AI in Bristol: A Short History, From the Transputer to Isambard-AI

Ask most people where British AI happens and they'll say London, Cambridge or Oxford. But AI in Bristol has a history that goes back nearly fifty years: chips designed for parallel computing in the late 70s, one of the UK's biggest robotics labs, a string of chip startups, and now the most powerful supercomputer in the country, sitting just outside the city.

I've lived in Bristol for years and worked in web development and SEO since 2010. I also have a first-class BSc in Artificial Intelligence, so the city's AI story is one I find genuinely interesting, not just good material for a blog. This is the short version: the companies, labs and people that turned Bristol into one of the best places in the UK to work on AI, and what that means if you run a business here today.

Before "AI" was a buzzword: Inmos and the transputer

The story starts in 1978, when Inmos was founded with government backing and set up its design team at Aztec West, on the northern edge of Bristol.

Inmos built the transputer, a microprocessor designed from the ground up for parallel computing: lots of small processors working on a problem at once, passing messages to each other. That sounds very familiar now. Training a modern AI model is exactly that kind of problem, spread across thousands of chips. In the 1980s it was well ahead of its time.

Inmos never made a profit, and it was sold to SGS-Thomson (now STMicroelectronics) in 1989. But it did something more lasting than any single product: it brought a generation of chip designers to Bristol, and many of them stayed. That talent pool is the root of almost everything else in this post.

HP Labs and the research years

In the early 1980s Hewlett-Packard opened a research lab in Bristol, which grew into a major HP Labs site at Stoke Gifford. Over the following decades it worked on security, trusted computing and the semantic web. HP Labs Bristol created Jena, a framework for working with linked data and knowledge graphs that's still used today.

Knowledge representation (how you get a computer to store facts and reason about them) was a huge part of AI long before neural networks took over. A lot of today's "AI with a knowledge base" ideas have roots in that kind of work.

Bristol Robotics Laboratory: AI with a body

Most AI lives on a server. Robotics is where it has to deal with the real, messy world.

The Bristol Robotics Laboratory grew out of an intelligent autonomous systems lab set up in 1992. In 2005 the University of the West of England (UWE) and the University of Bristol won joint funding to turn it into the BRL, and it was officially launched in its current home on UWE's Frenchay campus in 2012. It describes itself as the largest academic centre for multi-disciplinary robotics research in the UK.

The BRL works on everything from swarm robotics to assistive robots for healthcare, and on the ethics of autonomous systems. It's also why Bristol produces so many engineers who are comfortable with both hardware and software, a rare mix.

The University of Bristol and the AI talent pipeline

A tech cluster lives or dies on talent, and the University of Bristol has been a steady source of AI researchers and engineers for decades.

In 2019, UKRI funded the university's Centre for Doctoral Training in Interactive Artificial Intelligence, led by Professor Peter Flach, to train at least 50 PhD researchers in machine learning, intelligent systems and human-computer interaction, with responsible AI running through the whole programme. Those researchers are now working in Bristol's startups, in big tech and in research labs.

Put that next to UWE's robotics work and SETsquared (the university business incubator that has backed hundreds of startups) and you get a city that keeps producing people who can build AI, not just talk about it.

The chip makers: XMOS and Graphcore

Remember those Inmos engineers? Some of them went on to build the next generation.

XMOS was founded in 2005 as a University of Bristol spin-out, co-founded by David May, the transputer's chief architect. Its processors ended up in voice-enabled devices like Amazon's Alexa products, and more recently it has focused on running AI directly on small, low-power chips ("AI at the edge") rather than in the cloud.

Then came Graphcore. Founded in Bristol in 2016 by Nigel Toon and Simon Knowles, Graphcore designed the IPU (Intelligence Processing Unit), a chip built specifically for machine learning and pitched as an alternative to Nvidia's GPUs. At its peak it was valued at around £2 billion, one of the UK's most talked-about startups.

The story didn't go entirely to plan. Competing with Nvidia is brutal, and in 2024 SoftBank acquired Graphcore for far less than its peak valuation. But Graphcore kept its Bristol headquarters and its team, and it put Bristol firmly on the map as a place where serious AI hardware gets designed.

