Alibaba unveiled the Zhenwu V900, a new AI chip it says is meant to support both model training and inference, as the company set out a broader plan to expand the computing behind its cloud business.

At Alibaba's annual conference in Hangzhou, CEO Eddie Wu called the V900 China's most powerful AI chip and said it can deliver three times the performance of the preceding Zhenwu M890. The Associated Press reported that Alibaba uses its Zhenwu chips in data centers that provide computing for the company and its cloud clients. Those performance claims are Alibaba's, rather than an independent benchmark.

The chip announcement came alongside an ambitious infrastructure target. Alibaba said it expects to have more than 20 gigawatts of computing capacity by 2032, citing sharply rising demand for AI computing. It also said it plans to train a future AI model in the range of 5 trillion to 10 trillion parameters; its latest Qwen3.8-Max model has 2.4 trillion parameters, according to AP's report.

The practical question is how fast that roadmap can become deployed capacity. Wu said shortages across the AI data-center supply chain are already limiting the pace of expansion. That caveat matters because a chip announcement and a capacity target describe an intended buildout, not a guarantee that the systems will arrive on the schedule or at the scale Alibaba has outlined.

The announcement also lands amid a larger push by Chinese technology companies to develop more domestic AI hardware and models while U.S.-led restrictions limit access to some advanced chips and chipmaking equipment. Alibaba's new processor is a concrete part of that strategy, but its competitive performance will need to be judged after the company moves beyond its own claims.

Alibaba unveiled the Zhenwu V900, a new AI chip it says is meant to support both model training and inference, as the company set out a broader plan to expand the computing behind its cloud business.

At Alibaba's annual conference in Hangzhou, CEO Eddie Wu called the V900 China's most powerful AI chip and said it can deliver three times the performance of the preceding Zhenwu M890. The Associated Press reported that Alibaba uses its Zhenwu chips in data centers that provide computing for the company and its cloud clients. Those performance claims are Alibaba's, rather than an independent benchmark.

The chip announcement came alongside an ambitious infrastructure target. Alibaba said it expects to have more than 20 gigawatts of computing capacity by 2032, citing sharply rising demand for AI computing. It also said it plans to train a future AI model in the range of 5 trillion to 10 trillion parameters; its latest Qwen3.8-Max model has 2.4 trillion parameters, according to AP's report.

The practical question is how fast that roadmap can become deployed capacity. Wu said shortages across the AI data-center supply chain are already limiting the pace of expansion. That caveat matters because a chip announcement and a capacity target describe an intended buildout, not a guarantee that the systems will arrive on the schedule or at the scale Alibaba has outlined.

The announcement also lands amid a larger push by Chinese technology companies to develop more domestic AI hardware and models while U.S.-led restrictions limit access to some advanced chips and chipmaking equipment. Alibaba's new processor is a concrete part of that strategy, but its competitive performance will need to be judged after the company moves beyond its own claims.