Back
Technology

AI Token Spending Increases Amidst Questions of Measurable Productivity Gains

Generated on: Last updated: 2 sources
View source

Tokenmaxxing: The Hidden Cost of Corporate AI Adoption

Corporate adoption of artificial intelligence (AI) has led to a significant increase in the use of AI "tokens"—the basic units processed by AI chatbots, roughly equivalent to three-quarters of a word each. The practice of maximizing AI token usage, sometimes referred to as "tokenmaxxing," has become a topic of debate among executives and technology professionals, with conflicting reports on its effectiveness and cost.

Observed Inefficiencies and Sprawl

Reports indicate that the rapid adoption of AI tools has led to "AI sprawl," where workers use multiple programs, often duplicating efforts and wasting tokens. A survey of 6,000 digital workers by the company Glean found that:

  • 77% use multiple AI programs weekly
  • 33% use four or more
  • 60% re-run prompts across tools when initial outputs are unsatisfactory

While the survey reported that workers save an average of 11 hours per week individually, only 13% of respondents stated that these savings have "significantly improved" overall company performance.

"That link is not there yet, right?" — Uber COO Andrew Macdonald, on connecting token usage to productivity gains

Lee Senderov, chief transformation officer at Travelport, reported an instance where one worker used 160 times the tokens of the next highest user over a four-day period. Some technology professionals have publicly questioned the return on investment. Akshat Bubna, CTO of AI startup Modal, posted on X that 50% of internal token spend is "completely useless." Engineering manager Karthik Hariharan stated that tokens were "burned for millions of dollars without any real significant ROI."

Rebecca Hinds of Glean's Work AI Institute described the situation as a "tragedy of the commons," where individual optimization may degrade collective team performance.

Executive and Corporate Statements

Uber COO Andrew Macdonald stated in an interview that he has not observed direct productivity improvements from increased AI token usage. "That link is not there yet, right?" Macdonald said, noting that while more products may be shipped, it is difficult to attribute a specific productivity increase to token usage. His comments reportedly gained over 2 million views on X.

Google CEO Sundar Pichai said at Google's I/O conference that chief information officers are concerned about budget overruns on AI, adding, "I think the problem is going to get worse as we go through the year."

"I think the problem is going to get worse as we go through the year." — Google CEO Sundar Pichai

Investor Michael Burry called tokenmaxxing a "crazy, rushed, temporary phase" and warned that Nvidia stock could face an "aggressive" decline.

Y Combinator CEO Garry Tan defended tokenmaxxing, stating "we've been tokenmaxxing longer than most people."

Corporate Responses

Several companies have modified their AI policies in response to its use and costs:

  • Amazon removed an internal AI leaderboard after employees reportedly gamed the rankings
  • Palantir CEO Alex Karp compared tokenmaxxing to a porn addiction
  • Duolingo revised its policy on AI use in performance reviews
  • Meta and AT&T are reported to have reduced AI use due to rising costs

Visa tracks employee AI use, rewards teams for faster AI-driven development, and reports a monthly token spend of nearly 2 trillion. A report from engineering intelligence company Jellyfish found that the top 10% of Claude Code users consumed about ten times the median developer's tokens but produced only about twice the output. The Jellyfish report recommends tying costs to concrete metrics like pull requests rather than rewarding or penalizing raw token consumption.

Scholarly and Analytical Perspectives

Researchers and analysts have identified potential issues related to trust and coordination. Past research by BetterUp found that AI-generated documents without oversight reduced coworker trust. Kate Niederhoffer, head of BetterUp Labs, stated that companies are not answering "the big why" regarding AI adoption.

"Organizations exist to coordinate workers and align them toward shared goals." — Emily DeJeu, Carnegie Mellon professor

Emily DeJeu, a professor at Carnegie Mellon, noted that organizations exist to coordinate workers and align them toward shared goals. Lee Senderov suggested that companies need to centralize AI workflows to avoid duplication.