【#Tech24H】On June 30, Zhonghao Xinying officially launched “Xuyu”, a new generation of fully self-developed high-performance TPU AI-dedicated computing chip, alongside the “Taize 2.0” AI high-performance computing platform, a software-hardware integrated intelligence base built around the Xuyu chip. As comprehensive upgrades from the inaugural “Shana” chip and the original Taize server, Xuyu and Taize 2.0 deliver leapfrog improvements in underlying architecture, peak compute performance, on-chip memory, cluster interconnect, and computing energy efficiency. According to the company, Xuyu continues Zhonghao Xinying’s fully self-developed TPU roadmap, achieving complete in-house development of IP cores, proprietary instruction sets, low-level operator acceleration libraries, and full-system software, with no reliance on overseas core technologies.