PyTorch Conference China 2026: Advancing the Open Source AI Stack
Exploring the future of AI through open-source collaboration and innovation.
1Introduction to PyTorch Conference China 2026
The PyTorch Conference China 2026 was held in Shanghai from September 8 to 9, alongside KubeCon + CloudNativeCon and OpenInfra Summit. The event featured technical discussions on models, frameworks, distributed training, inference, hardware, and cloud-native infrastructure. The theme 'Open Source for the AI Era' highlighted the collaborative efforts across these layers.
2Keynote Highlights
Mark Collier, Executive Director of the PyTorch Foundation, delivered a keynote on September 8, emphasizing an open ecosystem where researchers, developers, model builders, and hardware companies collaborate. The session connected PyTorch with the path from experimentation to production, new accelerator architectures, and the infrastructure that runs AI workloads.
3New Members of PyTorch Foundation
On September 8, the PyTorch Foundation welcomed Alibaba Cloud, Ant Group, and Cambricon as new members. Alibaba Cloud joined as a Platinum Member, Ant Group as a Gold Member, and Cambricon as a Platinum Member. These new members bring expertise in cloud infrastructure, production AI at financial-services scale, and MLU accelerators, respectively.
4Technical Sessions and Demonstrations
The conference included co-located sessions, keynotes, technical demonstrations, and community meetings. Topics covered hardware adaptation, training and serving, open infrastructure, accelerator integration, and collaboration across open-source communities. Notable sessions included discussions on Kubernetes-based auto-scaling and device-agnostic PyTorch infrastructure.
5Contributions from New Members
Cambricon's contributions to PyTorch span torch.compile, Eager Operators, Device Runtime, Distributed Computing, AMP, Dataloader, and Profiler. Alibaba Cloud focuses on cloud infrastructure and the Qwen model family, while Ant Group brings experience with production AI at financial-services scale.
6Open Source AI Stack
At the press conference, the PyTorch Foundation outlined an open-source AI stack spanning applications and agents, building and delivering intelligence, running and scaling AI workloads, infrastructure and isolation, and heterogeneous compute. Key software layers include PyTorch, vLLM, Ray, Kubernetes, KServe, Kueue, OpenTelemetry, and llm-d.
7Collaborative Ecosystem
The PyTorch Foundation emphasized that no single organization builds the whole system. Together, their members span across Any Model, Any Chip, Any Cloud, and Any Agent. This collaborative ecosystem enables the advancement of AI technology through shared expertise and resources.
8Impact and Future Directions
The PyTorch Conference China 2026 showcased the importance of open-source collaboration in advancing AI technology. With contributions from over 100 China-based developers across 40+ organizations, the event highlighted the growing influence of China in the global AI landscape and set the stage for future innovations.
Original article: PyTorch Blog — PyTorch Conference China 2026: Advancing the Open Source AI Stack