The announcements sent Alibaba's Hong Kong-listed shares soaring 5.1% on Tuesday to their highest level in a month, as investors responded to the technology giant's plans to expand across the entire AI ecosystem.
Alibaba's strategy extends from the development of foundation models and advanced semiconductors to the construction of massive data-centre infrastructure capable of training and operating increasingly sophisticated AI systems.
The announcements were made at Alibaba Cloud's annual Apsara conference in Hangzhou, highlighting the company's efforts to strengthen its position as Chinese technology companies race to develop domestic alternatives to Nvidia's AI processors.
The push has gained urgency as Washington continues to tighten restrictions on China's access to advanced American-made semiconductor technology.
Alibaba Chief Executive Eddie Wu portrayed the company's investments as part of a much larger technological transformation.
“The truly groundbreaking products of the Machine Intelligence era have not yet arrived,” Wu said, describing the development of AI as the beginning of an era comparable to the Industrial Revolution.
Wu predicted that machines could eventually generate more than 1,000 times the “thinking” of all humanity, compared with less than 3% today.
Alibaba targets 10 trillion parameters
At the centre of Alibaba's latest AI push is a new generation of Qwen models that could be several times larger than the company's current flagship system.
Wu said Alibaba's Qwen team plans to train a model containing between 5 trillion and 10 trillion parameters, with the aim of enabling AI systems to handle “more complex, longer-horizon tasks”.
The company has said it is pursuing artificial superintelligence — AI systems designed ultimately to exceed human capabilities across a broad range of tasks.
Alibaba's current flagship, Qwen 3.8 Max, has approximately 2.4 trillion parameters. Parameters are among the measures used to describe the scale of an AI model and its ability to process complex information.
In a separate statement, Alibaba said its next-generation Qwen 4 model is currently being trained, while future versions, including Qwen 4.5 and Qwen 5, are expected to scale toward the 5 trillion-to-10 trillion-parameter range.
The company is also working to reduce the amount of human involvement required during the development and improvement of its models.
Wu said the Qwen team had made “meaningful” progress in enabling AI systems to identify their own weaknesses, conduct experiments and generate training data with limited human intervention.
The development could become an important part of Alibaba's strategy as the global AI industry shifts toward increasingly autonomous systems capable of performing longer and more complicated sequences of tasks.
New Chinese AI chip
Alibaba also used the conference to unveil a new AI processor developed by its T-Head semiconductor division.
The Zhenwu V900 is designed to provide three times the performance of its predecessor, the M890, according to Wu.
The company said as many as 500,000 of the new processors could be linked together in clusters to train and operate some of the world's largest AI models.
The chip is scheduled to enter mass production and commercial release in the first quarter of 2027.
Wu said Alibaba expects “significant growth” in annual shipments of its AI processors.
The company launched its previous-generation M890 chip in May, underscoring the rapid pace at which Chinese technology companies are attempting to develop domestic computing capacity for AI.
Alibaba's semiconductor push comes as Chinese companies face restrictions on obtaining some of the world's most advanced AI processors from U.S. suppliers.
That has increased pressure on China's technology industry to develop domestic alternatives capable of supporting the enormous computing requirements of modern AI systems.
Alibaba Cloud expands data-centre ambitions
Alibaba is also planning a major expansion of the infrastructure required to support its AI ambitions.
Wu said Alibaba Cloud aims to increase its global data-centre capacity to more than 20 gigawatts by 2032.
Demand from customers seeking AI computing capacity was “exceptionally robust”, Wu said, adding that the surge was accelerating revenue growth at Alibaba Cloud.
The company is nevertheless facing supply-chain constraints that are limiting how quickly it can expand its infrastructure.
“The industry's mid-to-long-term demand far outpaces our supply capabilities,” Wu said.
Alibaba Cloud plans to begin bringing its AI supernodes online at commercial scale during the current quarter, according to Wu.
The expansion reflects the enormous infrastructure requirements of advanced AI. Training increasingly sophisticated models requires large clusters of specialised processors, while operating those models for millions of users requires additional computing capacity.
AI's next phase
Wu suggested that today's most visible AI applications may represent only the beginning of a much larger technological shift.
He compared AI-powered coding to the light bulb during the development of electricity — an important early application, but not necessarily the technology's ultimate breakthrough.
For Alibaba, the strategy is therefore broader than building another chatbot or foundation model.
The company is attempting to establish a vertically integrated AI business spanning models, chips, cloud computing and data-centre infrastructure.
The scale of the planned Qwen models and the development of Alibaba's own processors also illustrate the intensifying competition between Chinese and U.S. technology companies over the future of artificial intelligence.
With its shares rising sharply following Tuesday's announcements, investors appeared to welcome Alibaba's expanded AI ambitions. The bigger test, however, will be whether the company can turn its enormous investments in computing infrastructure, chips and increasingly powerful models into commercially successful AI products.