Alibaba plans AI model with 5 trillion to 10 trillion parameters, unveils new chip

Alibaba's Qwen team plans to train a new AI model with between 5 trillion and 10 trillion parameters, according to a report published by wtvbam.com on September 21, 2026. The plan was described by Alibaba chief executive Eddie Wu. According to the article, the model is intended to tackle 'more complex, longer-horizon tasks,' and the company is pursuing artificial superintelligence. The report also states that Alibaba unveiled a new chip. The excerpt provides no other specification for the planned model.

The article quotes Wu saying that customer demand for AI was 'exceptionally robust,' and that this demand was accelerating Alibaba Cloud's revenue growth. No figure was given for that revenue or for the rate of growth. The excerpt does not state how long training would take, when the model might be completed or released, whether it would be published openly, or what the new chip is designed to do. It gives no parameter count or performance figure for the chip.

The report presents the 5-trillion-to-10-trillion parameter target as its headline figure and links the effort to a stated pursuit of artificial superintelligence. It also ties the plan to commercial results, with Wu describing AI demand as 'exceptionally robust' and crediting it with accelerating Alibaba Cloud's revenue growth.