Alibaba AI chip Zhenwu V900 promises three times its predecessor’s performance as the Chinese technology giant expands across chips, models and data centres.
HANGZHOU: Alibaba AI chip development has taken another major step with the unveiling of the Zhenwu V900, a new processor designed to provide the computing power needed to train and run increasingly large artificial intelligence models.
Alibaba introduced the processor at its annual Apsara Conference in Hangzhou on September 22. Developed by its semiconductor unit T-Head, the V900 delivers about three times the performance of its previous generation, according to the company.
Alibaba describes the V900 as China’s most powerful AI processor. The chip can also be connected into large computing clusters, allowing thousands of processors to work together on demanding AI workloads.
Mass production is expected to begin by early 2027.
Why is Alibaba building its own AI chips?
AI models require enormous computing resources, making access to high-performance processors increasingly important for technology companies.
China’s push to develop domestic alternatives has also intensified as US restrictions have limited Chinese access to some advanced American semiconductor technology.
The Alibaba AI chip is part of a broader strategy spanning processors, cloud computing, data centres and the company’s Qwen family of AI models.
Alibaba chief executive Eddie Wu said at the conference that the company wants to develop the full AI technology stack rather than depend entirely on outside suppliers.
That strategy places Alibaba alongside Chinese technology groups including Huawei that are investing heavily in domestic AI computing infrastructure. Huawei said last week that demand for its AI computing equipment in China was exceeding available supply.
Alibaba is planning a much larger AI model
The chip announcement came alongside another ambitious target.
Alibaba plans to train a next-generation artificial intelligence model containing between five trillion and 10 trillion parameters. That could make it several times larger by parameter count than the company’s current flagship Qwen 3.8 Max model.
Parameter count alone does not determine how capable an AI model will be, but the proposed scale demonstrates the computing demands Alibaba expects future systems to require.
The company is preparing infrastructure to match that ambition. Alibaba Cloud is targeting more than 20 gigawatts of global data-centre capacity by 2032 as demand for AI computing increases.
Alibaba shares rose 5.1 per cent to a one-month high following the announcements.
For the Alibaba AI chip strategy, the bigger significance is therefore not one processor alone. Alibaba is attempting to control more of the technology chain behind AI, from silicon and computing infrastructure to the models and applications running on top.
Impact to expect
Competition. The V900 could strengthen China’s domestic AI computing ecosystem while giving Alibaba greater control over the infrastructure required to build increasingly powerful Qwen models.
