Alibaba unveils plans for artificial superintelligence

Alibaba Group plans to train a new artificial intelligence model with between five trillion and 10 trillion parameters.

Chief Executive Eddie Wu announced the plan on Tuesday, as the Chinese technology giant outlined a sweeping strategy encompassing AI models, chips and data centres.

Speaking at Alibaba Cloud's annual Apsara Conference in Hangzhou, Wu said the company's Qwen team was continuing research into model architecture and data optimisation, with the goal of completing "more complex, longer-horizon tasks" and advancing toward artificial superintelligence, or ASI.

In a separate statement, Alibaba said its next-generation model, Qwen 4, was currently in training, adding that the upcoming Qwen 4.5 and Qwen 5 series were projected to scale up to five trillion to 10 trillion parameters.

Parameters are the variables a model learns during training and are a rough gauge of its size.

Alibaba's current flagship model, Qwen 3.8 Max, has 2.4 trillion parameters, meaning the planned model would be roughly two to four times larger.

Wu said Alibaba's proprietary M890 AI supernode already handles inference for models above two trillion parameters, a capability he said only "a handful of companies globally" possess.

The Qwen team has made "meaningful progress" on recursive self-improvement, in which models identify their own limitations, design experiments and synthesise data to drive a cycle of self-evolution, he said.

Alibaba also introduced the Zhenwu V900, a next-generation AI chip from its T-Head semiconductor unit, which Wu called the most powerful AI chip in China.

It delivers three times the performance of its predecessor, the M890, and a single cluster built on it can support up to 500,000 cards for frontier model training and inference, he said.

The chip is scheduled for mass production and commercial release in the first quarter of 2027, the company said in the statement. The M890 was released in May.

The company expects "significant growth" in annual AI chip shipments, Wu said.

The launch comes as Chinese firms race to build domestic alternatives to Nvidia's processors amid US export restrictions.

Wu set a target for Alibaba Cloud's global data centre capacity to surpass 20 gigawatts by 2032.