Every time you ask ChatGPT to write an email, summarize a document, or generate code, you’re burning through tokens. Every time a corporation runs thousands of customer service queries through an AI model, it’s burning through a lot more of them. And now, economists are starting to pay very close attention to how many tokens the entire economy is consuming.
AI companies have long used tokens as their fundamental billing unit, the way a utility company charges for kilowatt-hours. But the metric is quietly graduating from corporate invoices to something much bigger: a macroeconomic indicator for tracking how fast AI is spreading through the economy.
Tokens as the new kilowatt-hour
Here’s how it works. In AI, a “token” is a chunk of text, roughly three-quarters of a word, that a language model processes. When OpenAI or Anthropic or Google charges enterprise customers, they meter usage in tokens. Input tokens for what you send the model, output tokens for what it sends back.
The parallel to earlier technological shifts is hard to miss. Economists once tracked electricity consumption to gauge industrialization. Then they tracked cloud computing spend to measure digital transformation. Token consumption is shaping up to be the equivalent metric for the AI era.
