The UK government has unveiled plans to construct a cutting-edge supercomputer with a budget of £900 million (equivalent to US$1.1 billion). This supercomputer, named Isambard-3 in honor of the 19th-century British engineer Isambard Kingdom Brunel, is slated for installation later this year at the National Composites Centre in Bristol. Bristol University, a member of the GW4 group of research-intensive universities that includes Bath, Cardiff, and Exeter, is set to host both the supercomputer and the newly established AI Research Resource (AIRR, also referred to as Isambard-AI). The funding for these initiatives stems from the government’s AI investment announcement made earlier this year in March.
Isambard-3 is projected to be composed of thousands of state-of-the-art graphics processing units (GPUs), positioning it as one of Europe’s most powerful supercomputers, as stated by the Department for Science, Innovation, and Technology (DSIT) in a recent announcement. Michelle Donelan, the Secretary of State for Science, Innovation, and Technology, emphasized the significance of this investment in British innovation, anticipating that Isambard-AI would be instrumental in propelling scientific discovery and maintaining the UK’s leadership in AI development.
Regarding specific technical details such as the number of cores and processor types, the University of Bristol indicated that it is currently premature to confirm such specifications. Nevertheless, the university already provides multiple supercomputer clusters for research purposes, all running on the Linux operating system. Notably, its BlueCrystal Phase 4 system is optimized for large parallel tasks using Nvidia P100 GPUs and boasts 32 GPU nodes, each equipped with two cards. Additionally, it features 525 Lenovo compute nodes equipped with Intel E5-2680 v4 (Broadwell) CPUs.
In related news, the announcement of Isambard-AI precedes the UK’s upcoming global AI Safety Summit, scheduled for November 1 and 2 at Bletchley Park. Originally introduced by Prime Minister Rishi Sunak during a visit to Washington to meet with US President Joe Biden in June, the summit aims to convene government officials, AI companies, and researchers. Its primary objectives include formulating a framework for international collaboration on AI safety, identifying potential areas for collaborative AI safety research, and highlighting the constructive applications of AI technology in various domains.