- Zhenwu V900, the latest Alibaba AI chip, is due for mass production in Q1 2027, with manufacturing details undisclosed.
- CEO Eddie Wu says demand is outpacing Alibaba’s ability to supply data centre capacity.
Alibaba showcased the Zhenwu V900, a new Alibaba AI chip, at its Apsara Conference in Hangzhou on September 22. At the same event, CEO Eddie Wu set a target for Alibaba Cloud to operate more than 20GW of data centre capacity worldwide by 2032.
Wu said customer demand for AI remained “exceptionally robust” and was accelerating Alibaba Cloud’s revenue growth. But he acknowledged that shortages across the global AI data centre supply chain were limiting how fast the company could expand, according to Reuters. “The industry’s mid-to-long-term demand far outpaces our supply capabilities,” he said.
What the Alibaba AI chip offers
The Zhenwu V900 was designed by T-Head, Alibaba’s chip unit. Alibaba said it delivers three times the performance of the Zhenwu M890, released in May. The chip carries 216GB of memory and 1,200GB/s of chip-to-chip bandwidth, and it natively supports FP8 and FP4. These are low-precision number formats that let a chip run AI inference, the work of answering queries, faster and at lower cost.
Alibaba said the V900 is scheduled for mass production and commercial release in the first quarter of 2027. As per a Reuters report, Wu called it the most powerful AI chip in China.
The chip will sit inside an upgraded supernode server that Alibaba said can scale to clusters of up to 500,000 cards. A supernode links large numbers of accelerators so they operate as a single system. Wu said Alibaba Cloud would begin bringing its AI supernodes online at commercial scale this quarter.
Questions the V900 claims leave open
Alibaba did not publish the benchmarks behind the threefold performance figure. It also did not say who will manufacture the V900 or on which process node. That detail matters because US export controls restrict Chinese companies’ access to advanced chips and to the tools used to make them.
The company said its Zhenwu chips serve more than 650 customers in sectors including automotive, finance, energy and manufacturing. That figure covers T-Head’s existing chips, since the V900 is not yet on sale.
Alibaba also described an experiment in which its Qwen3.8-Max model spent more than 60 hours running chip-design software autonomously. According to the company, the model produced chip bus modules with 42% less area and no loss in performance. The results are Alibaba’s own, and it did not say whether those modules are used in the V900.
Bigger models drive the compute demand
The hardware plans track Alibaba’s model roadmap. The company said Qwen 4 is in training, and that the Qwen 4.5 and Qwen 5 series are projected to reach 5 to 10 trillion parameters. Parameters are the values a model learns during training and a rough measure of its size. Alibaba’s current flagship, Qwen3.8-Max, has 2.4 trillion parameters, making the planned model roughly two to four times larger, Reuters reported.
Wu argued that machine “thinking” is less than 3% of the volume of human thinking today, and could eventually reach 1,000 times human capacity. He said the major products of this era are still to come. “The truly groundbreaking products of the Machine Intelligence era have not yet arrived,” he said.
Malaysia anchors Alibaba Cloud in Southeast Asia
The capacity push extends to Southeast Asia. In June, Alibaba Cloud opened a public cloud region in Johor with two new data centres, bringing its Malaysian footprint to five facilities, its largest in Southeast Asia. The company said the expansion is part of the US$53 billion AI and cloud infrastructure investment it announced in 2025.
Alibaba Cloud has also said it plans to roll out agentic AI services in Malaysia in the second half of 2026. At Apsara, it unveiled AgentCore, a platform for building and managing enterprise AI agents, along with Agent Context, which it said cuts token usage by up to 67%. Alibaba did not say when these services will reach its international regions.
Alibaba also did not say whether Zhenwu chips will be deployed in its data centres outside China.
Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events, click here for more information.
Tech Wire Asia is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.
The post Alibaba unveils AI chip and 20GW plan as CEO admits supply squeeze appeared first on TechWire Asia.