June 30 — Zhonghao Xinying released a next-generation, fully self-developed
high-performance TPU AI compute chip, Xuyu®, and an integrated software-hardware
compute base, Taize® 2.0 AI high-performance compute platform. Compared with
first-generation Shana® chip and Taize® compute servers, the new products
improve underlying architecture, peak compute, on‑chip storage, cluster
interconnects and compute energy efficiency.