Interview with Interview with Sir Andrew Hopper

Andrew (Andy) Hopper was born to Polish parents in Poland in May 1953. His parents divorced and his mother married William Hopper, an Englishman, who adopted Andy. They came to live in the UK in 1964.

Andy attended Quintin School, a grammar school, in St John’s Wood. He was unable to sit the Eleven Plus when he arrived in the UK as he did not speak English; he spoke Polish and his second language was Russian. He says: “I was excluded from a whole bunch of schools because, not taking it (the eleven-plus) was the same as not getting it, and, that was one of the barriers to my education in this country.”

The school became a comprehensive school as Andy joined the sixth form. He says he enjoyed his school years and his favourite subjects were those involving physical, tangible things such as the physical sciences, physics and electronics.

Despite not performing in school at the same level he had when in Poland, Andy was determined to go to university. In 1971 he went to Swansea to read computer technology which comprised computer science and electronics, electrical engineering, alongside economics and accountancy. Andy says: “As an undergraduate, in your first year getting some underpinnings in how to do a balance sheet, and some vague idea of marginal costs, revenue, pricing and stuff like that, as a minor part of this computer technology course, was fantastic. In 1971 David Aspinall, who was from Manchester (UMIST) and had gone to Swansea for his first chair, had the foresight to put together a course which, I think was amazing, and, it was right on the money, both intellectually and financially.”

Andy excelled at university, enjoying both his studies and the accompanying student lifestyle and new environment. During the summer holidays Andy travelled and started his life-long love affair with travel. He drove himself to Tehran in Iran and made trips to Brazil and the Amazon.

Andy’s first introduction to computers was with an ICL machine, he explains: “In 1971 when I first arrived, the university had the ICL 1900 or whatever it was.  In other words, it was behind a big cabinet, and you couldn’t really get at it. Then fairly quickly in that first year they said, ‘Oh here is a bit of a board we’ve prototyped.  See if you can make the light flash,’ or whatever.  Today it’s called Raspberry Pi; at that time, it was called Intel 8008.”

In his third year, Andy ran a project to reprogram a four-processor piece of hardware which had been constructed as a research project by Professor Aspinall and his postdoc and PhD students.  He adds: “So I got a little bit of exposure to slightly more complex systems and parallel processing and so on. It wasn’t networking, but how to have them working together into the memory system and all this sort of stuff. As soon as the microprocessor comes along, they plonk it front of the students, me and others, and let us get our hands on it.  We get our hands dirty.  Get a good feel for it.  Ah, marvellous.”

In 1974, having gained a good degree, Andy decided to study for a PhD and sent applications to universities in the Alps hoping to combine his love of study with his love of skiing. However, he didn’t receive any replies and his tutor, Professor David Aspinall, suggested he should make an appointment to see David Wheeler at Cambridge. He was interviewed by Maurice Wilkes and Roger Needham and offered a place.

Andy’s degree had given him a good rounded experience of building hardware, software and applications, he explains: “I knew how to use a soldering iron, and I knew what the TI catalogue was with all the 7400 chips inside it.  At Cambridge they had just completed building the CAP machine and the Cambridge Ring was only just getting going, and I turned up as a person who says, ‘Oh I like a bit of hardware construction, design construction.’  So, they welcomed me…. what I said I’d like to do was in line with the lie of the land at the department, as it turned out.”

Maurice Wilkes was the head of department at the time, and David Wheeler became Andy’s supervisor. Andy was allocated a space in both the PhD students’ shared laboratory and the lab space with the technicians and engineers who had built the CAP and had been involved in EDSAC 2. Andy explains how it worked: “I started building little bits of hardware and they were very happy there was somebody up there. Equally, it was a little unusual for a PhD student to be sitting with that engineering group and support group, but that meant I got to know them quite well.  So later on, I helped out with the Cambridge Ring design, where those people were building it. David Wheeler had designed a lot of it, Maurice Wilkes had the conceptual thoughts about it, Roger Needham did the distributed system around it. I wore short trousers and helped out with little bits and pieces, but nevertheless, those little bits and pieces did help with the construction.  It was in the bosom of the engineering and technology technical team, and I could also do my PhD which was to design a VLSI version of the Cambridge Ring and other local area networks.”

In the 1970s there was a proliferation of microprocessors and microcomputers, plus peripherals which came with the challenge of how to link them together effectively without needing to use the local PTT or telephone company’s wires and exchanges. The Cambridge Ring was one such solution; a very early local area network. It was a slotted ring.  It featured tens of nodes on the ring and each node could generate a packet. That packet would be sent around the Ring to a recipient; the recipient would copy the packet, which would then make its way back to the source which would mark it empty.

