AI 推理的隐性存储账单：后端数据远超输出
原标题：AI’s hidden storage bill: What inference leaves behind
AI 摘要
MIT Technology Review AI 刊文指出，AI 推理的存储成本常被忽视：WD 测量发现，一段 8 秒、用户可见仅 3.97MB 的 720p AI 视频，后端会留下超过 500MB 的日志、缓存、渲染、备份与安全记录等数据，放大约 130 倍。WD 委托 IDC 对 763 名 IT 与业务决策者的调查显示，近 95% 受访者称过去 12 个月生成式 AI 导致数据量增长，约 61% 增幅达 25% 以上；McKinsey 预计到 2030 年推理需求年复合增长 35%，超过训练的 22%。文章建议企业按 AI 数据全生命周期规划分层存储，并将归档数据视为 RAG 与推理的活跃资产。
正文节选
Sponsored Provided byWD Ask most executives to picture the cost of AI and they’ll likely see GPUs—racks of accelerators with gigawatts of power—and a waitlist for silicon. But that picture leaves something out: Data storage costs. What happens to data after a model responds? Much of the back-end data that’s generated—from billions of daily prompts, images, and videos—needs to be stored. But data storage is one of the harder infrastructure costs to plan for, and it’s often overlooked in initial A