Alibaba Unveils New AI Chip and Qwen Roadmap
Alibaba has unveiled a new in-house AI chip and outlined plans for much larger future Qwen models and data-centre capacity.
Alibaba is pushing deeper into every layer of the AI stack, from chips and models to cloud infrastructure and data centres.
What happened
The company unveiled the Zhenwu V900, a new in-house AI chip that Alibaba says delivers roughly three times the performance of its predecessor.
CEO Eddie Wu also outlined plans to train future Qwen models with 5 trillion to 10 trillion parameters, compared with 2.4 trillion for the company's current flagship model.
Alibaba is targeting more than 20GW of global data-centre capacity by 2032.
The company says Qwen 4 is already in training, with subsequent generations expected to scale further.
Why it matters
Alibaba is not competing only as a model developer.
Owning more of the chip, cloud and data-centre stack can reduce dependence on external suppliers and give the company more control over cost, deployment and product integration.
That matters particularly in China, where access to advanced foreign AI hardware remains strategically sensitive.
The bigger picture
The largest AI companies are increasingly converging on the same vertically integrated strategy.
Model quality still matters, but infrastructure control can be just as important when training and inference require enormous amounts of compute.
Alibaba's roadmap shows that the AI race is becoming a contest between full-stack platforms capable of financing chips, models and physical infrastructure together rather than standalone software labs.