Vincent Gusdorf and industry experts warn that the 'tokenmaxxing' trend in artificial intelligence is reaching a point of diminishing returns.
The corporate trend of 'tokenmaxxing'—maximizing the usage of AI tokens to drive productivity—is facing a backlash as businesses realize that high token consumption does not always equate to increased value. Vincent Gusdorf, head of AI analytics at Moody’s Ratings, noted that while it is easy to create unnecessary content with AI, the rising costs of these tools are forcing a more disciplined approach. Industry leaders like Alex Karp and Satya Nadella have expressed concerns over the high costs of tokens and the potential for companies to pay twice by feeding proprietary data into models. To address this, many firms are shifting toward 'tokenminimizing' and 'model routing,' where simpler tasks are assigned to cheaper, more efficient systems. Economists Nicola Borri, Aleh Tsyvinski, and Yukun Liu have utilized data from 380 trillion tokens to track the 'AI Factor' in financial markets. Their research suggests that investors expect a wide range of industries, beyond just tech, to benefit from AI, creating an 'AI premium' for stocks in sectors like utilities and manufacturing.