Alibaba is planning an AI model with as many as 10 trillion parameters and unveiled a new AI chip as the Chinese technology giant expands across nearly every layer of the AI infrastructure stack.
The company said today that future versions of its Qwen model could contain 5 trillion to 10 trillion parameters, up from 2.4 trillion for its current flagship Qwen 3.8 Max, Reuters reported. Parameters are one rough measure of a model’s size, although a larger model is not automatically a better one.
Alibaba also introduced the Zhenwu V900, an AI processor developed by its T-Head semiconductor unit. CEO Eddie Wu said the chip delivers three times the performance of its predecessor and can be connected in clusters of as many as 500,000 processors.
The processor is expected to enter mass production and commercial release in the first quarter of 2027.
The announcements matter beyond model size. Alibaba is trying to control more of the technology needed to build and operate AI, including models, chips, cloud services and data centers. Chinese technology companies are also under pressure to develop alternatives to Nvidia processors as U.S. export restrictions limit access to some advanced chips.
Alibaba said it is also targeting more than 20 gigawatts of global data-center capacity by 2032. Wu said customer demand for AI remains exceptionally strong but that supply constraints are limiting expansion.
“The industry’s mid-to-long-term demand far outpaces our supply capabilities,” Wu said, according to Reuters.
The company also said its Qwen team has made progress toward models that can identify their own weaknesses, conduct experiments and generate training data with limited human involvement.
Alibaba shares in Hong Kong rose 5.1% to their highest level in a month.