Alibaba scales AI compute with new chip and 20 GW data center target

Alibaba's push into proprietary AI chips and massive data center expansion shows its commitment to a full-stack AI strategy. The Zhenwu V900, set for 2027, and the 20 GW data center target by 2032 are aggressive moves. This internal development aims to control more of the computing stack, reducing reliance on external suppliers and potentially improving cloud gross margins over time. Alibaba's adjusted EBITA for cloud services rose 133% year-on-year to RMB 5.7 billion in the June quarter, indicating early success in this strategy.

This strategy positions Alibaba as a stronger competitor in Asia's cloud and AI markets. Its investment in custom silicon and vast infrastructure directly challenges regional rivals like Tencent Cloud and Huawei Cloud. The scale of 20 GW by 2032 suggests a long-term play for dominance in AI infrastructure. Increased internal chip adoption could also give Alibaba a cost advantage, impacting pricing and profitability across its cloud offerings in China and Southeast Asia.

The thing to watch is Alibaba's free cash flow. It was negative RMB 44.7 billion in the June quarter, worsening from RMB 18.8 billion a year earlier. While cloud profitability improved, the capital expenditure required for this infrastructure buildout is substantial. Sustaining this level of investment while improving overall cash generation will be the key test for Alibaba's ambitious AI strategy through 2027.