Key facts
- Alibaba has launched its first suite of AI models for robots, the Qwen Robot Suite.
- The suite includes models for navigation, environmental simulation, and physical task execution.
- Alibaba's Qwen 3.7-Max large language model is designed for sustained, multi-step agentic operations.
- Alibaba Cloud introduced the Panjiu AL128 Supernode Server to handle high-frequency AI agent inference.
- T-Head, Alibaba's chip subsidiary, unveiled the Zhenwu M890 AI processor, its most powerful to date.
- Alibaba projects AI-related product revenue to surpass conventional cloud compute sales by year-end.
Alibaba Group Holding has launched its first suite of artificial intelligence models for robots, signaling a significant shift in China's technology sector from chatbots to AI agents capable of interacting with the physical world. Developed by Alibaba's AI research unit, Tongyi Lab, the Qwen Robot Suite is designed to enable machines to perceive, reason, and interact with physical environments.
The suite includes Qwen-RobotNav for vision-language navigation, Qwen-RobotWorld for video simulation, and Qwen-RobotManip, a vision-language-action model built on the Qwen3.5-4B architecture for physical execution. Alibaba's latest large language model, Qwen 3.7-Max, is optimized for agentic workloads, capable of sustained, multi-step operations. In internal benchmarks, it outperformed a manufacturer's official version by tenfold.
Alibaba Cloud introduced the Panjiu AL128 Supernode Server, designed to handle the high-frequency inference requests of AI agents, featuring Agentic RL for continuous model refinement and safety guardrails. Complementing these software advancements, T-Head, Alibaba's chip design subsidiary, unveiled the Zhenwu M890, its most powerful AI processor to date, offering three times the performance of its predecessor and supporting various precision formats for efficient agent workloads. T-Head has shipped over 560,000 Zhenwu chips to more than 400 customers across 20 industries.
Alibaba CEO Eddie Wu indicated a strategic repositioning, forecasting that AI-related product revenue will surpass conventional cloud compute sales. The company projects its model and application services platform annual recurring revenue (ARR) to reach RMB 30 billion (approximately US$4.1 billion) by year-end, up from a projected RMB 10 billion (US$1.4 billion) in the June quarter.