Alibaba has expanded its artificial intelligence (AI) platform with new cloud infrastructure, AI chips, and large language models designed to support autonomous AI agents that can perform complex tasks with minimal human supervision.
The biggest update is Qwen3.7-Max, Alibaba’s latest large language model (LLM) built for coding, reasoning, workflow automation, and long-running AI operations. The model can handle software engineering tasks, office automation, debugging, and multi-step workflows involving hundreds or even thousands of actions.
Alibaba described Qwen3.7-Max as a foundation model for AI agents that can operate continuously for up to 35 hours and manage more than 1,000 tool calls without performance degradation. The model is optimized for agent frameworks such as OpenClaw, Hermes Agent, Claude Code, Qwen Paw, and Qoder, making it easier for developers and enterprises to build AI-powered services.
The company also upgraded its cloud infrastructure to address increasing demand for AI computing. Alibaba Cloud introduced the Panjiu AL128 Supernode Server, which integrates 128 AI accelerators in a single rack to support large-scale AI training and inference workloads.
Powered by Alibaba’s Zhenwu M890 AI processors and ICN Switch 1.0 networking chips, the system delivers petabyte-per-second bandwidth to process large volumes of simultaneous AI requests. The infrastructure is now available through Alibaba’s Bailian model service platform in China.
Alibaba also added a reinforcement learning feature called Agentic RL to Bailian. The system uses feedback from AI agent execution to continuously improve model performance. Built-in safety controls are included to keep autonomous agents operating within defined boundaries.
T-Head, Alibaba’s semiconductor design subsidiary, introduced the Zhenwu M890 AI accelerator, which delivers three times the performance of the earlier Zhenwu 810E processor, according to the company.
The processor includes 144 gigabytes (GB) of GPU memory and 800 GB per second of inter-chip bandwidth. It also supports multiple precision formats ranging from FP32 to FP4, allowing companies to balance computing power, speed, and operating costs for AI workloads.
Alongside the chip, T-Head introduced the ICN Switch 1.0 networking processor, which delivers up to 25.6 terabits per second of bandwidth for large AI computing clusters while reducing latency and congestion.
Alibaba said more than 560,000 Zhenwu chips have already been delivered to customers. More than 400 companies across 20 industries, including automotive and financial services firms, are using the processors for AI-driven operations.
The expansion shows intensifying global competition in enterprise AI infrastructure as technology companies race to build platforms for autonomous AI agents. For businesses in Southeast Asia, including the Philippines, Alibaba’s growing AI ecosystem could provide an alternative to Western AI platforms for cloud-based AI development and deployment.