The University of Bristol also spun out Ultrahaptics in 2013. Its technology used ultrasound to let you "feel" things in mid-air. It went on to buy the US hand-tracking company Leap Motion in 2019 and rebranded as Ultraleap: a Bristol startup buying a Silicon Valley one, which doesn't happen every day.

Isambard-AI: the UK's most powerful supercomputer

If you need one fact to show how seriously to take AI in Bristol, it's this one.

In July 2025, Isambard-AI was officially launched by the University of Bristol. It's a £225 million national facility built by HPE with 5,448 Nvidia GH200 Grace Hopper Superchips, and it's the most powerful supercomputer in the UK. It's also one of the most energy-efficient in the world; its first phase came second on the Green500 list in 2024.

It was named after Isambard Kingdom Brunel, which is about as Bristol as it gets. Researchers and companies across the UK use it to train and run AI models that would be impossible on normal hardware. It's a long way from the transputer, but it's the same idea: lots of processors, working in parallel, designed in this corner of the country.

Bristol's tech scene right now

The infrastructure keeps evolving. The Engine Shed at Temple Meads, home to SETsquared Bristol and many of the city's startups for more than a decade, closes in December 2026, and SETsquared is moving into the University of Bristol's new Temple Quarter Enterprise Campus. Science Creates supports deep tech startups in Old Market, and the Bristol & Bath Science Park sits on the city's edge.

From where I sit, Bristol is massive for tech, and AI in Bristol is no longer a niche research topic. There are chip companies, robotics, creative and digital agencies, fintech, aerospace and a huge number of software teams. For anyone working in AI, whether you're a developer, a researcher or a business owner trying to work out what to do with it, there's a lot of opportunity here.

A quick timeline of AI in Bristol

What this means for businesses in Bristol

All this history is great, but most of the businesses using AI in Bristol aren't building chips or training giant models. They want to know something simpler: can AI save us time, and can we trust it?

My honest take after years of building web apps: sometimes AI is the right answer, and sometimes it isn't. Plenty of business processes are better handled by a clear set of rules or a decision tree. They're predictable, cheap and easy to audit. AI earns its place when the input is messy: emails written in a hundred different ways, documents that don't follow a template, customer questions you can't predict.

The problem is that a lot of AI tools are black boxes. You can't see why they did something, and you can't easily stop them doing the wrong thing. That's the part I wanted to fix.

Where Digital HQ fits in

Digital HQ is my studio here in Bristol, and in its current form it has only just launched. I've been building websites since 2010 and React web apps since 2019, and we build custom software and web applications that remove friction from everyday operations. Recent clients include Norwich Executive Travel and HaloTag (FX Moto).

The new part is AI agents you can actually see into. Digital HQ offers an AI agent API and a multi-tenant AI agents service where:

• You control the rules and instructions. Nothing is hidden. You can see what each agent has been told and change it.
• You connect it to your own systems by adding API routes, so the agent can look things up and take real actions in your business.
• Those actions are governed. Agents only do what you've allowed, so you get an intelligence layer on top of your business without handing over the keys.

And if a simple decision tree would do the job better, I'll tell you. The point is to fix the workflow, not to put AI on everything.

If you're a Bristol business (or anywhere in the UK) wondering where AI fits into how you work, get in touch. I'm always happy to talk it through, and you'll get a straight answer. If you want to keep up with the bigger picture, our AI industry changelog tracks the major developments as they happen.

FAQs

Is Bristol a good place for AI jobs?

Yes. Between the University of Bristol, UWE, the Bristol Robotics Laboratory, chip companies like Graphcore and XMOS, Isambard-AI and a large startup scene, Bristol has one of the strongest AI ecosystems in the UK outside London and Cambridge.

Where is Isambard-AI?

Isambard-AI is run by the University of Bristol and is located on the edge of the city. It launched in July 2025 and is the UK's most powerful supercomputer.

What's next for AI in Bristol?

More of the same, at a bigger scale. Isambard-AI gives UK researchers and companies serious compute on the city's doorstep, the universities keep producing AI talent, and new spaces like Temple Quarter keep startups close together. The bigger shift will be AI moving out of the labs and into everyday businesses.

Do small businesses in Bristol need AI?

Not always. Start with the problem, not the technology. If a process follows clear rules, simple automation is often cheaper and more reliable. AI is worth it where the work involves messy, unpredictable input, and it's best when you can see and control what it's doing.