Alibaba has reported its financial results for the second quarter of fiscal year 2027, posting 60.973 billion yuan in AI cloud and computing power service revenue—a 53% year-on-year increase. The company also disclosed progress on T-Head's new-generation training-and-inference integrated AI chip, Zhenwu V900, which is slated to enter mass production and go on sale in the first quarter of next year. The V900 delivers three times the computing power of its predecessor, the Zhenwu M890, and comes equipped with 216GB of memory, 1,200GB/s inter-chip interconnect bandwidth, and native support for FP8 and FP4 low-precision computing—capable of handling large-model training and inference tasks as well as high-load scenarios. Powered by its in-house solution, the chip enables thousand-card full-bandwidth high-speed interconnect; a single cluster of the new-generation Alibaba Cloud AI super node equipped with the V900 can scale up to 500,000 cards at maximum, with full-stack collaboration reducing communication losses. The Zhenwu M890 has already achieved large-scale commercial deployment, and with the delivery of the V900, Alibaba Cloud's self-developed computing power supply capacity will be further strengthened. T-Head also plans to launch the next-generation J900 chip in the Zhenwu series in 2028.
