Alibaba expands AI ambitions with larger Qwen models and new chip

Tuesday 22 September 2026 - 15:47

By: Azzat Manal

Alibaba expands AI ambitions with larger Qwen models and new chip

Alibaba is significantly expanding its artificial intelligence strategy, announcing plans for much larger Qwen models alongside a new generation of domestically developed AI processors as Chinese technology companies accelerate efforts to strengthen their computing capabilities.

The announcements were made at Alibaba Cloud’s annual Apsara conference in Hangzhou, where the Chinese technology group outlined an AI strategy covering everything from foundation models and semiconductor technology to the data-center infrastructure required to train and operate advanced systems.

Alibaba said its Qwen team is preparing models with between 5 trillion and 10 trillion parameters, a major increase compared with the company’s current flagship system. The larger models are intended to handle more sophisticated tasks that require longer sequences of reasoning and greater computational capacity.

The company is also developing the next generation of its Qwen family. Alibaba said Qwen 4 is currently being trained, while subsequent versions, including Qwen 4.5 and Qwen 5, are expected to reach the 5 trillion-to-10 trillion parameter range.

The expansion reflects the growing importance of large-scale AI models in the global technology industry. Parameters are commonly used as a broad indicator of the scale of an AI model, although a higher parameter count does not necessarily translate directly into better performance across every task.

Alibaba Chief Executive Eddie Wu said the company is pursuing increasingly capable machine intelligence and described the development of AI as the beginning of a technological transformation comparable in scale to earlier industrial revolutions.

The company also highlighted progress toward systems capable of improving aspects of their own development process. According to Alibaba, its Qwen researchers have been working on methods that allow AI systems to identify weaknesses, conduct experiments and generate training data with less direct human involvement.

Alongside the model announcements, Alibaba introduced the Zhenwu V900, a new AI processor developed by its T-Head semiconductor division. The company says the chip offers roughly three times the performance of its predecessor, the M890, and can be deployed in large clusters for training and running advanced AI systems.

The Zhenwu V900 is expected to enter mass production and commercial deployment in the first quarter of 2027. Alibaba also anticipates a substantial increase in its annual AI chip shipments as it expands its semiconductor capabilities.

The move comes as Chinese technology companies seek to develop domestic alternatives to advanced AI processors supplied by US manufacturers, particularly Nvidia. Washington’s export restrictions on sophisticated semiconductor technology have encouraged Chinese firms to invest more heavily in locally designed chips and computing infrastructure.

Alibaba is also planning a major expansion of its cloud infrastructure. Wu said the company aims to increase Alibaba Cloud’s global data-center capacity to more than 20 gigawatts by 2032, reflecting expectations of sustained demand for computing resources as businesses adopt AI applications.

The company said demand for AI-related cloud services has been particularly strong and is contributing to faster revenue growth. However, Alibaba acknowledged that supply-chain limitations could restrict the speed at which additional computing capacity can be brought online.

Alibaba’s announcements were welcomed by investors, with its Hong Kong-listed shares rising more than 5% during Tuesday’s trading session and reaching their highest level in roughly a month.

The company’s latest strategy illustrates how the competition in artificial intelligence is increasingly extending beyond software. Model development, semiconductor design, data centers and access to large-scale computing power are becoming closely interconnected as technology companies compete to build increasingly capable AI systems.