 Andy explains: “It was a little bit towards the sort of synchronous behaviour that you have in PTT style networks which existed at that time and continued to exist later.  It had a certain fairness of access, so that the slots allowed arbitration prioritisation and fair sharing, and allocation of bandwidth. It was a great design, and, Maurice conceived it, and David Wheeler designed it.”

The Cambridge Ring was renowned for its robustness which Andy explains: “Because the department had built EDSAC 1, EDSAC 2, the CAP machine, it had a tradition of building things that stood the test of time, in other words, were useful to others and provided a computing platform of some kind.  Maurice Wilkes had visited a company called Hasler in Switzerland who had, I think, some telecoms background and had also built a register-insertion ring. There was innovation in the way the protocols worked, so it wasn’t robust in the PTT sense, but it still worked well for the purpose.”

At the same time in the US, Xerox, was developing an alternative solution called the Ethernet which worked differently. Andy explains: “At the time we thought we had a great design, it worked. My contribution was in the way the maintenance was done, all packets were marked with some parity bits for maintenance purposes, and the degradation could be localised and bounded. In other words, the performance was a little bit closer to the synchronous behaviour than in Ethernet in comparison.”

Andy says that through his PhD, he was able to make “one or two minor, well, not major, contributions to the design of the Cambridge Ring”. He spent a lot of time on a chip version (VLSI). He explains; “When I look back on it, there was a Government funded project for doing a chip version of the Cambridge Ring, because it seemed that this was in every way going to provide faster networks, good local area networks, good intellectual impact, financial, commercial impact.  But it paid for one person, me; literally one person., So the chip version of the Cambridge Ring was one person; while the chip version of the Ethernet was Intel, Xerox and DEC.  This, now looking back, is barking mad, because, we did OK, but it was, not resourced properly.”

 He continues: “But there were all sorts of corollaries which are extremely positive, essentially, this is the precursor to why ARM exists, why the chip line in Cambridge exists in many different ways. The real precursor of why there was expertise, starts with this particular project, even though the later things are of a different kind.”

In the battle of the two technologies; Ethernet and the Ring, the Ethernet finally won.

Following his PhD Andy started a career as a lecturer at the University of Cambridge, and then went on to develop a simultaneous career pathway working in industry. In this way he co-founded 13 companies, including: Qudos, Virata, IPV, ABL, Adaptive Broadband, CBL, RealVNC, Level 5, and Ubisense among others.

Andy explains: “I have always had a normal academic job, with normal load, normal everything.  The best technology transfer is when somebody does simultaneous career and then they just talk to themselves. I had the good fortune of simultaneously having an industrial and an academic career with PhD students; I had the space to do that.”

In 1978, after gaining his PhD and having become an assistant lecturer, Andy met Hermann Hauser at a disco in Darwin College. The pair decided to start Orbis which would later be absorbed by Acorn which would lay the foundation for ARM.

They set out to start selling the Cambridge Ring, Andy explains: “By then the Cambridge Ring was being sold by three companies.  Maurice would hand out the drawings, I think it was for £10,000, drawings as seen, non-exclusive, they are copies, get on with it.  Cash in hand, good luck, non-exclusivity.”

The company, which was selling the board version while working on the VLSI version, took on Mike Muller as their first employee. Mike stayed through the transitions and is still the CTO of ARM to this day (at the time of this interview).  The company’s first customer was what was then called ICI in Runcorn in Cheshire.  The mark up on the hardware was seventeen times.  Andy says: “We had an early adopter who wanted to show ICI how higher speed networks, local area networks, at ten megabits per second could be used in their business. We worked our guts out constructing this and delivering it and all the rest of it and at the same time, we told them the future technology pathway with a chip version.”

Andy attributes Hermann with the entrepreneurial drive that started the company while Andy was “running around trying to make this chip work or design it”. In 1979, Orbis was absorbed into CPU, Cambridge Processor Unit, which Herman Hauser and Chris Curry had started a few months earlier than Orbis.

Andy explains: “When CPU stopped being a consultancy doing fruit machine, digital replacements and upgrades and started getting into home computing ….. it became obvious that this networking business was important, in general, all the sort of networking distributed systems stuff is happening there.  I’m doing a particular bit, but many others are doing many other bits.  So, we decided to absorb Orbis into CPU.  Acorn was always the sort of trading brand name; CPU was an underlying company.”

