The Hangzhou-headquartered e-commerce giant has developed a semiconductor specially designed to run AI models and applications, which is being manufactured by a Chinese firm, sources told The Wall Street Journal. (The Wall Street Journal)

Talking point: Chinese tech firms need their own chips because Beijing and Washington have each cut them off from the Nvidia graphics processing units (GPUs) used by other AI developers. The Trump administration won’t let the chipmaker sell its most advanced products to Chinese customers, but has let it export the less powerful H20 processor. But Beijing has told Chinese firms not to use H20s. Alibaba has released a series of open-source AI models as well as generative applications, and needs specialized chips to train and run its systems. It’s now reportedly trying to have the hardware manufactured in China, instead of by foreign foundries like TSMC.