Talent over tokens: AI models are becoming more expensive to run, and productivity gains are limited — efficient workers might be the solution to strained budgets

Enterprise AI is getting more expensive as token and compute costs rise, squeezing company budgets.

Examples from Nvidia, Uber, Anthropic, and Microsoft suggest some firms are seeing limited productivity gains from AI.

If costs keep climbing without clear returns, businesses may slow AI adoption and favor cheaper, narrower tools or human hiring.