Alibaba plans 10-trillion-parameter AI model, unveils new chip

A representative picture.Alibaba Group plans to train a new artificial intelligence model with 5 trillion to 10 trillion parameters as the Chinese technology giant expands its push across AI models, chips and data centres.Speaking at Alibaba Cloud’s annual Apsara Conference in Hangzhou on Tuesday, Chief Executive Eddie Wu said the company’s Qwen team was continuing research into model architecture and data optimisation, with the aim of completing ‘more complex, longer-horizon tasks’ and advancing towards artificial superintelligence (ASI).Parameters are variables that a model learns during training and are considered a rough measure of its size.Alibaba’s current flagship model, Qwen 3.8 Max, has 2.4 trillion parameters, making the planned model roughly two to four times larger.Wu said Alibaba’s proprietary M890 AI supernode can already handle inference for models with more than 2 trillion parameters, a capability he said only ‘a handful of companies globally’ possess.The Qwen team has also made ‘meaningful progress’ on recursive self-improvement, a process in which models identify their own limitations, design experiments and synthesise data to drive a cycle of self-evolution, according to Wu.Alibaba unveils Zhenwu V900 AI chipAlongside its plans for larger AI models, Alibaba introduced the Zhenwu V900, a next-generation AI chip developed by its T-Head semiconductor unit.

Wu described it as the most powerful AI chip in China.The V900 delivers three times the performance of its predecessor, the M890, while a single cluster built on the chip can support as many as 500,000 cards for frontier model training and inference, Wu said.Alibaba expects ‘significant growth’ in annual AI chip shipments, he added.The launch comes as Chinese companies race to develop domestic alternatives to Nvidia’s processors amid US export restrictions.Alibaba Cloud targets over 20 GW capacity by 2032Alibaba is also expanding its computing infrastructure, with Wu setting a target for Alibaba Cloud’s global data-centre capacity to exceed 20 gigawatts by 2032.Customer demand for AI remains ‘exceptionally robust’ and is accelerating Alibaba Cloud’s revenue growth, Wu said.

However, shortages across the global AI data-centre supply chain are limiting the pace at which the company can expand.‘The industry’s mid-to-long-term demand far outpaces our supply capabilities,’ Wu said, adding that Alibaba Cloud would begin bringing its AI supernodes online at commercial scale this quarter.Wu described the current shift as the beginning of an era of ‘Machine Intelligence’ comparable to the Industrial Revolution.

He predicted that machines would eventually produce more than 1,000 times the ‘thinking’ of all humanity, compared with less than 3 per cent today.He compared AI coding with the light bulb of the electrical age, describing it as an early application rather than the breakthrough product of the new technological era.‘The truly groundbreaking products of the Machine Intelligence era have not yet arrived,’ Wu said.