Alibaba Expands AI Ambitions With New Chip and Larger Model, Shares Rise 5 percent

Alibaba Group is stepping up its artificial intelligence ambitions with plans for a significantly larger AI model and a new generation of processors, as Chinese technology companies accelerate efforts to build domestic alternatives to advanced Nvidia chips. At its annual Apsara conference in Hangzhou on Tuesday, Alibaba said its AI research team is working toward models that could eventually reach between 5 trillion and 10 trillion parameters. The announcement helped lift the company's Hong Kong-listed shares by about 5%, taking them to their highest level in roughly a month.

The planned model would be considerably larger than Alibaba's current flagship Qwen 3.8 Max, which has around 2.4 trillion parameters. Parameters are commonly used as a broad indicator of the scale of an AI model, although model size alone does not determine its overall performance. Alibaba said its next-generation Qwen 4 model is already being trained, while future versions such as Qwen 4.5 and Qwen 5 could expand toward the 5-trillion-to-10-trillion-parameter range. The company expects these larger systems to handle increasingly complicated tasks that require longer sequences of reasoning and planning.

Alibaba CEO Eddie Wu said the company's Qwen team is also making progress in developing AI systems capable of identifying their own weaknesses, conducting experiments and producing training data with less direct human involvement. The broader goal, he said, is to move toward artificial superintelligence, referring to systems that could eventually exceed human capabilities in a wide range of tasks.

Alongside the model announcement, Alibaba introduced the Zhenwu V900, a new AI processor developed by its T-Head semiconductor division. Wu said the chip offers roughly three times the performance of its predecessor, the M890. The new processor is designed to work in large clusters, with Alibaba saying configurations could eventually connect as many as 500,000 chips for training and running highly demanding AI models. Mass production and commercial availability are expected to begin in the first quarter of 2027.

The chip announcement comes as Chinese technology companies face growing pressure to develop their own advanced computing hardware. US restrictions on the export of sophisticated AI processors to China have increased the importance of domestic semiconductor development for the country's technology sector.

Alibaba's strategy extends beyond AI models and chips. The company is also expanding the computing infrastructure required to train and operate increasingly powerful systems. Wu said Alibaba Cloud aims to increase its worldwide data-centre capacity to more than 20 gigawatts by 2032. According to the company, demand from customers for AI computing services remains particularly strong and is contributing to faster growth in Alibaba Cloud's business. However, supply-chain limitations could restrict how quickly the company can expand its infrastructure.

Alibaba also plans to begin deploying its AI supernodes at commercial scale during the current quarter. The company views these systems as part of the infrastructure needed to support increasingly large AI workloads. Wu compared the current development of AI coding tools with the early use of electricity, describing it as an important early application rather than the ultimate breakthrough that AI could deliver.

The announcements highlight Alibaba's attempt to build a broad AI ecosystem covering chips, foundation models, cloud computing and data-centre infrastructure. As competition intensifies in China's AI industry, the company's ability to scale these technologies will be closely watched by investors and technology companies alike.

Disclaimer: This image is taken from Reuters.