February_AI Optical Communication｜Structural bottleneck of CPO optical interconnection technology in AI mega-scale computing architecture
Author: Mr. Lin Weizhi, Executive Vice President, Ji-Pu Industrial Trend Research Institute
隨著生成式 AI 與大型語言模型的訓練規模持續擴張，資料中心正快速邁向十萬乃至百萬加速器節點等級的運算形態。當叢集規模突破傳統設計假設，系統瓶頸已不再僅來自單顆晶片算力，而是節點間資料交換效率與整體能耗結構。互連系統的功耗、熱密度與穩定性，開始直接決定整體運算平台的經濟價值與可持續性。在此背景下，傳統插拔式 pluggable optics 所帶來的功耗與空間壓力逐漸浮現，因此光電共通封裝（CPO）被視為
For more details, please register or log in.Member Login.
Previous
January CES _2026 Feature｜CES 2026 is not about new product announcements, it's about Jen-Hsun Huang setting the anchor for the next decade of AI openness.
January CES _2026 Feature｜CES 2026 is not about new product announcements, it's about Jen-Hsun Huang setting the anchor for the next decade of AI openness.
