Rohan Paul· @rohanpaul_ai · X·· 2 小时前AI 评分35
AI 导读
Ramp 上 AI 支出排名前 1% 的企业，把超过三分之一的 AI 预算花在了原始 GPU 算力上。 Ramp 将 LLM 订阅、编程智能体订阅、API token 计为 LLM 支出， 而 GPU 算力则涵盖为硬件容量本身付费（即重度用户租用 GPU 来训练、微调或托管自己的模型，而不只是调用 API。）
正文 · 原文
Ramp’s top 1% of AI spenders send more than a third of their AI budgets to raw GPU compute.
Ramp counts LLM subscriptions, coding agent subscriptions, API tokens as LLM spend r
while GPU compute covers paying for the hardware capacity itself (i.e. heavy users rent GPUs to train, tune or host their own models, not just call APIs.)
Interpret this how you will, but more than a third of AI spend by the top 1% of firms is already going to compute在 X 查看被引用的帖子
来源：Rohan Paul · x.com