Sam Altman：本地推理浪费算力，数据中心每token成本占优
Sam Altman on local inference.
Sam Altman on local inference.
Sam Altman 谈本地推理。
"from a global allocation perspective, it's quite terrible to have everybody, a bunch of the world's limited supply of DRAM sitting in their own computer and being used 5% of the time"
"从全球资源分配的角度来看，让全世界每个人都把有限的 DRAM 供应囤积在自己的电脑上，且仅使用 5% 的时间，这简直糟透了。"
per token, the datacenter wins easily. A local model at batch size 1 is wasteful by any utilization metric, and it’s also the only part of the stack nobody can deprecate on you. https://x.com/morqon/status/2115838856524120128/video/1
按每个 token 计算，数据中心轻松胜出。任何利用率指标下，批处理大小为 1 的本地模型都是一种浪费，而且它也是整个技术栈中唯一无法被他人淘汰的部分。https://x.com/morqon/status/2115838856524120128/video/1
更进一步：量化金融体系
看懂新闻只是起点——沿量化金融路径，把它变成能交付的工程能力