Empirical agent-based modelling is a key use case for national HPC (high-performance computing) infrastructure. It generally means more complicated model structure, and larger numbers of runs for calibration, sensitivity and uncertainty analysis. Accessing HPC is important to empirical agent-based modellers for improving model quality, increasing the scale of simulation from local to national (and even global), and obtaining and analysing results in short timeframes for workshop settings. The UK's HPC landscape is changing over the coming years, moving from being primarily focused on use cases in the physical and environmental sciences to covering all research areas. However, the infrastructure is also shifting from primarily CPU-based to primarily GPU-based computing. Kay Yeung, from UKRI, will be sharing updates on UKRI’s vision of the national compute ecosystem and compute investments, and is keen to meet researchers in the social sciences to explore barriers to accessing local and national infrastructure. Though focused on the UK context, learning from experiences of social simulation researchers in other countries is also valuable.