Alibaba Cloud, Cambricon, Ant Group Join PyTorch Foundation at Shanghai Conference

The PyTorch Foundation announced on September 8 at PyTorch Conference China 2026 in Shanghai that Alibaba Cloud and Cambricon joined as Platinum members and Ant Group as a Gold member. The conference ran alongside KubeCon + CloudNativeCon + OpenInfra Summit China 2026, and the new memberships took effect with the announcement. Huawei, already a long-standing supporter, remains a member, and representatives from all four companies delivered keynotes on the open AI stack, covering hardware, models, and infrastructure.

Membership details and governance

Platinum membership gives Alibaba Cloud and Cambricon one seat each on the PyTorch Foundation Governing Board and one seat each on the Technical Advisory Council. Both companies said they will work with the community to strengthen support for hardware and accelerators across the PyTorch ecosystem. Ant Group, as a Gold member, joins efforts to advance production-grade platforms for AI workloads. The foundation said more than 250 organizations across China contribute to PyTorch Foundation projects including DeepSpeed, Helion, PyTorch, Ray, Safetensors, and vLLM.

Keynotes map the stack

Alibaba Cloud's Qwen infrastructure team opened the keynote arc at the platform level, describing how they serve Qwen at massive scale. Huawei's talk explored hardware-software co-design and the interoperability needed between Chinese and global AI technology stacks, a question that grows as more accelerators enter production. Cambricon detailed how it is hardening PyTorch's device-agnostic foundation so any backend can offer broader reach and richer native capabilities. Ant Group showed through two linked talks how existing cloud native building blocks like Kubernetes Agent Sandbox and Kata Containers can be assembled into a secure, on-demand runtime for AI agents.

What this means for inference engineers

For engineers serving open-weight models, the new memberships signal a push to make PyTorch work natively across a wider range of accelerators. Cambricon's focus on device-agnostic infrastructure aims to reduce the friction of supporting multiple backends. Huawei's keynote reiterated its goal of native support for Ascend NPU, including native development, release, and distribution, so developers get the same PyTorch experience regardless of the accelerator. Alibaba Cloud's talk on serving Qwen at scale with multi-cluster AI infrastructure on Karmada points to practical patterns for running large models in production across heterogeneous hardware. The conference keynotes moved from silicon to software to deployment, giving attendees a full-stack view of how open source is being built into every layer of AI in China.

The PyTorch Conference North America is scheduled for October 20-21 in San Jose, with registration open.