Why measuring AI token usage is a trap for business leaders

Synthesia, a London-based AI video firm, expects staff to leverage AI across all roles, but the company actively discourages 'tokenmaxxing' as a performance metric. Gonzalez argues that management should instead focus on outcomes: whether the technology helps teams ship products faster, improves customer call preparation, or accelerates the onboarding process for new hires. While some organizations initially used high token burn rates as a proxy for successful AI adoption, this approach is losing favor as costs mount.

Financial realities are forcing a broader shift in corporate strategy. Pylon CEO Marty Kausas reported that his company’s annual bill for Anthropic is projected to climb from $400,000 to $1.4 million, prompting him to declare that the era of aggressive token consumption is ending. Meanwhile, companies like Coinbase have implemented strict weekly spending caps ranging from $500 to $5,000 to manage the surge in AI expenses. Despite these constraints, experts note that access to AI remains a relevant perk for recruitment. Recruiter Jeff Hyman suggests that asking about token allocations is becoming a standard inquiry for job seekers, much like asking whether an employer provides a company laptop.