China’s AI Giants Take Board Seats at PyTorch
3 min readThe PyTorch Foundation announced in Shanghai this week that Alibaba Cloud and Cambricon have joined as Platinum members, and Ant Group as a Gold member. The Platinum tier is not a logo on a website. It comes with a seat on the Governing Board and a seat on the Technical Advisory Council, which means two Chinese companies now help steer the framework most of the world trains its models on.
What the PyTorch Foundation Is
PyTorch began at Meta and was handed to the Linux Foundation in 2022. The foundation is now a vendor-neutral home for a wider stack: PyTorch itself plus vLLM, DeepSpeed, Ray, Helion and Safetensors. Huawei joined in 2023 and helped start the Accelerator Integration Working Group, which is how its Ascend NPU became the first platform beyond Nvidia to get native treatment in PyTorch.
The foundation says more than 250 organizations across China already contribute to its projects, so this formalizes a presence rather than creating one.
Silicon, Models, Then Deployment
The announcement landed alongside keynotes at KubeCon plus PyTorch Conference China 2026, and the speaker list reads as a full stack. Alibaba Cloud’s Qwen infrastructure team covered serving its open-weight models at scale. Cambricon, a Chinese designer of AI accelerators, spoke on hardening PyTorch’s device-agnostic foundation so any backend can plug in. Huawei covered hardware and software co-design and the interoperability needed between Chinese and global AI stacks. Ant Group took the application layer, showing how Kubernetes Agent Sandbox and Kata Containers can be assembled into a secure runtime for AI agents.
Alibaba Cloud CTO Feifei Li framed membership as a natural step after years of running PyTorch across heterogeneous hardware. Cambricon’s Elton Gong said the goal is a native, out-of-the-box developer experience across a broader range of backends, per the PyTorch Foundation.
Why It Matters
Cambricon has been on the US Commerce Department’s Entity List since 2022, and it now holds a governance seat at the foundation behind the world’s dominant AI framework. That is the tension worth watching. Export controls restrict what silicon crosses borders, but open source governance is a separate arena with its own rules.
The practical effect is that PyTorch keeps drifting away from being a single-vendor accelerator story. If Cambricon’s MLU chips and Huawei’s Ascend NPUs get first-class support upstream, developers outside China benefit from the portability too. Watch how the device-agnostic work lands in the next PyTorch releases.