Andy joined Hermann and Chris as a director and shareholder of CPU in return for his shareholding in Orbis. With Andy’s contribution, the company decided to do a networking system for low end computers, which Andy describes as “more of an Ethernet based system than the Cambridge Ring”.

The company, which was designing chips for the BBC Micro and later developed the RISC processor (the forerunner to the ARM microprocessor), worked collaboratively with the University as Andy explains: “When it was still the BBC Micro, the interface between the university department and the company was fantastic.    The whole capability to design chips, which were crucial to the BBC Micro, came from that pathway……We had learnt on the Cambridge Ring chips how to do VLSI on chips, and those learnings, and experience, that confidence, that toolkit, those people with the ability to do it, became available within Acorn.”

Andy goes on to recall how Acorn decided to develop the RISC processor with insight it gained from its close connection with the university. This relationship allowed it to spot, identify and test the potential through research into other projects around the world. He says: “Through my academic connections, I was an assistant lecturer, I investigated, got some people to send me the papers, and decided that this was actually something we could contemplate.”

Olivetti bought Acorn and Hermann Hauser became Vice-President of Research at Olivetti. In 1986, he invited Andy to head up the Olivetti Research Laboratory, which they co-founded as an independent company. It was 100 per cent owned by Olivetti but run entirely independently with a separate board and Andy as Managing Director. This enabled Andy and Hermann to have complete freedom in their projects; he likens it to Google’s DeepMind today.

Andy explains: “The freedom led us to innovate by using local talent, and using the local intellectual power, which was the very thing that was the reason for ARM starting in Acorn.” The company grew up to 50/60 people with Andy as director; a role that he maintained entirely separately from his academic role as a lecturer.

Andy says: “It permitted innovation in a way that’s not possible in a university situation for all kinds of reasons. It succeeded because Olivetti understood very well how to leverage the local culture.  They provided the framework which enabled it to happen.”

Andy and Herman proposed a new way of working that would allow and support spin-out companies based on the practical research and products the lab was producing. Olivetti agreed to take 20 per cent of the spin-out companies and Andy, in his private capacity, became a founder and found the venture money for the companies. The labs also used open source to keep alive projects that could not be spun-out, rather than lose their potential.

The biggest success was the design of the ATM network which was rooted in the design of the Cambridge Ring originally and led to the formation, in 1993, of the spin-out company called Advanced Telecommunications Modules Ltd (ATML).  Andy continues: “We span out a company (ATML) to commercialise ATM networks. That was a modest success, funded by American venture capital, we did a pivot as it’s called today.  Turned it into a DSL chip company; Virata, with a market cap at its peak on Nasdaq to five billion.” The company was acquired by Conexant Systems in 2004.

Other spin-outs included Telemedia Systems Limited (1995) which became IPV, and Adaptive Broadband (2000). Andy says: “It’s a remarkable thing. I’m being paid a salary as managing director of research of this lab, and yet I’m spinning out and becoming a founder and doing my bit for these companies.”

In 1985, Andy, along with Hermann Hauser and Professor Haroon Ahmed, Professor of Microelectronics at Cambridge, were involved in co-founding Qudos (Quick Design on Silicon). It did software for chip design and hardware prototype, using electron beam direct-write on the Ferranti gate arrays that were used in the BBC Micro and earlier in the Cambridge Ring.

Andy says the company was not as successful as it could have been. It did prototyping and built their own electron beam machines which Andy describes as “fantastic”. They became the dominant part of the business but because they were bespoke and could only be manufactured one at a time, they did not generate enough business growth. Andy says: “We should have concentrated on the CAD, and we would have been called Cadence, but we didn’t. We tried to do too much.” Qudos continued until a few years ago as an engineering centre having been become part of the Rutherford Labs, an e-beam side prototyping facility.

AT&T bought the lab from Olivetti and decided to shut it down making everyone redundant; this was just at the time the team was working on a project to develop a new broadband phone.

Exasperated, Andy set out to try to prevent this. He approached MPs, the House of Lords, he even found two interested buyers, Intel and BT, however, AT&T were adamant that the facility had to go. He describes the period as “the worst moments of his life”, he explains: “The Lab is going great guns, in fact it’s actually just about to miss the iPhone, because we did the iPhone before the iPhone, which is the world’s most profitable product ever, and, they’re shutting the Lab. I mean it’s nuts …. A bit of jinking and funking around and there would have been value there big time, because we were pretty experienced about how to do all this stuff by then, and the chances are if you’re experienced, you’ve got the talent, you’ve got the business method, things do happen. It’s not magic!”

