Alibaba also highlighted Qwen 3.7-Max as its flagship AI model for real-world tasks, adding to its broader push into enterprise AI infrastructure amid tighter pressure on Nvidia products in China.
Alibaba said its new Zhenwu M890 artificial intelligence chip is three times faster than the previous model and is designed for AI agents that need heavy memory use and system coordination. At the Alibaba Cloud Summit, the company also launched the Panjiu AL128 server system with 128 accelerators in one rack and said domestic customers can access it through Bailian, Alibaba Cloud’s model platform. Alibaba’s chip unit, T-Head, has delivered more than 560,000 Zhenwu chips to over 400 customers across 20 sectors, including automotive and finance. The company also introduced Qwen 3.7-Max, which the new report described as Alibaba’s flagship AI model for real-world tasks and said could automate complex work, improve productivity, and intensify competition in AI. Separately, reports said China blocked Nvidia’s China-focused RTX 5090D V2 graphics card from customs processing, adding to supply restrictions around advanced chips in the country.