The Zhitong Finance App learned that according to reports, Pingtouge, a semiconductor subsidiary of Alibaba (09988), plans to go public in Hong Kong in 2027 and plans to raise at least US$2 billion to US$3 billion, or HK$15.6 billion to HK$23.4 billion.
According to data, Pingtouge Semiconductor was founded in 2018 and is a wholly owned semiconductor business entity under Alibaba Group. It has end-to-end capabilities from chip architecture design and R&D to product implementation. At present, the company's product layout has covered various fields such as CPU, GPU, NPU, network, and storage, focusing on building a complete chip product system around data centers and AI computing power infrastructure.
In the field of AI computing power, Pingtou has continued to accelerate product iteration in recent years. In September 2026, Ali unveiled the next-generation AI chip Zhenwu V900 at the Yunqi Conference. According to Ali's disclosure, the chip's performance is about 3 times higher than the previous generation Zhenwu M890. It has 216 GB of video memory and 1200 Gb/s inter-chip bandwidth, supports FP8 and FP4 precision calculation, and can connect up to 500,000 chips to form a hyperscale cluster for cutting-edge large-scale model training and inference. Ali expects the Zhenwu V900 to be mass-produced and commercialized in the first quarter of 2027.
It is worth noting that Pingtou's commercialization progress is no longer limited to Ali's internal ecology. Ali previously disclosed that as of the beginning of August 2026, Zhenwu series chips have served more than 650 customers, covering various industries such as automobiles, finance, energy, manufacturing, etc., and supernode instances based on the M890 have also been launched on Alibaba Cloud for large-scale sales. This means that Pingtou is gradually expanding from the group's internal computing power supplier to the external market.
Judging from the industry background, Pingtou is seeking an independent listing at this time, which coincides with global investment in AI computing power infrastructure continuing to heat up. With the expansion of large model parameters and the accelerated implementation of applications such as AI Agents and embodied intelligence, the market demand for high-performance GPUs, servers, network connectivity and memory chips continues to grow. At the same time, US restrictions on advanced chip exports to China are also further pushing Chinese technology companies to speed up the construction of a domestic computing power supply chain, providing new market space for domestic AI chip manufacturers.