A new supercomputer set to be one the most powerful in Europe is to be built in Bristol, in a move to drive pioneering AI research and innovation in the UK.
The UK government has confirmed the University of Bristol will host the new AI Research Resource (AIRR), which will serve as a national facility to help researchers maximise the potential of AI and support critical work into the potential and safe use of the technology.
The world-class AIRR cluster will vastly increase the UK’s compute capacity – essential to achieving the UK’s AI ambitions and securing its place as a world-leader in harnessing the rapidly developing technology. The cluster, which will be made up of thousands of state-of-the-art graphics processing units, or GPUs, will be able to train the large language models that are at the forefront of AI research and development today.
Compute refers to the systems assembled at scale to process complicated tasks, and is integral not just to the science and technology ecosystem but to the running of modern economies. This new national facility will help to underpin the UK’s next-generation compute infrastructure, in line with the recommendations of the independent Future of Compute Review.
The new AIRR will be dubbed Isambard-AI after Isambard Kingdom Brunel, the engineer whose groundbreaking creations – including Bristol’s famous Suspension Bridge – transformed Britain and revolutionised transport and construction.
Science, Innovation and Technology Secretary Michelle Donelan said:
We are backing the future of British innovation, investing in a world-leading AI Research Resource in Bristol that will catalyse scientific discovery and keep the UK at the forefront of AI development.
The Isambard-AI cluster will be one of the most powerful supercomputers in Europe, and will help industry experts and researchers harness the game-changing potential of AI, including through the mission-critical work of our Frontier AI Taskforce.
Plans for the AIRR were announced in March, backed by a £900 million investment to transform the UK’s computing capacity and establish a dedicated AI Research Resource.
Bristol already plays host to cutting-edge computing technology, with the Isambard 3 supercomputer due to be installed later this year to support research in AI and machine learning, while the University of Bristol is home to the UKRI Centre for Doctoral Training in Interactive Artificial intelligence. Both Isambard 3 and Isambard-AI will be based at the National Composites Centre, in collaboration with the GW4 group of universities – an alliance made up of the Universities of Bath, Bristol, Cardiff and Exeter.
Simon McIntosh-Smith, Professor of High Performance Computing at the University of Bristol and project lead, said:
We’re delighted to be chosen as the site to host the UK’s first ever Artificial Intelligence Research Resource.
Isambard-AI will be one of the world’s first, large-scale, open AI supercomputers, and builds on our expertise designing and operating cutting-edge computational facilities, such as the incoming Isambard 3.
Professor Phil Taylor, Pro Vice-Chancellor for Research and Enterprise at the University of Bristol, said:
AI is expected to be as important as the steam age, with ramifications across almost every area of academia and industry. Bristol’s proud to be at the forefront of this revolution.
To be selected to host a new national AI supercomputer speaks to the University’s cutting-edge research into AI and machine learning.
We have unique expertise in rapidly building and deploying large-scale research computing infrastructure and we’re excited to play an integral part in establishing the UK as an international hub for AI.
The UK will host the world’s first AI Safety Summit on 1 and 2 November, bringing together leading countries, technology organisations, academic and civil society to discuss the risks created or exacerbated by the most powerful AI systems, and how to address and mitigate them. The Summit will also look at how the benefits of safe AI can be unlocked to improve lives. Photo by mikemacmarketing, Wikimedia commons.