AI Token Demand Rises as Companies Introduce Spending Limits
A 36Kr report and commentary by Kuaidao Qingyi connect sharply rising AI token demand with difficulty for model companies to generate profits. OpenRouter data cited by 36Kr show tokens routed through the platform rising about 250-fold to 128.4 trillion over 20 months. In the second week of September, open-weight models handled 72% of those tokens but accounted for only 27% of platform spending. The commentary reports that Salesforce introduced role-based token budgets, Apple reduced some engineering quotas, and Uber imposed per-person, per-tool limits after exhausting its annual budget in four months. It also argues that outcome-based pricing remains difficult for general-purpose AI because business results can be hard to define and attribute among AI, people and other factors.
The reports show that enterprises are adding formal controls to AI spending and that strong token usage does not automatically translate into proportional revenue or clearly measurable business value. They also highlight why general-purpose AI providers face pricing and attribution challenges.
2026年的AI行业,正卡在一个反直觉的尴尬期。
一面是当前的AI需求在一路疯涨。据OpenRouter数据,其平台调度的Token处理量在20个月内增长了约250倍,达到128.4万亿枚;算力租赁市场也随之水涨船高,H200的租赁价格年内上涨了160%,B200上涨120%。
然而,落入口袋的利润却异常惨淡。同样是OpenRouter数据显示,开源权重模型在9月第二周处理了72%的Token,却仅获得27%的平台总支出。
不仅平台端分不到钱,模型公司自己更是在流血。以智谱为例,其2026年上半年经调整净亏损近20亿元;而大洋彼岸的OpenAI,第一季度的运营亏损率也高达122%。
暴涨的需求,硬是没有拉起模型公司的利润。这不禁引人深思,卡住AI回收利润的到底是什么?
01、被替代的旧活与算不清的新账
AI进军市场,最先受到伤害的是那些高度标准化的专用工具。
#OpenRouter#Salesforce#Apple#Uber#Visa#Tencent#Anthropic#OpenAI#Zhipu AI#Stripe#Agentforce