done with the Graviton 2 – and buys the chips directly from TSMC they get the most expensive component in their servers at half the price. This is why 50 per cent of the new Amazon rollout of servers is now Arm-based.
The same argument is true for Microsoft and all the other data centre providers. This can make Arm serious competition to Intel in the server market.
They may now wish they had allowed us to change the 286 pin-out, so that we would have never bothered to design the ARM in the first place!
ARM is an exceptional company in many ways including its management. In its 32 year history it has only had 3 CEOs, Robin Saxby, Warren East and Simon Segars. Each of them increased the value of ARM by $10bn. All three CEOs were UK educated engineers who must rank among the best technology CEOs in Europe.
I wish the new CEO Rene Haas every success and hope he can repeat the tradition of outstanding leadership of ARM for the next ten years.
Stay informed of the latest news and features