Shanghai – September 10, 2026 -- Alibaba Cloud and Cambricon have joined the PyTorch Foundation as Platinum members, while Ant Group has signed on as a Gold member, the Linux Foundation-backed body announced at the first KubeCon + CloudNativeCon + OpenInfra Summit + PyTorch Conference China 2026 in Shanghai.
Three Chinese AI giants formalize ties to the open source framework governing global model deployment
The moves add Alibaba Cloud and Cambricon to the Foundation's Governing Board and Technical Advisory Council, each securing one seat on both bodies. Ant Group's Gold membership grants it a role in advancing production-grade AI infrastructure platforms. The three companies join Huawei, which has been a Foundation member since 2023 and helped launch the Accelerator Integration Working Group that brought Ascend NPU support into PyTorch as its first additional hardware platform.
More than 250 Chinese organizations already contribute to core PyTorch projects
The Foundation said upward of 250 organizations across China are contributing to projects under its umbrella, including DeepSpeed, Helion, PyTorch, Ray, Safetensors, and vLLM. Mark Collier, Executive Director of the PyTorch Foundation, said the new members strengthen the global open source AI ecosystem developers rely on for training, optimizing, serving, distributing, and running models.
Keynotes trace the AI stack from silicon to agent deployment
Alibaba Cloud's Qwen infrastructure team detailed how it serves the Qwen model family at scale using multi-cluster infrastructure on Karmada. Cambricon presented work to harden PyTorch's device-agnostic foundation, aiming to let any hardware backend offer broader reach and richer native capabilities for its Machine Learning Unit ecosystem. Ant Group delivered two linked talks on assembling Kubernetes Agent Sandbox and Kata Containers into secure, on-demand runtimes for AI agents. Huawei addressed hardware-software co-design and interoperability between Chinese and global AI stacks.
Executives frame membership as a bet on heterogeneous hardware support
Alibaba Cloud CTO Feifei Li said the company has run PyTorch at scale across heterogeneous hardware and wants to help raise the bar for AI infrastructure. Cambricon Vice President of Software Engineering Elton Gong said the goal is a native, out-of-the-box developer experience across a broader range of backends. Ant Group CTO Zhengyu He said the membership continues the company's long-term participation in global open-source collaboration to lower barriers to AI development.
Huawei signals continued push for native Ascend support inside PyTorch
Fred Li, Head of Huawei's Computing Open Source Development Team, said the company will keep working toward native development, release, and distribution for Ascend so developers get the same PyTorch experience regardless of accelerator choice. The Shanghai conference runs September 7-9, 2026, with PyTorch Conference North America scheduled for San Jose on October 20-21.