Sustainable high performance computing in the UK: a position paper on public investment, accountability, and the path to net zero
File(s)
Accepted version
Author(s)
Type
Journal Article
Abstract
The UK invests significantly in High Performance Computing (HPC), yet has no coherent national strategy that addresses sustainability, accountability, or
long-term infrastructure planning. This paper proposes and promotes three key messages: first, that there turn on public investment cannot be assessed without
consistent public reporting on utilisation, research outputs, and cost-effectiveness for all publicly funded HPC systems; second, that publicly funded HPC systems should be subject to mandatory, standardised reporting on energy consumption, carbon emissions, within formation on what happens when a system is decommissioned; and third, that the conflation of AI and HPC in UK policy is a strategic error. Instead, HPC is the underpinning infrastructure, and AI is one of many types of applications using that infrastructure, alongside numerical modelling and simulation. Treating AI and HPC as separate, competing workloads risks under-serving the wider research community that depends upon that shared infrastructure, resulting in a weakening rather than strengthening of the UK’s national compute capability.
long-term infrastructure planning. This paper proposes and promotes three key messages: first, that there turn on public investment cannot be assessed without
consistent public reporting on utilisation, research outputs, and cost-effectiveness for all publicly funded HPC systems; second, that publicly funded HPC systems should be subject to mandatory, standardised reporting on energy consumption, carbon emissions, within formation on what happens when a system is decommissioned; and third, that the conflation of AI and HPC in UK policy is a strategic error. Instead, HPC is the underpinning infrastructure, and AI is one of many types of applications using that infrastructure, alongside numerical modelling and simulation. Treating AI and HPC as separate, competing workloads risks under-serving the wider research community that depends upon that shared infrastructure, resulting in a weakening rather than strengthening of the UK’s national compute capability.
Date Acceptance
2026-10-08
Citation
Journal of Supercomputing
ISSN
0920-8542
Publisher
Springer
Journal / Book Title
Journal of Supercomputing
Copyright Statement
Copyright This paper is embargoed until publication. Once published the author’s accepted manuscript will be made available under a CC-BY License in accordance with Imperial’s Research Publications Open Access policy (www.imperial.ac.uk/oa-policy).
License URL
Publication Status
Accepted