Following the shut-down, Andy continued with his team to co-found companies based on projects that had been underway in the lab. He says: “RealVNC (2002), Ubisense Ltd (2002), and Level 5 Networks (2002, now Solarflare Inc), among others, are the ones I co-founded with the teams that were the actual teams in the Lab……. Those companies started because there was tremendous value on the table.”

In 2004, when Andy became Head of the Department of Computer Science and Technology, he decided to not participate in any company that was a spin out of the department as he explains: “I felt, there was a conflict between being head of an institution and participating in companies that are the fruits of that.”

Andy says that he chose to adopt a volume strategy which he describes as very unusual in a university system. He says: “A volume strategy aims to get as many companies started as possible and wishes them well and supports them in every possible way.  You don’t take any financial participation but get them going.” In order to achieve this, the department set up a business club called the Computer Lab Ring, a name suggested by Maurice Wilkes.  He adds: “This is different to the technology transfer offices; this is our own thing. This Ring refers to the networking group, but it’s a business club, there are mentors, people with huge amounts of money as investors, companies who have done well, less well, and so on.”

In order to join the club, members have to have been through the Computer Lab, they are given the opportunity to start their companies within the department with free office space and mentoring from Andy and others. Andy does not participate in their companies. He explains: “I’m very proud and pleased that this has worked extraordinarily well, because, there are now just over 260 companies. This is probably the biggest thing in Cambridge cluster source as a house of innovation.  Half of them are still going. If you include ARM, which backdates to that, eighteen per cent have been sold for a total of just over $40 billion, so this is heavy duty.”

He continues: “In my experience, which is similar to Hermann’s, we can’t predict which one’s going to be successful, so it’s no good picking winners.”

While he has not entirely ruled out more co-founding of more companies from his own research, he is, in the meantime, really enjoying his involvement with the Royal Society where he is a Fellow and Treasurer. He says: “I’m finding that the door is so wide open, being Treasurer is very interesting in its own right, but being able to be effective, not having electric fences or barriers around me, is very appealing.  So, I’m not slowing down, because I have found a situation where I can be effective in my own way, without having to do politics or all that sort of stuff.”

“I’m not a sort of person who has everything planned. I grab my opportunities mainly through enthusiasm. To some extent I learnt it. Hermann Hauser he’s one of my best buddies and business partner. He sometimes calls me his little brother. His enthusiasm infects me and has infected me, from when we were at Cambridge as PhD students.”

Andy says: “I don’t think I made any mistakes, but you can’t have your cake and eat it.  So, by combining academia with industry, you can’t have the top position on both sides. It just doesn’t work like that, because on the industrial side, unless you go with a company, you’re going to lose your shirt at some point as an academic.  So, I’ve lost my shirt more often than I would have done if I had gone with any of these companies where I would have been in a principal position in them to negotiate better.”

He also tells the tale of a very big missed opportunity, involving the iPhone. He explains: “The biggest miss is the iPhone. We designed something that was based on an iPAQ, which was a little HP hand-held with a Wi-Fi network, with a pen screen, where we prototyped, what today are called apps, and they literally look exactly the same, the layout’s the same. They were triggered by tapping. We called them web snacks, now they call them apps.  I remember trying to say to AT&T, ‘Look, we ought to set up a development unit in Cambridge, we’ve got something interesting here, not quite sure what it is,’ and so on, but it fell on deaf ears, so that was a miss, because the iPhone and its eco-system is the world’s most profitable product ever.”

Andy has over 6,000 hours as a pilot with his own plane and airstrip. He has flown to more than 700 different airports; he says: “I’m a flying nut. ……It’s adventure flying.”

On the subject of which university created the first computer, Andy says: “Well actually both did in their own way. Everybody’s claim on the first computer is true, not just ours, those two, there are others, and they’re all true because they’re all slightly different.”

In 1996 Andy was made a Fellow of the Royal Academy of Engineering

In 2005, Andy was awarded an Honorary Fellowship of Swansea University.

In 2006, he became a Fellow of the Royal Society

In 2007, he was awarded a CBE for services to the computer industry

In 2010, Andy was awarded an Honorary Degree from Queen’s University Belfast

He was President of the Institution of Engineering and Technology in 2012

He collected the Royal Academy of Engineering MacRobert Award in 2013 followed by the Royal Society Bakerian Medal in 2017

Interviewed by: Richard Sharpe on the 21st June 2018 at the Royal Society

Transcribed by: Susan Hutton

Abstracted by: Lynda Feeley