Alibaba Group Holding and Tencent Holdings are leading a strategic shift among Chinese tech majors, moving their generative AI investments from digital chatbots toward embodied AI systems that power physical robots. The pivot, accelerating through May 2026, marks a new competitive front in the global AI race — one fought not on benchmark leaderboards but on factory floors and warehouses.

Alibaba's Qwen3.7-Max and the robotics toolkit

Alibaba Group Holding launched its Qwen3.7-Max model last week, featuring “tool-calling” capabilities that enable the model to act as a digital brain, triggering external software and hardware components. According to the company, the model can control robots by orchestrating physical actions including navigation, obstacle avoidance, and task planning.

Beyond the flagship model, Alibaba has released a suite of supporting AI tools for robotics — a robotic gripper agent, a navigation model, and a vision-language system built specifically for physical-world interaction. The move positions Alibaba as a full-stack embodied AI provider rather than a pure language model developer.

Tencent's OpenClaw enters mass production

Embodied AI start-up Zeroth announced earlier this month that its M1 humanoid robot had become the first mass-produced robot to integrate Tencent Holdings' OpenClaw AI agent framework. The integration allows large language models to interpret human speech and immediately translate commands into physical robotic movements — a milestone that bridges conversational AI with real-world actuation at scale.

The M1's mass-production status is significant: it signals that embodied AI is moving beyond laboratory prototypes into commercial deployment, putting pressure on global competitors to accelerate their own hardware-software integration timelines.

Why it matters: from digital to physical intelligence

“The past few years of large language model development have mainly focused on solving problems in the digital world,” Wu Bangyi, chief data officer at Tianyu Shuke, was quoted as saying by Securities Daily. The shift toward embodied AI represents an attempt to extend that problem-solving capacity into the physical environment — a domain where latency, sensor fusion, and mechanical reliability introduce challenges that pure software cannot address.

Industry observers note that the pivot also reflects intensifying commoditisation pressure on large language models, where differentiation through benchmark scores alone is no longer sufficient to command premium pricing or enterprise contracts.

The competitive backdrop

The embodied AI push by Alibaba and Tencent arrives as global players including established robotics firms and well-funded start-ups compete to define the software stack that will run the next generation of autonomous machines. China's robotics sector has attracted significant attention from investors and analysts at firms including UBS and Goldman Sachs, who have flagged the country's manufacturing scale as a structural advantage in deploying humanoid robots commercially.

Institutions such as Tsinghua University and the China Academy of Information and Communications Technology have also been active in setting technical standards and conducting foundational research that underpins many of these commercial deployments.

What's next

With Alibaba's full robotics model suite now publicly available and Zeroth's M1 already in mass production with Tencent's framework embedded, the near-term question is whether Western robotics and AI companies can match the pace of hardware-software co-development coming out of Shenzhen and Beijing. The companies most exposed are those still treating robotics as a hardware problem rather than an integrated AI systems challenge.