Alibaba targets 20 GW of data center capacity by 2032 as V900 chip scales to 500,000 GPUs

Alibaba says it plans to expand the capacity of the data centers operated by Alibaba Cloud worldwide past 20 GW by 2032. The target was reported by the South China Morning Post. The company has not said whether the figure refers to IT load or total facility power.

At its annual Apsara Conference in Hangzhou, CEO Eddie Wu introduced the Zhenwu V900, an AI GPU designed by Alibaba's T-Head chip unit. Wu called it “the most powerful AI chip in China today.” Alibaba says the V900 delivers three times the performance of its predecessor and can be used in clusters of up to 500,000 GPUs; the figure describes a claimed scaling capacity, not a confirmed deployed system.

Reporting on the launch, Zhidx listed 216 GB of memory and 1,200 GB/s of chip-to-chip bandwidth. Alibaba has not published figures for FLOPS, manufacturing process, power use or internal memory bandwidth, limiting comparisons with rival chips. The V900 was previously expected to reach clients in Q3 2027, but its timeline has moved forward by two quarters. T-Head vice president Gao Hui said the company plans to release one generation each year, with the J900 due in Q3 2028.

Alibaba also cited a May 2026 T-Head statement that it had shipped just over 560,000 Zhenwu chips in total. The company says medium- and long-term demand exceeds its supply, while shortages across the AI data center supply chain are limiting how quickly it can expand computing infrastructure.