Alibaba unveils the Zhenwu V900 AI chip, claiming three times the performance of its predecessor, with mass production due in early 2027
The chip from Alibaba's T-Head unit has 216 GB of GPU memory, the company says. It has not published the benchmarks behind the threefold claim.
At its Apsara Conference in Hangzhou on 22 September, Alibaba unveiled the Zhenwu V900, a processor for training and running AI models from its chip unit T-Head. Alibaba's release says the V900 “delivers three times the performance of its predecessor, the Zhenwu M890 (released in May)”. It gives the chip 216 GB of GPU memory and says it “is scheduled for mass production and commercial release in Q1 2027”.
Alibaba has not published the benchmarks behind the threefold claim, so the comparison is the company's own. The V900 follows the M890 by about four months, by Alibaba's own dates.
The Switchboard's 25 September edition covered other news from the same conference: the agent products AgentCore and an agentic computer, the Qwen 4 training run and the roadmap to Qwen 4.5 and Qwen 5, and chief executive Eddie Wu's goal of more than 20GW of data-centre capacity by 2032. The Zhenwu V900 was not in it.
A figure of more than $53bn of AI spending over three years circulated with the conference coverage, but neither Alibaba's release nor Wu's keynote gives an investment amount, so it is not repeated here.
- Claimed Alibaba says the Zhenwu V900 delivers three times the performance of the Zhenwu M890, released in May. Alibaba Cloud
- Claimed Alibaba says the V900 has 216 GB of GPU memory and is scheduled for mass production and commercial release in Q1 2027. Alibaba Cloud
- Confirmed Neither Alibaba's release nor Eddie Wu's keynote transcript gives an investment amount. Alibaba Group
Models & releasesMoney & markets
Caught up on October 17, 2026, after it was missed at the time · front page