Token usage
person concept tool org talk claim — click a node to jump to its page; hover an arrow for the relation
Token usage, in the context of AI-augmented organizations, refers to the volume of AI model calls consumed by individuals or companies as a proxy metric for AI work performed — analogous to headcount or compute as a measure of organizational capacity.
Token usage as a substitute for headcount
Tom Blomfield argues forcefully that companies should reframe their resource thinking around tokens rather than employees, coining the phrase "burn tokens, not head count." His position is that firms will soon be constrained not by how many people they can hire but by how many tokens they can consume. This view is supported by empirical signals from the Y Combinator portfolio: YC companies are reaching demo day with approximately 5x more revenue per employee than 18 months ago, a shift attributed directly to AI-driven productivity. Blomfield's framing implies that token spend is becoming the new hiring budget — a capital expenditure replacing a labor expenditure.
Pedro Franceschi extends this logic to founding: if he were starting a company today, he says he would begin with the premise of "why can't it be just me," accepting that token consumption would be correspondingly high as a trade-off for minimal headcount.
Uneven token consumption inside organizations
Token usage is not uniformly distributed across a company. A three-tier model of AI adoption describes the reality inside most organizations:
- Token-maxers — "engineers that are pushing a bunch of code" and other power users who drive the bulk of token consumption and extract the most productivity.
- Average users — employees getting a fraction of the potential productivity gain.
- The rest — the majority of the company, interacting with AI only in what is called "Google search mode": a chatbot with a few MCP connections, far below the productivity frontier.
This tiered pattern is framed critically — the concentration of token usage among a small group of power users represents an organizational failure to diffuse AI capability broadly, leaving most of the company's potential productivity gains unrealized.
Implications
The concept reframes organizational efficiency: a company's AI leverage is visible in its token burn rate relative to headcount and revenue. High revenue-per-employee ratios, as seen in top YC cohorts, may increasingly correlate with high token-per-employee ratios, making token usage a leading indicator of competitive advantage.