China has rapidly expanded its national and academic computing infrastructure to support data-intensive and AI-driven scientific discovery. This talk provides a 2026 update on the evolving landscape of supercomputing and AI infrastructure for science in China, covering recent system deployments, architectural trends, and representative application breakthroughs. We review the transition from traditional HPC systems toward integrated HPC+AI platforms, including the growth of large-scale GPU clusters, heterogeneous architectures, high-speed interconnects, and energy-efficient system design.
The talk highlights how these infrastructures are enabling new research paradigms in AI for Science (AI4S), spanning life sciences, climate modeling, materials discovery, and engineering simulation. We will discuss the convergence of simulation, data analytics, and foundation models, as well as the emergence of unified software stacks and workflow ecosystems that support large-scale, cross-disciplinary collaboration.
In addition, we examine key challenges and opportunities, including resource sharing across institutions, software and talent development, sustainability, and the role of national and regional computing centers in supporting open science. The talk concludes with an outlook on future directions for HPC and AI infrastructure in China and its role in accelerating global scientific innovation.
