Alibaba plans larger Qwen model and new AI chip

Alibaba plans to train a new AI model with between 5 trillion and 10 trillion parameters, chief executive Eddie Wu Yongming said at Alibaba Cloud’s annual Apsara Conference in Hangzhou. The company is pushing its Qwen team to improve model architecture and data optimisation so future systems can complete more complex, longer-horizon tasks and move toward artificial superintelligence.

Alibaba’s current flagship model, Qwen 3.8 Max, has 2.4 trillion parameters, making the planned system roughly two to four times larger. Wu said Alibaba’s proprietary M890 AI supernode already supports inference for models above 2 trillion parameters, and said Qwen has made meaningful progress on recursive self-improvement, where models identify limitations, design experiments and synthesise data for self-evolution.

The company also unveiled the Zhenwu V900, a next-generation AI chip from its T-Head semiconductor unit. Wu said it delivers three times the performance of the M890, and that a single cluster can support up to 500,000 cards for frontier model training and inference. Alibaba Cloud aims for global data centre capacity to surpass 20 gigawatts by 2032 as demand remains strong, though supply-chain shortages continue to constrain expansion.