Alibaba Cloud Comprehensively Upgrades for the Agentic Era, Unveils New “Chip-Cloud-Model-Inference” Technology System
Summary
Alibaba Cloud announced a full-stack "chip-cloud-model-inference" upgrade for the Agentic era, launching the QianWen Cloud site, an M890-based super-node server, and the Qwen3.7-Max flagship model. The company presented a self-developed data-center chip matrix and networking/storage chips, plus a reworked Agentic Cloud that exposes Skills and CLI modules so agents can call cloud capabilities like functions. Alibaba also detailed the Bailian inference platform with pooled GPU scheduling, context caching, elastic throughput, Agentic RL and built-in safety controls for long-running, high-concurrency agent workloads. The release signals a strategic shift toward token-metered, agent-driven AI services and accelerated model iteration, and Alibaba highlighted early commercial traction and a strong market reaction.