AI Computing Power

AI computing power encompasses the infrastructure needed to train and run inference on large models, including GPUs, memory, networking equipment, substrates, cooling systems, electricity and data centers. New models such as Kimi K3 continue to drive demand, Anthropic has signed a major computing capacity agreement, Nvidia is expanding its startup investments, and suppliers are accelerating capacity expansion and shifting their service models. At the same time, the concentration of computing resources, energy consumption and regional competition are drawing scrutiny. Tracking this topic offers insight into AI investment trends, changes in hardware supply and demand, and the evolving landscape of technology competition.