[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"tag-tpu":3},{"tag":4,"articles":11},{"id":5,"name":6,"slug":7,"article_count":8,"description_zh":9,"description_en":10},"5722c5dc-3dbc-42fe-a9b5-5c6ff7dfada7","TPU","tpu",3,"TPU 是 Google 為機器學習設計的專用加速器，常見於訓練與推論大模型、處理長上下文與多模態工作負載。隨著雲端供給、電力與成本成為 AI 基礎設施瓶頸，TPU 也逐漸成為觀察算力布局的重要指標。","TPUs are Google’s tensor processing units for machine learning workloads, used for training and inference at scale. They matter because model builders now weigh TPU capacity, cloud availability, power use, and cost alongside Nvidia GPUs when planning large AI systems.",[]]