Germany’s High-Performance Computing Center Stuttgart (HLRS) has highlighted collaboration with technology companies and research partners as central to the future of supercomputing, as the organisation explores new computing architectures beyond traditional CPU and GPU systems.
The centre said its newly established Future Computing Group will investigate emerging technologies, including quantum computing, neuromorphic systems and alternative accelerator architectures, to assess their potential for future HPC workloads. "Ultimately, HLRS collaborates with other HPC centers because it enables us to provide the best possible portfolio of resources and services — locally, nationally, and across Europe," commented Professor Michael Resch, Director of HLRS.
According to HLRS, the initiative comes as conventional performance-scaling approaches are approaching physical and economic limits, particularly as GPU-based supercomputing systems become increasingly expensive and energy-intensive.
Johannes Gebert, leader of the Future Computing Group, said the organisation is working closely with hardware developers to evaluate whether new architectures can effectively support scientific and industrial HPC applications. The group will examine how emerging hardware performs across different algorithms, workflows and software environments as computing systems become more heterogeneous.
One of the first collaborations under the programme involves Openchip, a Barcelona-headquartered company developing high-performance accelerator technologies based on the RISC-V architecture. HLRS said the partnership will assess how Openchip’s vector-processing approach could support large-scale HPC workflows while improving sustainability and reducing the complexity associated with porting applications to GPUs.
The announcement forms part of HLRS’s wider strategy to prepare for the next generation of supercomputing infrastructure. The centre is currently operating its new Hunter supercomputer, which entered service in 2025, while planning deployment of the exascale-class Herder system in 2027.
HLRS said future HPC systems will increasingly need to support converged workloads spanning simulation, artificial intelligence and large-scale data analytics. The organisation also emphasised the importance of improving energy efficiency as demand for AI and scientific computing infrastructure continues to grow.
