It’s like having 600,000 laptops – but what will the UK’s most powerful AI supercomputer actually do?

By Marc Ambasna-Jones 25 Jul 25 Reading time: 5 mins

“If you’re an astronomer, you need a telescope. If you’re a data scientist, you need compute,” said David Hogan, VP Enterprise EMEA at Nvidia, during last week’s official launch of Isambard-AI’s new supercomputer. And by that Hogan means massive compute.

Housed at NCC in the Bristol and Bath Science Park and built and run by the University of Bristol Centre for Supercomputing (BriCS), Isambard-AI is certainly massive. It is currently ranked 11th fastest in the world and fourth globally for energy efficiency. With the equivalent processing power of 600,000 high-end laptops running in parallel, Isambard-AI can, promises a BriCS statement, perform in one second what it would take the entire global population 80 years to achieve.

So what does that mean? At this scale and expense (it’s backed by £225m in government funding) it needs to deliver more than just raw performance. According to Professor Simon McIntosh-Smith, director of BriCS, the goal is to give researchers and businesses the ability to run AI models and simulations at a speed and scale that was previously out of reach.

For Nvidia’s Hogan, it’s about plugging a critical infrastructure gap. “The UK has some of the world’s best data scientists and AI start-ups,” he said. “What’s missing in the flywheel is compute.” He may have a point. The UK government has made high-end compute a national priority, and Isambard-AI plays a central role in that vision.

Working alongside the Dawn supercomputer at the University of Cambridge, Isambard forms part of the UK’s AI Research Resource (AIRR). Together, they are expected to deliver a combined compute capacity of 23 AI exaFLOPs. They are cornerstones of the UK government’s newly launched Compute Roadmap, which sets out a plan to scale the country’s sovereign compute capacity 20-fold by 2030 and position the UK as an “AI maker” capable of delivering breakthroughs in science, healthcare, and clean energy.

And just to prove the point, a wave of early use cases is already emerging. The Rosalind Franklin Institute is using Isambard-AI to analyse 3D placental scans to improve pregnancy outcomes and reduce stillbirths. Researchers at the University of Liverpool are applying it to sift through 68 million chemical combinations in search of greener industrial processes. UCL, meanwhile, is using the system to develop scalable AI tools for prostate cancer screening, based on MRI scans.

According to McIntosh-Smith, the life sciences sector is showing the strongest immediate demand, especially in areas like drug discovery and design, where AI models trained on large datasets can help reveal patterns that lead to new treatments. “They’ve been experimenting and now they’re ready to go,” he told BI Foresight. “These are the kinds of projects that could improve health and lifestyle outcomes within the next five years, not just in the distant future.”

Other early applications include engineering, product design, and advanced manufacturing, where AI is being used to optimise component design, wind turbine placement, and even additive manufacturing methods. “If this is your end goal, what’s the best way to build it?” said McIntosh-Smith. “That’s exactly the kind of problem AI is good at solving.”

More than 300 applicants responded to the first access call, with allocation decisions made by DSIT and UKRI. It’s enough demand to fill the machine for a year. With a shorter, more agile allocation model than traditional high-performance computing systems, Isambard-AI is designed to enable rapid experimentation, so this may act as a key criterion for applications being accepted.

“You can apply now for time that starts next week,” said McIntosh-Smith. “If that goes well, you can apply for more time or try something completely different. We’re able to turn things around much faster because AI doesn’t wait.”

The significance extends beyond performance metrics. With Bristol now expected to be one of three national supercomputing centres alongside Edinburgh and Cambridge, McIntosh-Smith sees Isambard-AI as a long-term catalyst for the region’s innovation economy, one built on deep roots in chip design, HPC, and advanced research.

McIntosh-Smith sees Isambard-AI as an anchor for regional and national innovation, with Bristol’s legacy in semiconductors and system design making it one of only a few places in the UK capable of delivering a supercomputing facility of this complexity. Bristol, Edinburgh, and Cambridge are expected to form a three-centre national network, with Bristol focused specifically on AI workloads.

“The reason this is happening in the south west is because we have such a strong ecosystem of high tech,” he said. “There aren’t many cities in the world where the companies actually developing the technology are local to where the supercomputer is being built but Bristol is one of them.”

Finally, he reflected on the speed of the build itself (just over one year for what normally takes four or five) as a lesson in what’s possible when assumptions are challenged.

“There’s a lot of prevailing wisdom that’s just wrong now,” he said, adding “We proved you can do something that normally takes three or four years in just over one.” McIntosh-Smith also noted that the centre has already welcomed visitors from around the world keen to learn how they built it so fast.

What happens next won’t be measured by benchmarks alone. The real test is whether researchers, start-ups and industry can make meaningful use of Isambard-AI to experiment, accelerate innovation, and bring forward ideas that previously would have struggled to leave the page. That, ultimately, is the true measure of progress.For Peter Kyle, Secretary of State for Science, Innovation and Technology, who launched Isambard-AI last week, this is about putting “the most powerful computer system in the country into the hands of British researchers and entrepreneurs” and propelling the UK to the “forefront of AI discovery.”

Whether it lives up to that ambition remains to be seen. But what it does do is help create a culture of no excuses. The UK now has the tools and that can only be a good thing.

Marc Ambasna-Jones / Editor-in-chief

Working as a technology journalist and writer since 1989, Marc has written for a wide range of titles on technology, business, education, politics and sustainability, with work appearing in The Guardian, The Register, New Statesman, Computer Weekly and many more.