The Isambard-AI supercomputer has officially entered service at the University of Bristol, becoming the most powerful AI supercomputing system built in the United Kingdom. The system is based on 5,448 NVIDIA GH200 Grace Hopper Superchips and delivers performance of 21 exaflops for AI workloads, placing it eleventh globally according to the latest TOP500 list cited in the source.
The system was unveiled during a ceremony held at the Bristol Centre for Supercomputing, attended by UK Secretary of State Peter Kyle, along with leaders from academia, industry and government, including Simon McIntosh-Smith, director of the Bristol Centre for Supercomputing, and Neil MacDonald, executive vice president and general manager of the server business at Hewlett Packard Enterprise.
Computing capacity directed at Britain’s priorities
Isambard-AI’s capacity is more than ten times that of the United Kingdom’s next-fastest supercomputer, and its computing capacity exceeds the combined capacity of the other British supercomputers. According to the source, the system ranked fourth globally in energy efficiency, making it a platform intended not only to expand AI capabilities but also to reduce the operating cost of high-performance infrastructure.
The project received government investment of £225 million and was implemented in cooperation between NVIDIA, HPE, the University of Bristol and other parties. Targeted areas of use include AI-powered drug discovery, advanced climate modelling, materials science, and the development of large language models that account for British languages and laws.
Early projects on the system
- Nightingale AI: A sovereign, multimodal foundational healthcare model trained on data from the UK National Health Service to support early diagnosis and personalised care.
- BritLLM: A British language-model project supporting languages such as Welsh alongside English, with the aim of promoting inclusion and improving the delivery of public services in healthcare, education and other areas.
- UCL Cancer Screening AI: A scalable system for detecting prostate cancer through magnetic resonance imaging scans, targeting faster diagnosis and personalised treatment.
- EIMCRYSTAL: A University of Liverpool project that uses AI to screen 68 million chemical formulations for more sustainable industrial materials that rely less on rare or toxic inputs.
- EgoAI: A University of Bristol project that analyses recordings from wearable cameras and other smart devices to help people perform tasks at home, with future potential to assist people living with dementia.
Rapid construction and resource efficiency
Work on the project began in June 2024, and the complete 5-megawatt system was ready for operation in June 2025. The project team attributes the speed of implementation to running activities in parallel, early procurement of critical hardware—particularly graphics processing units—and the use of prefabricated modular data halls that were shipped and assembled within days. The modular data centre was also established within 48 hours, while the project progressed from concept to operation in just under two years.
Isambard-AI uses direct liquid cooling from HPE, enabling 440 graphics processing units to be installed in a single rack, alongside hybrid cooling towers that allow more than 90% dry operation without water. The system operates entirely on carbon-free electricity, with the possibility of reusing waste heat, while its power usage effectiveness metric is below 1.1.
Access and usage management
The UK Department for Science, Innovation and Technology and UK Research and Innovation will manage access to the system, with a focus on aligning use with national priorities, providing opportunities for smaller institutions and start-ups, and supporting research and commercial innovation. The operation of Isambard-AI forms part of a broader British effort to build national AI infrastructure.